The Vibe Coding Security Risk Nobody's Talking About ｜Ali Parandeh @oreilly Author

me to refresh the page, so give me one sec.

Yeah.

good. It's not interrupting them. Okay, great. Yep. Sounds great. Give me one sec. Okay, let's get started. Okay. Hello, everyone. Welcome back to TwoSet AI. We have ⁓ Ali Perende today. He was on a show before and we were talking about live coding and building production AI systems.

Welcome back.

Since then, he's left his job and started his own company. Ali, welcome back.

Thank you very much for having me.

Okay, maybe start with the latest, well, recent news. I'm really curious about your opinion on this. So open claw. You know, we know it exploded, right? 250k GitHub stars, know, employees are installing it on their work machines and connecting to corporate systems. People are letting AI run overnight, building products while they sleep. That's my dream too. What do you think? When you see that.

I mean, it is extremely exciting times. A lot of people are feeling, ⁓ becoming like builders right now. They are building their own AI agents. It's becoming more democratized. Like anyone can do it these days. You can have, you can have your own AI agent sitting in your house, doing work for you automatically. And all you need is buy yourself a Mac mini or as mini mini computer that can run these AI agents and People have found they can unlock a lot of free time for themselves on doing their admin for them, doing their finances, doing all sorts of various tasks like setting up their own personal assistants where it's connected to all their internal systems and they can essentially just talk to them on WhatsApp. And I have one of my friends who set this up for himself and he, for instance, use it as a fitness trainer. So he just sends him a message and says, I've got this in my fridge. How can I? What do I do I make for today? Well, I'm going to the gym. What should I have for workout plans? And it's very personalized to him because of the way he sets it up. Even orders him delivery food and things like that from, ⁓ from these delivery apps. Yeah. So it's, it's, it's crazy times we live in.

Are you using it yourself or do you? ⁓

I'm not using it myself now because I have my own set of apps that I use and I don't use it myself is because I don't trust, it doesn't have any safety guardrails. And also if it goes off tangents and also if it starts going off around, it can end up costing you thousands of hundreds of thousands of dollars or pounds. in API costs and tokens because these agents are not scaffolded enough. They could get into a loop and it can cost you a lot of API tokens. And at the moment, I didn't find the need in terms of how to use them to justify the need to purchase a Mac mini and set all of those things up, whereas I normally automate my workflows with existing tools that I've access to. I'm quite happy with those. And also I do want the privacy and the safety around that. So that's why I didn't set those open closed myself.

God. Do you feel your friends are happy with it?

⁓ They're happy with it. If you're using it for personal, let's say, if you're not running it for a business or anything like that, and you're setting it for personal, and you're happy with the risks, then you can set it up for your personal agent. Also, sometimes people may not necessarily want to pay subscription. So that's one of the main reasons they pay per, they set up OpenClo because they pay per token instead of paying a subscription. So you still need to connect it to some time of API. Most of the time people are using connecting into Claude. API or open AI API, but you're still paying by tokens. If you're a heavy user and you're paying for pro subscriptions with these tools, it might actually be more cost effective to not have a subscription and use OpenClaw instead, especially because it has tool use and most of the time it does a good enough job.

Mm. Even with using a Mac Mini, doesn't, for any, even for personal use cases, you're still going to expose your information and your privacy stuff to it, right, in order to make use of it. Like, you know, let it access my Gmail, then I'm exposing all my Gmail to the tool, right?

Yeah, especially if you're going to be putting it on the internet so you can use it through WhatsApp. These days, there a kind of authentication in place where you need to double authenticate with that agent to use it. But it's not going to be in a vacuum. are attack vectors available to them. Most people try to put guardrails around it by putting some system instructions to follow those is called ⁓ like very light alignment guardrails essentially, where you're providing system instructions to not hallucinate, do not do x, y, z, maybe even more descriptive than that. But it's still prompt guardrails are not as strong as actual guardrails that you put in as input and output guardrails that you may build in an enterprise agent. ⁓ Alongside will authentication authorization guardrails around. your MCP servers or model context protocol servers where the agent can communicate with to interact with tools.

Yeah. So I mean, it's still, it's still, don't know what stage we should call it. It's alpha beta stage that is still developing, right?

I think at the moment there is a hype around it. There's still some time before it becomes more adopted. Big players are starting to adopt the standard and start using it. are massive meetups happening around it. There's a lot of early adopters starting using it. Like for instance, there was a meetup I read about that people in that meetup have used it to automate certain investment decisions, like especially for stock markets and lot of personal use.

Right. Mm.

So for people who are a bit more adventurous and are actually like, you know, installing it on their systems and playing around with it, they have found really interesting use cases. Whereas before, for instance, people would sit down and create automations in NA 10 or in make.com in these kinds of platforms, create like multi-agent orchestrations. Whereas now they're doing it in, for instance, open claw because it's more like an open open source framework for these things. and it provides more integrations and abilities to perform various different tasks.

Do think this is, well, this is definitely going to be the future, right? But do you think this indicates the, you know, the up and coming truly autonomous AI employees?

Potentially, ⁓ it depends. think for it to obtain better adoption, they need to standardize the security layer and the safety layer much more significantly. The problem with some of these agents is they can end up exposing API tokens. They can end up exposing your system inadvertently. ⁓ Or they can be manipulated using social prompting or prompt injection, various different attack vectors. that could cause problems. And especially with people giving access to their agents to social platforms like Multbook, which is more like a Reddit for AI agents. And they just let these agents go wild on that system. ⁓ There could be all sorts of vulnerabilities. And there have been significant CVEs raised for OpenClaw as well in the past, like confirming these vulnerabilities at really critical stages. ⁓ It has been one of the main reasons I've not been using it because ⁓

Mm-hmm.

Even for my personal laptops, I have a lot of personal information on there and I don't want to install any system that has access to personal data or various different reasons, especially when I'm working with clients or anything like that, working on work laptops and work computers. Definitely it's an ergo for these systems. And you can't really suggest it to any clients because unless they're going to run it in some kind of vacuum isolated network, it's going to cause problems for them. And there's also a maintenance overhead. And not even talking about the co-token costs. I shared your link where there was someone racked up $800 or $300 of costs because their open AI agent, sorry, their open Claw agent ran into some kind of loop and couldn't decide to stop.

Get out. Yeah. That's crazy.

And you don't want to open your billing account and be surprised with a $300 bill. You might as well just pay $200 every month to claw clot or to open AI and just go crazy on the usage if you want. It's good enough to give you things that you need and you can also build agents on top of that if you want.

Right. Hmm. When do you think we'll actually be able to hire a reliable AI employee?

hire, you mean like from a marketplace. I think there might be some entrepreneurs creating marketplaces for AI employees where you go in and start using someone else's AI. There might be these ecosystems coming in so I can see that happening. And I think it would be useful. It's similar to the GPT storing, OpenAI, and ChatGPT, or the skills hub for Claude, where you can get skills.

Yeah.

You know, like for instance, in Notion, can go to the template or Figma or Notion, they have community stores where you can download templates or things like that. would just be like that. Someone else builds an agent, connects it to certain services and you can start using it. I've seen that Notion as well as going down that route where you can now install and configure custom AI agents in Notion and already comes with pre-configured integrations where you can integrate it to major systems like your... like your Google Docs, your Google Workspace, your MS 365, Microsoft 365, and even add your own MCP on top and customize that NGB, the system prompt. honestly, I found these third party tools really interesting and really useful because like software notion, for instance, if you build custom agents in it, it's got all the security infrastructure in place. They've got ISO 27001, 9001, they've got SOC 2. Like I'm more confident putting more, putting sensitive data in Notion because it's got all the standards and it's a security infrastructure and I can back that up, back that data up and I can use their API to integrate things compared to an open source where I need to build the entire infrastructure myself. And basically spend more time doing that than actually running a business.

Yeah, it will only become more powerful though. Like let's imagine like notion AI plus isn't wrong. Once we address the security issues, then it's

Yes, absolutely. Yeah, absolutely. It used to be like, for instance, before to build agents, had to use a lang chain, lang chain framework. Then Pydantic AI came in like learning lessons from lang chain and creating a framework for building AI agents in Python. And now we have OpenClaw. So it's like the next revolution. But as you said, OpenClaw sits somewhere between alpha and beta stage at the moment, from my opinion. So it's not something you want to use production grade in your business.

Yeah. but it's coming.

It is coming. If you're a small business, you're an adopter, you're a solophenate, definitely something to play with to actually start using it. Maybe I should start using it, but play around with it and figure out which workflows may not necessarily be super sensitive that I can start putting in it. But honestly, I didn't find the need to. Every day-to-day workflow, like with emails, with tools, there is fair part of those I can subscribe to that doesn't cost a lot of money. that does the job way better or there is a custom agent so can build in Notion for myself.

Tell me about it, right? Last time we talked about how AI can turn a 10x engineer into a 100x engineer. And now you are a solo founder now and you don't have a team at iZoom yet. So you must be building your own automation stack, right? Can you share what you're using and how does it fit into your day?

Yeah, absolutely. the first things that normally solopreneurs or people coming into the AI space will face is there's a lot of AI tools out there. And for someone who's new to AI, it is absolutely overwhelming. There are AI cheat sheets. So I have created my own cheat sheet that I can share with you. But essentially, ⁓ you don't.

You should, you know, your tea sheet, you shouldn't put it on the website and I can share it with this episode and then it will, you should add a call to action to it. And then people can either sign up, either you have a newsletter or whatever you have, right? Give a call action and that to me.

Yeah, definitely. Yeah, definitely. I will put it on the website and it can be used as a lead magnet. Essentially, you don't need hundreds of tools. You need a few cornerstone tools. For instance, you do need one AI general assistant, like for instance, Claude or ChatGPT, like Pick Your Pick, that you subscribe to for day-to-day conversations. That's something you can easily bring up on your phone and talk to. ⁓

Yeah, exactly.

You do need a way to talk to all your notes and documents. So note taking is very important for shoulder panels. Like you're going to be creating a lot of documents. You're going to create your own SOPs. You're going to, you need some kind of transcriber that transcribe your meeting so you have, can come back into them and do actions. So if everyday notes, need accounting software, but essentially there is a handful of AI tools that you need for day-to-day work. And you don't need more than that. Everything else is more like on a neat. on neat basis if you want to. For instance, if you're working in a creative industry, you may need a lot of like ⁓ Canva AI or like these other tools that allow you to generate images. But for most business owners and even engineers, like you just need a general assistant that you pick and that could do a lot of your workflows for you to some extent. More advanced use cases, you may go down to workflow automation tools like NA10 or make.com. make.com might be much easier to learn ⁓ and then go from there. And some businesses that don't want to use a lot of third party tools, Microsoft 365, for instance, or Google Gemini, they provide enough automations in there that allows you to automate most of the workflows that you have to help you go forward. So I, myself, I use Copilot 365 for certain, I've created my own custom Copilots for certain scenarios. For instance, Whenever I need to expense, either create invoices or put expenses to clients, I use Co-Pilot to convert my receipts into a spreadsheet that is approved by the client in a format that they like. And I can upload sensitive information into it because the data stays within my 365. But for general day-to-day workflows, I use Notion as my transcriber, as well as my knowledge management system. So all documents live in Notion. ⁓ It's more like my SharePoint and I can... Essentially, it has all the models for available like Gemini, GPT, Claude, and I can basically talk to the entire document set. And I've structured my notion in databases. So I have a database of documents. I have a database of customers. I have a database of leads, database of XYZ. And these databases essentially link to each other. And it allows me to essentially create new records as I like and talk to all of them. And as I'm working on a document, I can talk to my AI assistant. It can also help me write the document up. So most of the time when you're working in cloud, for instance, when you use an artifact, it just sits in a conversation. You can't do much with it. Whereas in Notion, I can get Notion to produce 50 documents for me while I'm talking to the Notion AI.

⁓ So for Notion, are you chatting with Claude and then generating the artifact into Notion? Is that how you're connecting it? Or you're just chatting with him?

Yeah. yeah. So Notion AI Chat essentially has Claude model. ⁓ It communicates using Claude. For instance, you can select the model. But essentially, if you ask it to create your documents, let's say you have a long conversation, for instance, for a business strategy, you can then ask Notion to create you the SOP or the Claude and just creates your page. And it can edit the page. can edit that artifacts.

I

And it's very similar to the experience you get with ChatGPT Canvas mode, where you really go to the document and you can edit the document by talking to the chat.

Is it just connecting your Cloud API to Notion? that how it works?

⁓ So Notion has their own service agreements with Cloud, Gemini, these models. the data essentially goes through their own API. So you don't need to provide your own API tokens or anything. You pay a subscription fee for Notion AI. And it is not cheap. It's quite expensive. But essentially, I think I pay like 250 pounds a year or something like that. But I get good amounts of usage I can use it for.

Do you feel that? Go ahead. Okay.

So you can use it to create your databases. You can use it for creating artifacts, editing documents, reviewing across various different pages, take transcriptions of meetings. ⁓ And you can create your own CRMs in there. It's very, very useful. And I can even connect my website to Notion through their API. So whenever new leads come in, it just creates a new record in my Notion database.

Do you feel the chat experience within Notion, even though you're using Claude within Notion, is similar to chatting within Claude?

So I have Claude as a subscription separately, and I have Notion. Notion is my main to go to because whenever a conversation creates an artifact, that artifact can then get generated in my Notion space in a correct database. And I can come back to it and I can search it. If I'm creating, let's say, a landing page, or if I'm creating a UI or a prototype, Claude and Claude Cowork are very useful because with Claude Cowork in the Cowork mode, You can point it to a directory and it can just literally create on that directory documents on your default system. whereas cloud itself can create you like prototypes and artifacts that loads in your browser. So if you're creating, say utility tool, let's say you want to merge bunch of CSVs and text files and there is no, you don't want to download anything and you don't want to upload stuff to a online website to do that. Utility. You can easily create yourself a utility tool using cloud. And it can just give you an HTML file that sits on your computer. You open it up in a browser, and it allows you to easily upload files. And the files don't go anywhere. It's just on your browser. And it allows you to merge those files together if you just want a really quick, simple utility tool or some kind of dashboard or anything like that. So there is use for Cloud as well subscription, especially because a third mode that it has is Cloud Code that allows you to essentially talk to your terminal.

For using, it's usable. Yeah, good point. It's just my experience is like, you know, the third party tools are integrating with cloud or OpenAI. There's somehow behind the scene calling the API. the issue with calling the API versus the chat experience is the chat experience within cloud or chat GBT, I know they have a lot more optimization within the chat that doesn't just come with the API calling. So I stayed within the cloud space ecosystem.

yeah.

for that reason.

Yeah, yeah. The Cloud desktop app itself will have ability for you to add your own custom MCPs or anything like that. But I would say some of the other third parties are getting way better. They allow you to switch between models. So for instance, Gemini might be really good at image understanding. Cloud is really good at copywriting. Chat Chippity might be really good at strategy planning or anything like that. So different models might have slightly different strengths. So you may want to switch between them. Having a third party tool that allows you to switch between APIs is normally useful for this type of tasks. Especially with Notion, like dropping new features, like meeting transcriptions, I don't know, like ability to generate images inside your documents, ability to ⁓ create databases, summarize things, all of those things inside one tool is normally good because you don't want to keep switching tasks and contexts.

Yeah, yeah, good point. It doesn't limit me if I want to like switch, you know, I'm switching between try to be T called and Gemini all the time sometimes. Yeah, so yeah. ⁓

Yeah, exactly. And for people who work in Google Workspaces, have Gemini and Gemini and Google Notebook LM that allows them to essentially have the notion experience in Google Workspaces. Microsoft is trying to do the same with Co-pilot, but I would say Microsoft is still little bit behind because they have a lot of legacy infrastructure, and they're trying to comply that enterprise-grade guardrails, and that's going to slow them behind. ⁓

⁓ Can we make this?

The moment copilot is not producing the same level of quality and output that third party tools does. That's just where Microsoft is. They have a lot of legacy infrastructure like Microsoft Graph API to integrate with. you can't really talk to your documents in SharePoint as well, but they do provide you tools like copilot agent studio that allows you to build custom agents and connect them to the web, connect them to their credit, deep research agents, things like that.

What else do you use other than notion cloud?

⁓ so I use Figma and Figma make, so because I do design user interfaces, I do presentations, all of my training slides, for instance, I do them in Figma. The reason being is that Figma allows me to adjust the tone of the content on the fly. Allows me to generate presentation notes, allows me to generate images for slides. And it comes with really good templates that easily can make really super nice looking templates in Notion and Figma. ⁓ It has a FigJam product that allows you to do whiteboarding. So for instance, when I've run workshops, people create, especially when they're online workshops, people can give them access without a login to that whiteboard. They can drop sticky notes to brainstorming and then they can highlight all of their sticky notes and ⁓ arrange them by topic or arrange them by sort them in any way you like. And they all use AI to do that. And it's very useful for like quickly.

Right.

extracting value and extracting patterns from those sticky notes.

You're not using gamma for slides or the...

At the moment, no, but I may look into gamma as well. I have been essentially recommending gamma to a few clients to use because they were looking into presentations. However, I think ⁓ it depends on the use case. If you're to slides completely, making them look good, yes, use an AI tool like gamma. I feel mostly confident with making good slides manually or semi-manually. So I've been sticking with Figma. I didn't want to. rack up lot of subscription costs. But one thing I really like about Figma is the ⁓ Figma Make, which allows me to produce really good looking user interfaces for prototypes when I'm trying to show to a client, for instance, a vision that I have for an app that I could build for them. Or even when I'm iterating over product ideas with users, with potential users, I just send them, I just... ⁓ go have a bit of back and forth, like 50, 60 turn conversations with Figma make, which under the hood uses Claude again. But Figma, because it's a UI tool, it has a really good understanding of UI principles and you can make really nice looking UIs compared to other vibe coding platforms. LavaBoard allows you to build full apps.

I love what about lovable,

But the problem with Lovable that I've been telling people about, so I had one of the past clients approaching me. They wanted me to help one of their friends. This friend was a teacher. She had built a whole app in Lovable connected to super-based database and fully working. They wanted to start selling the app to schools, and the school wanted enterprise grade. And she had no idea how to code. She was mostly vibe coding the tool. The school wanted her to deploy the app into Google Cloud or Microsoft Azure. And she had no idea how to do it. And lovable doesn't let you do that. you need to switch the database to Google Cloud SQL. You need to deploy it into containers. None of those things she can do because ⁓ it's completely different. lovable doesn't let you do that. Lovable just creates your super base because it's super easy to create databases with it. And it provides an easy API to work.

You can't just assume. Yeah. If you're a developer, can you move lovable to the stuff that you build from lovable to a proper text rush?

Yeah? Yeah, if you can export the code base, you will rewrite all the connectors for database. So the database connected to SuperBase will need to be rewritten. All the OR and logic object relational mapper. All of that will need to be rewritten to work with Google Cloud SQL. And you need to containerize the app itself into a Docker containers where you can deploy them into Cloud Run or App Engine.

Right. Right. Right. So in that, know, so it's doable. It's just maybe your friend's, you know, teacher who has no experience doing it. But if she wants to, she could try to talk with Claude and how to do that exportation. Right. To start? Yeah. Yeah.

yeah, that's what I told her. Basically there is a bit of a pain to rewrite the logic to be cloud ready and also to learn the cloud bits, like understand what Google Cloud is, understand how to deploy there. But the main benefit of Vibe coding, like for instance, this UI tool that I, the UI prototype that I did when I'm validating my product idea, it's all done with Figma Make. It uses React code, but I can download all that code and reuse them. I may need to just adjust them and...

Right.

connect them together in exactly the architectural patterns I like to make it more maintainable. But it's really good code. Like even my website, I use the AI tools to create it. Even my own website that I re-skinned and recreated, I use AI to write the copy and help me iterate over the copy. So I spent two or three days with ⁓ motion AI to create the copy for my website, make sure it's tight and knit and it's really good. And it follows a 21-letter sales letter framework. ⁓ Then ⁓ I used relume.ai, so R-E-L-U-M-E, .ai to essentially iterate over various different designs at scale on different sections. So I have like 100 hero section designs, and it generates the entire copy for you. So you can easily see how it looks like. It then generates the design system. All of that is AI generated. And then once I'm happy with that, I then export the React code for each bit.

Mm-hmm. It's crazy.

I then string them together in my professional IDE, integrated development environment, ⁓ like VS Code or PyCharm. And then in PyCharm, I use JetBrains AI to help me, for instance, to animate those sections. So I get JetBrains and I tell it use, for instance, frame emotion to fade in the hero section, ⁓ the hero title, add like a typewriter effect into it. Like when I load the page, like have delays around animations, like this animation is to come in first from this site, then that one is to come to insert. When you scroll down, animations need to pop up. So I don't sit down and do the animations, which is mostly CSS work by hand. I just get a guide to do all the CSS work.

Right, right. Yeah, yeah, that's crazy. I mean, you're streaming a bunch of tools, but you know, my follow up is one thing is you have to be in the middle, right? You are the human in a loop from one step to another. You're in between figma make, relume, jet brains, right? You have to be there to make the call. Like, okay, this is the final, right?

Yeah, exactly. If you have experience, you know how to string them together like a pipeline. It's like manufacturing pipeline. Like you design the pipeline and then you get the tools to add to the actual grant work, but you are quality controlling the outputs between each stage. If you just get the AI to do the entire pipeline for you, you're not going to get a hundred percent. You're going to get 70%, 80%, maybe if it's very optimized. Most of the time you're not going to get something you like. It's just not going to have your signature on it. Whereas if you're quality controlling the pipeline, then you can use different tools or you can use the same tool. You can get the quality you want. And the main thing is you don't want to delegate your entire thinking and your entire craft to AI tools because first of all, you lose your communication skills. You forget what you've done and also your critical thinking skills. You're going to start losing it. Look at, look at people doing multiplications right now. If you go and sit in a restaurant and you need to split the bill, can you do it in your head? These days you can't.

Yeah. That's a really good comment. Yeah, really good comment.

Yeah, if you are like, let's say you are five people or six people, how are going to split the bill in your head if it's like 36.44?

Like, ask Kaji Witte and then they hallucinate. They're like, you pay the most bill, please.

Calculators have made all of us really bad at my arithmetic mathematics now in our heads. We have to rely on calculators and chat chippity is like that for language. So sometimes it's still worth spending time actually learning a craft manually. So doing things that don't scale in the beginning. like, for instance, the other day I had to learn video editing with CapCut because I started doing more videos for social media.

That's crazy.

And it was my first time video editing and CapCut provides a lot of AI features, for instance, with eye adjustment, with cutting arms and M's, fill it with words from the video and all of those things is AI driven now. And I was sitting down and literally learning it and I spent like three hours doing a 10 minute video. But I knew that that three hours that I spent learning video editing by hand will then allow me to speed up much faster down the line with editing for future videos and understanding what AI does and quality controlling that.

Yeah, you know, I've been doing video editing for two, two years, no video editing experience. And I just learned on the, on a fly when I started a channel and I still spend three hours on 10 minutes. I can tell you.

It's such a pain. It's so hard to create it. I've been following the market as well to see new video editing coming too. There was a guy that I follow, name is Daniel Priestley. He's a business

It's so hard! I love him. I really love him. I'm a big fan. He's really good.

He's a really good, he's the guy to go to for any entrepreneurship education. knows stuff. Yeah, especially if you're a solopreneur. And this guy has launched many SaaS apps. One of them is called Scorecard, where you basically create online consists. I didn't go with that. I created my own Scorecard, a mini app. I got a new one.

Yeah, especially if you're starting solo, he's the guy. I wanted to use it. very tempted. Yeah, I want to use it. ⁓ okay wait are you releasing yours? can i use yours? maybe yours is cheaper

No, no, it's not a SaaS app. It's basically for my website, I said, I want something like a scorecard, but I don't want to pay scorecard. What do I do? So I just created this quiz with HTML CSS, and Claude did all of that for me with my design system that I gave it. So I just copy-pasted the entire CSS of my website into Claude, and I said, go build me a design. Go build me a quiz with reporting. And the questions need to be like using this framework that Daniel Chrisley teach, which is 10 questions on best practices, 10 questions on current reality, 10 questions on ideal reality, and it gives you a report. So if I had to the question system. I'll just release one of these frameworks. I learned that framework and then I used AI to generate the questions and I then go and edit the questions and create a quiz and lead magnet essentially using that and make sure that the reporting and all the questions actually make sense in the context.

You know what? going do it myself. Yeah. This is the most valuable thing, The frameworks and the way you think. It's actually not the coding part, right?

Exactly. The coding part and the generation of the copies and stuff, you don't need to become a copywriter anymore. You can use Claude, which is the best copywriters to help you with the copy. And you can still own the copy. You still have to sit down and learn copyrighting to some extent to make sure that it's not hallucinated or anything, because these models hallucinate all the time. So you can't really just trust them out of the box. And you want to put your own voice and tone to it, but and your own domain expertise. But these days as a solopreneur, you don't need to hire an entire team. You can literally do all your marketing, all your sales with using AI tools. You could have AI employees essentially. And there are companies like Querser that have scaled to millions of users and millions of pounds in revenue per month, sorry, per year, just by using just with only a few employees because the rest AI does some of the job. And solopreneurs are also building million pound businesses just using AI tools.

I'm looking forward to reaching that. Good luck.

Yeah, me too. So yeah, I'm using a lot of AI tools for various like whenever I whenever I have grant work to do, I asked myself, can I use AI? And then the other thing I asked myself is, is it better to use AR? Is it better I do it manually in the early stage? For instance, right? Yeah.

You actually asked a question? You actually asked a question?

Yeah, I mean, it's not like a sit down and go through a checklist. It's like in my head, I sit down and for instance, I need to write a mission statement. Should I use AI? like that kind of stuff, you need to sit down and think. You don't want to delegate that to AI. So I have a go at the thinking process. And when I hit the roadblock, and then I get help from the AI to ask me questions and do back and forth to finalize it. That's my workflow. So I always try to put down my own head, then use AI to enhance it. Not the other way around.

⁓

Only when I'm clueless and I use AI, like for instance, if I'm writing a copy, I'm not a marketer. So copywriting is new to me and I need to sit down and spend probably three years in studying copywriting to become really good. So in those areas, like for instance, I have an attempt and then I use AI to help me enhance the approach. But essentially you do need ⁓ to obsess over the quality by spending manual effort on

Mm-hmm.

the output on the artifact. And once you figure that system out, then use AI to scale it a bit or enhance it and find the gaps. AI is ⁓ more like an AI is a thinking partner that helps you.

Mm. Absolutely. Yeah. I was interviewing Jay Alec, who's a LinkedIn coach yesterday. And we were talking about people throw everything to AI first. That AI first thinking is slowing our own thinking down. It's just as you said, like, you know, we can't do calculation. We can't talk anymore. I don't know how to talk.

Yeah I mean, there was the MIT study that was done on the activity. So if you remember, they broke people into three groups. One group could use AI and child GPT to write an essay. Another group only do web searches on Google to write the essay. And another group had to use their brain. They had nothing else other than pen and paper. And they measured their brain activity and the scan of the activity of the brain group.

yeah.

was showing a lot of red lines, which shows very high activity in critical thinking and in memory and communication. The people who were using web search had less of those red lines. And the people who were using ChatGPD and LLMs, they had only maybe one red line or two red lines. Most of their brain was kind of deactivated. And this is also why they concluded that it will damage your critical thinking skills, your communication skills. And you also don't forget that you've produced an artifact. Let's say you write an email, an important email on sales calls or anything like that. You forget that you sent it. You forget what you said like weeks ago or months ago because you just use the AI to reply or generate things up. So it's very important to own any app with that AI generates. Like you need to sign it essentially.

Yeah, yeah, that's a Yeah, it's very dangerous. feel like human where we go trending, right? We're, don't know.

It is, Yeah, I mean, when I teach engineers AI and they are very risk-averse, they don't really trust it, essentially, I say that AI will not do 100 % of the job. It will do 80%. You don't want to overuse it. You want to use it to save you time on repetitive work, on manual work, on work you want to avoid, on work that can be automated, all of those things. In any work that's time-consuming, but also, you don't want to go off the other side. where you don't use it or go to the other side where you use it a lot and use your skills. So you want to be a bit of a mix. And in engineering, normally every artifact that gets produced, like a report, you will need to go through a reviewer process, a checking process, authorization process, and people sign their names on it. And you want to do the same thing with AI. You want to produce artifacts with AI to help you save time, especially if you're only a single person company or you're by yourself. But you do want to make sure that you own it.

Right.

So you need to read everything that it generates and edit it and think of it as someone cooking something in the kitchen and making a mess. You go and clean the kitchen up first before you ask them to go and create another dish. You don't want to get them to keep creating dishes for you and literally explode the whole kitchen with a mess. Same thing with AI.

I can't. That's such a so visual analogy. But you know, I want to comment that this is this current generation thinking. What about the next generation and the next next generation? They grew up, you know, getting used to Chachi BT. You and me, we were trained pre-Chachi BT.

yeah.

We didn't have this. So you have to learn everything. You remember Stack Overflow? There's a website called Stack Overflow in the past. And you look everywhere trying to find answers and you can only get peripheral answers. Yeah. Yeah. How much research we have to do to figure something out. It's not the same.

Yeah, it was. where you had to like find answers in different programming language and then try to back, do thinking. Yeah. Go through all those blog posts and click on the cookie app banners to get rid of them. Block pass through all the ads to just get some piece of sentence after 200 lines of introduction of the answer question you might have. I remember those days. Yeah.

Right. What are you going to tell your kids? Right. You're going to tell them, okay, you know, if you want to exercise, build your muscle, you have to exercise, do the hard things. Are they really going to do it? What are you going to teach them?

I think it's the same problem with social media. Social media reduces people's attention span. You actually have to actively make yourself not bored and do practices to help you not have attention deficit. Read books, write stuff down on paper by hand, try to give yourself digital detox. It's the same thing with chat-chipity. You need to basically decide on certain activities that you're going to do it with your own brain. and with your own intuition and not rely on language models because one of the main problems that it does, it makes you rely on them on every day or any small and big decision. Like you want to go grocery shopping, use ChatGPT to make sure that you've done the right grocery list. You want to make big life decisions whether to go with job A or job B or should I marry this person? Use ChatGPT for those. ⁓

Ask Chad to be tea. Should I be the person? Ask Chad to be tea.

Yeah, any email you send to clients, to customers, to partners, to anyone, like any email you say you always want to double check with ChatGPT because you become super reliant on these tools and you don't want that to happen. You want to cut yourself enough. Anytime you feel like you're becoming too reliant, you need to have a digital detox of language models where you rely on your own brain. Like if you're writing emails, don't use copilot on email writers. Just write it by hand, even though it wastes a bit more time. but actually you end up not losing your skills. You're not losing your intuition. Intuition is more important. Intuition is basically your subconscious mind processing millions of data points that you don't consider getting inputted into your brain, but your brain is processing when you're solving complex problems. And you don't want to lose that intuition because the more you rely on AI, the more you lose the intuition.

I think it's against human nature though.

It is. It's a half of least resistance going on using AI because you just ⁓ delegate your thinking to it. The only valid time you want to use it a lot is if you're running against deadlines and you're pushed against deadlines and you're super stressed and you just want to get tick things off your to-do list. These instances normally happens when the project organization not being exactly correct or where ⁓ you like you're firefighting in those cases, yes, you're going to use AI to speed things up, to be able to meet your deadlines. Or if you overloaded with stuff, it's better to use AI at that point to get you over those lines. But you don't want to, you don't want that to become a repeating pattern because then over time it's like eating junk food. Uh, it's it is. Yeah. It's like, it's like, if you eat junk food a lot of times, it's going to degrade your body. If you eat it once or once or twice.

It is.

of times is fine. It's not going to happen to your weight. But if you eat it a lot of times, that's when your weight starts gaining and you start losing your health. And there is a mental health as well. Then there is a brain health, which is related to you need to start, you need to use your brain to write things down. You need to use your brain to create things because you get satisfaction from craft, especially if you're obsessed over quality. And if you write things down, because that process of writing will help you crystallize your vision and it will help you crystallize your thoughts. Without the process of writing, you don't get there. Like you will have ⁓ foggy vision of what you want or what the idea is. Well, unless you sit down and write things down, which a lot of business owners do these days, they sit down and write a business plan, even though no one else will read that other than themselves. That whole process of writing those things down, it helps them crystallize their market, which their approach is, what their vision is, what their mission is, that kind of stuff.

I think this is the right point to ask you, do you have a doomsday prediction in the era of AI? ⁓

So I'm thinking, essentially, it's not the singularity that's going to be much of a problem in terms of AI writing its own code and getting robots to kill humanity, that kind of stuff that you see in sci-fi. It's not necessarily those. When institutes like Alan Turing Institute or these big, big institutions and NGOs look at AI safety, it's more about a random business coming and starting using AI in safety-critical applications. that can harm humans, that can put people at jeopardy in terms of their ⁓ social life. For instance, if companies start using AI to score employees for performance and then start demoting or promoting them or ⁓ making them redundant based on an AI score, which essentially what the AI you act is trying to prevent. So the regulation in EU is trying to prevent companies misusing AI in the certain scenarios. In my best opinion, the best way to use AI, not necessarily to replace humans, is to assist humans. So AI assisted programs and software is the best use of AI, especially in safety critical applications where AI can essentially ⁓ reduce a layer of human error. So humans can make mistakes all the time, especially when you get tired, you make mistakes. For instance, there is an automation right now with TFL where, know, have you read in TFL trains in? ⁓ DLR trains in London? No. So there was a DLR trains in London where someone got drawn over by four different trains at night because their trains are all automated and the human operator normally because nothing happens most of the time, your brain kind of goes into autopilot and you make those mistakes. And the rail accident investigation branch recommended TFL to use AI, computer vision to detect if someone is on the train. So as things get more automated,

⁓ not yet. Tell me about it.

humans' brains go on autopilot, and AI can actually use as an additional layer of safety or assisted layer to help you find gaps, find conflicts across a lot of data and information. Like in software, AI can read all of your logs of your software and give you anomalies and critical information that you need instead of sitting there and read through millions of rows of text. That's where AI is useful.

Mm. There's definitely higher stake ones, ⁓ know, doomsday predictions, other than just, you know, our brains, you know, is going backward of ⁓ the critical thinking skills. But if you think about it, the scene is the same. I think the scene is potentially overly dependent on the AI, right? The high-stake business, the real ways, right? Even think about it, even if they had this... ⁓ AI assisted ⁓ watch or safe safeguard. What's going to happen is the human is going to be lax, right? The human is going to be like, you know, I have AI to, you know, cover my back. I'm not going to pay attention to it. So that's the, that still can happen.

Yeah, I mean, the more you integrate AI into systems, one of the things that can happen then yes, the operators of those systems and the people relying on it will become more reliant. They will less use their actual craft skills. That's why you need to also train them to make sure they don't become reliant on it. Like for instance, sometimes you may want to use a calculator. Sometimes you may want to do the calculation in your head. And I use the calculator examples because AI is like a calculator for human information.

Yeah.

is for information calculation. It's some form of calculation of knowledge workers. Numbers, you normally used to them by hand when you would go through primary school and when there was not many calculators. But now calculators, everyone just use them on their phone. And over time, you become reliant on it you lose the skill. But you need to be proactive and make sure that you keep practicing the craft. So it's more a training than to prevent, and help people understand that it's like cigarettes. You don't smoke too many because you get cancer. It has lots of health issues. You wanna keep the balance and moderation.

I don't know. Yeah, after talking with you on in this session, feel like I'm gonna I'm not I'm not gonna let my kids use use chat GBT too much like don't use it.

Yes, it's, it's the thing is they will then become, Chachapiti becomes more like their therapist and you don't need Chachapiti as a therapist. ⁓ My girlfriend is a therapist and she hates people who use Chachapiti to give themselves self-therapy because it's not a therapist. Chachapiti is trained to make you feel good about yourself. it will, yeah, it will essentially try to say things that make you

Yep. I go ahead and finish your thought.

feel like, I'm feeling understood and good and stuff like that. And it makes you also reliant on coming back to it all the time and take your point of view. I'm sure OpenAI is working around Gartra. It has worked Gartras around that to make it more neutral. Or if your system prompted well enough to not do that, you can get it fine-tuned towards doing that. But yeah, essentially, you don't want kids to become reliant on AI to self-validate.

Yeah, or use it to do homework.

Or you go home, which I think a lot of teaching organizations and universities will have this problem with. They will need to rethink their entire examination system. So similar to with calculators, they now let calculators, students bring calculators into exams. ⁓ You can still do exams without having to rely without students having access to their phone or anything like that. when you

Bye. Yes. Mm-hmm.

assignments, you need to make sure that the assignments you assume that they're going to use AI. So you probably perhaps increase the level of expectation from the output. When I was a head of engineering at the previous company, I then we hire, we give people a task to, for instance, build a dashboard on an API that fetch data or build a data pipeline. I would assume that they're going to use AI. So my expectation from them was first of all, It needs to be more impressive what they produce in those few days. And number two, they need to be able to understand the entire thing. So I will ask them questions on the architecture. I will ask them questions on decisions, the trade off and the decisions and what would they do if they had more time. And these questions normally help me to make sure that they understand what they did rather than just delegating the entire thing to AI. That's how I know that they're the right fit.

Yeah. Or they need to do it in classroom. Like do it, do it right there in person. Don't take it home. Right.

Yeah, exactly. At the end, we wanted to speed up the pace of hiring. So what we did was to replace that take-home exercise with an in-person exercise instead.

Yes. Yeah. Yeah. Yeah. I have ⁓ so much to discuss with you. I feel like I have to schedule a search session with you. ⁓ But speaking of automation, I always welcome you to come back, by the way. ⁓ let's finish up this automation topic. I I want to run by you of how I'm running my automation ⁓ stack for my use cases. For, you know, just ask for.

Stop second.

expert opinion. I use three layers. So first, use Cloud projects with skills I've written. And then second, I use co-work for schedule tasks. It has limited access to some third party tools like Gmail, maybe, and Calendar, things like that. The third layer I'm thinking of is using Cloud code to build integrations outside this Cloud ecosystem. Because some of the things I know that I'm just streaming things together, right? And I have to be in the middle, for instance. Ideally, I can say that, you know, we did this recording today. I did all my research. have the transcript of the recording and then I can somehow kick off AI flow to start editing. But editing software is not autonomous just yet. So I have to be in there. But there's a whole process of, you know, the point is ⁓ after I'm done with the recording, I want to, let's say I want to send you a thank you email through my Gmail. And then I would then connect a co-work to Gmail and start kicking that off, for instance. ⁓ That's what I'm currently using. So I'm completely staying within the ecosystem cloud, partially also because I'm doing a lot more content creation than pure coding. So I don't need tools like WinServe or Cursor or Copilot as much. ⁓

Yeah, I mean, those are for people who do vibe coding a bit more. ⁓ I think what you're describing there, normally the best way to figure out how to automate the entire workflow is to map it out on a block diagram. And the other thing is to play around with the existing tools you've got and understand their limitations and capabilities and integrations that you can do and understand the level of level of autonomous. So

Yeah.

What you're essentially describing is if you want the entire process to be automated, you're looking at workflow automation rather than an AI agents. So it could be multiple AI agents that hand off their work to another one. Now workflow automation tools like na10.com or make.com allow you to automate with ⁓ a user graphical interface that doesn't require as much of technical expertise. Or you're looking at a script that Claude can, for instance, produce. where there needs to be a trigger, a workflow, and an output. So you can think of your entire workflow, your entire ⁓ input to output as a series of chained input, workflow, output. Input, workflow, output that are connected together. could be in a sequence. It could be in parallel. Some bits could be a sequence. bits could be parallel. And people essentially been creating these type of advanced automations in NA10 or make.com ⁓ where AI is sitting in the middles, in the workflow middles. There are AI agents sitting down, which are essentially API calls to model providers that sends the data. And then the data comes back, and a certain tool gets executed, and so on. ⁓

There is a- I've got to finish your thought.

I would say like on the video side, from ⁓ what I've seen so far, yes, it's not as automated as most places. are certain third party tools that allows you to like repurpose content or edit it. But nothing, yeah, it's not fully automated because first of all, video content is very heavy. It's a heavy file. It's like hundreds of megabytes, gigabytes.

Mm-hmm.

And uploading them, downloading them, editing them with AI takes a lot of energy and water consumption and CO2 generation takes a lot of energy to produce. So most of that needs to run on your laptop and most of that needs to be still manually adjusted, but editing tools are integrating AI features in to help with the process a bit.

Yeah, yeah. I mean, current workflow has the Cloud Chat and Cold Work. I haven't started integrating with the Cold part yet. And actually, Cold Work is already pretty powerful. I mean, if I can connect with some third-party tools like NNN, then I can build workflow downstream in NNN. then just what I need is this chat interface to kick off the command. So if Cloud Cold, sorry. cloud, let's say, co-work, can kick it off and just say, start this NAM process or something like that and take my Ostrot data. Then I think that's kind of complete my pipeline.

Yeah. Yeah, like you can definitely get cloud code to build you some kind of web based app where ⁓ it can, for instance, run all the time and it can do the entire work. What I've seen most people do is they build a workflow in NA 10. The trigger for that would be listening to a folder in Google Drive, for instance. So you don't even need to start a chat. All you have to do is upload the file into Google Drive folder.

Mm. Good point. Yeah.

It gets triggered by event. So it's called an event-driven trigger. So you have different triggers. have HTTP triggers, where an application sends an HTTP request and triggers a workflow. You have ⁓ file-based triggers, like a file gets created, updated, or deleted from a Google folder. You event-driven triggers, so event-driven architectures, or manual triggers, where you click a button. These are the three different triggers that you can create.

Right, right.

You basically you need to design the workflow first on a piece of paper to figure out how to chain it together. Figure out the edge cases for each workflow. Like what happens if it fails? What do you do? Where would it fail? Like what do you need to validate before running the workflow? All of those things you need to by using your brain basically or using Claude to help you plan it. Once you have a workflow planned out,

Right.

Then you sit down and integrate it together. Now tools like N810 really help you with that because you don't need say a graphical interface. can easily connect them together. It's a bit technical, but you can still do it. Then if you want to do it, if you don't want to rely on those tools, then yes, you can use cloud to build it. So cloud basically writes Python code and React to do it. React will be the user interface. The Python will be the backend. And it essentially does what N810 does, but in code.

⁓ Yeah.

The UI will just essentially give you a button as a trigger or gives you an integration to Google Drive. And you just need to put that tool somewhere or make sure that tool keeps running all the time and listening.

Yeah, thanks for helping me out. I think this is very empowering for all levels of skill set in terms of coding.

Yeah, definitely. think one tip I would give is if anyone wants to build these workflows, start a chat in Claude, just a normal chat. Describe your idea in a, dictate your idea to Claude so you don't even need to write it. Dictate it for five, 10 minutes. And at the end, ask Claude to ask a bunch of questions about what exactly the work, like to help you understand a bit more about the workflow. Once you get interviewed by Claude and you provide enough more context, ask to generate you a mermaid diagrams, like block diagrams, sequence diagrams, system breakdown diagrams, and data flow diagrams. So you can ask it to generate these diagrams and ask it to use mermaid, mermaid ⁓ diagramming language, because it's just text. It can generate that text for the diagram, but then they get visualized inside Cloud. So you can see the entire visualizations. So you don't even have to sit down and create those visualizations yourself. And that code, can paste it into future chats or into cloud code and cloud code exactly knows how to create it.

Yeah, good point. Fascinating. you know, I only did half of my plant interview with you today. So my other half is about your product and your business. Can we do a session for that?

haha I'm happy to do it now as well if you have time. Otherwise, yeah, we can do another session.

Let's see. Did you have your dinner yet?

Um, no, I'll have to do that in another hour.

Okay, let me see. ⁓ We can we can keep going if you have time and then I'll just I probably I probably would have to have to split them because it's too long for one episode. Alright. Okay, so great.

Yeah, yeah, let's go.

That you shared how you're working. let's talk about what you are working on right now. So last time you were head of engineering, as you said, now you're running build your UI, right? What changed? Tell me about what changed.

Yeah, yeah, yeah. So I'm running now buildyourai.com. So buildyourai essentially is aimed at heavy industries, industrial organizations, construction, automotive, energy, aerospace, infrastructure, so those markets. The main reason I focus on this is because I'm a chartered mechanical, chartered software engineer myself with the engineering council in the UK. So these companies listen, they basically want someone to have enough AI expertise to guide them through the process. Most of these companies have this pressure to adopt AI. They know that their competitors are doing it. They know that other sectors have done it and there is good return on investment, but they are confused with, they're facing a lot of pressure in terms of clarity. Like what is exactly AI? What can it do? What are the limitations? What happens if things fail? And how can we do it safely? Right. They are very risk-averse and they want to make sure that they do it properly. And at the moment, that's the main reasons they are like two or three years behind AI adoption curves with the rest of the markets like IT finance, retail, advertising, creative industries have adopted this much faster ⁓ because they normally have the talent and ⁓ normally it doesn't require as much risk. with, mean, finance has regulations as well, which is good. Whereas with most of the engineering sectors, if you make a mistake, it can have a lot of problems. So I created Build Your AI to essentially, first of all, help increase the awareness in the sector about the technologies and how they work, their limitations and capabilities. And I do that through training through professional institutions that they trust, like Institutional Mechanical Engineers, Engineers Island, those type of institutions. And the second one is to ⁓ provide consulting services and advisory services so that they help them ⁓ map out and discover use cases, build pilots, and scale those pilots that are proving successful. So that's the kind of the current business model, in a sense, for the industry where we generate the revenue. But I'm also, on the same time, for certain areas where there is a lot of demand or where I think there needs to be a really good product, like in engineering, we have a lot of software that have been around for years.

I'm curious. Yeah, keep going.

They're not innovated. have been developed by incompetence and it's not innovated yet in the AI age. And I think in these areas, I am placing bets on products where I think these can be very successful if you integrate AI correctly as an AI assistant layer. So one product I'm working on right now is an enterprise requirements management software going after large, mostly aimed at large construction. and engineering projects to help them with analyzing their requirements, managing their requirements, preparing material pack for stage gates and where you have to comply with certain regulatory requirements where you need to produce a lot of documentation to show you understand the requirements well and you have verified and validated against them.

Mm. I'm curious, what would you say is the number one misconception these engineering executives have about AI? They want to adopt it, right?

Yes, they want to adopt it. And I think that a lot of these engineering companies at the moment, some of them are quite good. They have tried going deeper into AI systems and building their own. Some of them are enrolling the AI very slowly to their employees. And ⁓ their biggest misconception is that they cannot use AI because their data is not safe. or the data is not secure, or they don't know how the data will be used. Some of them only started adopting Co-Pilot. And Co-Pilot is the minimum amount of thing you can do to start using AI. It doesn't differentiate you from other companies, other companies or even companies in the engineering sector. It's the minimum amount like everyone is enrolling their employees into Co-Pilot now. And Co-Pilot is not even very good. It's not exactly the best AI tool we can use. Some companies even create their own private AI assistants that are integrated with more internal databases for better workflows and better context provided to the AI assistant. So that's where most companies are. They're just enrolling their employees to copilot. And most of these employees have no idea how to use copilot because they played around with it a bit and they found it like cool, it generates some text, but it's not necessarily exactly the quality that they want. Whereas it's more about like understanding the right way to create, customize it, and to personalize it, and to provide good, effective prompts to it to make it useful.

But I would think that that's not an executive would think about, this is the ROI of doing this project. Just letting my employees having access to Co-Pilot. And turns out majority of them are not really using it. Then it's kind of like a wasted effort.

Yeah. And some people who are more pro AI in the company, they may end up just using their own personal chatbots like ChatGPT or Cloud, which normally most companies are not using those systems. Most engineering companies, or pretty much all of them, they are using Microsoft 365 as an enterprise platform. So Copilot is the only approved chatbot that they can use.

It is the gap for adopting ⁓ AI more at a deeper layer, ⁓ the ability to spot real use cases. mean, Copilot is an easy, low-hanging fruit.

If you use it, you can still get good outputs from it, especially if you build your own custom co-pilots inside. Each employee maps out where they spending most of the time, where is time intensive, where it's repeatable or things where they can avoid tasks they can avoid. They can create very narrow customized co-pilots for themselves for very specific workflows. And that can be a massive time saver for them, especially if they have to do weekly reporting, quarterly reporting, like they need to produce something every, every, every period of time. Those things can really help these employees save a lot of like admin time. And I don't know, I think 20 to 30 % of their time engineers normally spend it on these type of admin work and an actual value at engineering. I was done by McKinsey, I believe.

Okay, so. So that's mostly employee productivity. That's the use case.

That's one part is employee productivity. The second part is integrating AI within like adding more custom AI solutions ⁓ and adding more, ⁓ going deeper into AI systems and building systems internally that works on top of their internal databases, integrates with internal databases and systems and documentation, providing them with better intelligence ⁓ or providing them, providing their products with more value to their customers.

Which one of these use cases do you see executives are actually willing to pay?

⁓ It's mostly where they can unlock additional value by integrating with systems that ⁓ are just a thing they're not getting much value from. So one of the main areas where a lot of engineering companies are looking into is predictive maintenance, for instance. So some of them have installed sensors on the production lines in manufacturing companies, but their equipment fail. And they want to understand when they fail or even predict in advance when those fail. And they can use sensor data to use AI to put their make, those predictions. The thing is most of this data is very bespoke, which means it needs a bespoke model built on top of that. So you build a machine learning model on top of it. You can't just place the data into core pilot and get it to work. Core pilot is not the right model for it. And there's a lot of data and The other thing I tell engineers is that you guys are sitting on a lot of custom bespoke data that only you understand. You need the domain expertise to understand that data and to build custom AI models on top to extract additional value and insight from it. You're not going to find off the shelf tools and consumer facing tools that solve that problem for you. You need to build your own bespoke. So that's where the leaders and the SLTs of these companies will probably will be spec allocating budget to discover what these use cases are and actually ⁓ develop pilots and scale them. So a lot of manufacturing companies, like for instance, ⁓ fiberglass company used the computer vision to detect when the fiberglass production line would fail and predict seconds in advance when they need to stop and fix it. Automotive companies have ⁓ used AI to predict in advance when their honing rings would fail when they're producing gearboxes. And these kind of like scenarios, essentially. Predictive maintenance is very high adoption at the moment where companies are looking into. But their biggest challenges, I think that O'Reilly found with these, with most companies is the use case identification, explore, like what use cases would be good. Most of them will have problems around that. And normally that comes from education. Like you need to understand different types of AI, limitations and capabilities. to help you with ideation. Number two is lack of quality data. So either the data isn't exist, they need to install sensors, or the data is not in a good shape. And number three is talent. They don't have any capability, most of them. Most of them haven't looked into allocating budget to create data science teams, or AI engineering teams, or even upscale existing engineers to become AI engineers, to help them build those internal tools.

Do they need to? ⁓ go ahead. Finish your thought.

⁓ I would say sometimes it makes sense to do it, especially if you want to be very serious and you want to do a lot of experiments. Sometimes you want to allocate, you want to basically up contract it to specialist consultancies to build it for you. But you do want to invest in that areas because it may help you save a lot of costs and maintenance downtime or helps you stay more competitive on the marketplace by increasing the value of your products, by integrating AI into it.

Do they need to see some established use cases before they'll move forward with investing in that? Because I can see that you're saying that you have to meet in the middle, right? They have the domain expertise, they know their problems, but they may not know how AI can solve that.

So one of the things I help them with is normally when I'm establishing the awareness and understanding with their teams is ⁓ these days I used AI deep research to basically research their entire market and their competitors and ideas for how they can start using ⁓ AI for various different use cases. In the early stages of where AI was still being developed, there weren't many use cases. You will find a couple of companies here and there that started adopting it. But these days there are pretty much a lot of case studies that can be found online ⁓ where engineers, engineering companies have started adopting AI across various different scenarios, starting from transport for London, all the way to Network Rail, all the way to manufacturing companies like Rolls-Royce, Boeing, all of them are starting to use AI, even in defense, even in ⁓ automotive or construction. Like in construction, AI is being used to help you predict project runs and cost overruns and things like that.

That seems like natural, ⁓ you know, involvement over time that people are seeing more use cases being successful, right? And then they feel like, you know what, we can do that too, right? Why don't we explore opportunities there? So that's maybe how it's getting on the, know, closing to the top of the adoption curve these days. But we still, you know, keep keep hearing most AI prototypes never make it to production. I've seen this use case to be successful, like predictive maintenance. But we can kick off an AI prototype solving that for our company, ⁓ but the majority still die. Are you still seeing that? Do you have to convince a CFO to fund something when this is the common sentiment. I'm guessing they may be on the fence, On the one side, they've seen success cases. On the other side, they hear like a lot of these fails, you know, what's happening there.

So the thing is business is mostly just a series of experiments. Sometimes you do need to experiment with novel ideas to innovate. Like what R &D does basically is sometimes R &D just comes up with ideas and experiments and you've got to invest in them. And some of that will not be fruitful, but you can't innovate without risk, taking a risk. There is always a risk involved where it may not be fruitful, but

Right.

With more experience of doing these projects and from best learning, best practices from other companies and other sectors, there are checklists and there are things you can follow to make sure that your projects has a higher chances of success. So for instance, yes, you can look for research papers and you can look for existing case studies that exactly has done what you're trying to do or in a similar area, but maybe a different market to give you more confidence. The other one is you can always limit the cost. You can always limit the scope and the experimentation until you're confident that it's good return on investment before you go forward. So one of the things I tell companies, for instance, is to use agile instead of waterfall when you're scoping AI projects. So instead of like sitting down and defining a scope and then allocating team and budgets to it, I would suggest to decide on a team and on a time and play around with the scope. within that timeframe. So there's only three of us, you have one month, what can we do? Like we can play around with the scope all the time. Like as we learn new stuff, we may change the scope all the time. So the scope is kind of like all over the place, but at the end of every week or every sprint, we have to have some kind of finding. So any finding, the way it says, this direction wasn't good, that direction wasn't good. Those are still good findings. Those are still money well spent. You basically found out that like this approach doesn't work with this data.

Hmm.

But that approach isn't good. So we may not go down that route at all. And we saved a lot of money for ourselves trying out an idea that doesn't make any sense to any business case for it. ⁓

So what you do, I'll go ahead and finish your thought.

⁓ Yeah, like that basically there is an approach where you can just limit the cost whilst you're finding things out. Finding things out is still valuable because it tells you which direction to not go down and which direction to spend more money on. The other thing as well is with all of these companies that have failed with their AI agents, normally the main problem is they haven't done this initial discovery properly. They rushed into... trying to build an agent and productionize it. They didn't ret-team it enough. So they didn't spend a good amount of time during the pilot stage and prototyping stage, making sure that first of all, ⁓ the edge cases and the bad cases are handled and that there is good return on investment. And number two, that the actual the users will adopt it. So they didn't provide enough of a... support and infrastructure to make sure that there is user adoption. Because if you build the tool and you spend all this money and users don't use it, that's a waste of money. And that's normally where these agents fail. those agents don't have enough security layer, they don't handle edge cases well, they're not reliable enough, they're not production grade. Or number two, they don't have an AI adoption strategy. They don't have a strategy how to get people to start using it. Some companies just enforce the usage. That's one strategy. Some other companies like try to do webinars and training to encourage people to use it more often.

It seems like this product mindset set, the product mindset is always needed everywhere, whether you're building a product facing external customers or internal customers, the same thing, right? Is this something that you're helping with, like the strategy of planning? How do you scope, yeah, projects and how do you get the best ROI?

Exactly. Yeah, exactly. So like in May, for instance, I'm going to be helping a company as a manufacturing company. So I'm going to be helping them 10 days to meet with discovery sessions to figure out like for each of the ideas we came up with, what, how, what's the best way to approach them. So figure out what's the best project management approach and what's the best way to explore the data sets and like the next steps and plan it out a little bit more essentially. So. It's ⁓ most engineering companies, they're to waterfall because most engineering projects, you've done them in the past, like building a railway A to B is always the same, right? Like you build a railway from A to B, you need to do certain amount of things. There are lots of standards to guide you and you just, as long as you're compliant with the standards and you have a couple of senior engineers with the experience on the project, their risk is semi-limited. So you can essentially plan waterfall and manage the project in waterfall format. have stage gates where you ⁓ do business planning, requirements planning, number two, you do preliminary design, then detailed design, then construction, then validation, and then you commission. And then maintain, then decommission. Essentially this step-by-step approach in terms of waterfall. But with AI, it's very iterative. Like with any R &D project, you don't know what you're doing. It's like you're in a dark room. not well lit with normal engineering projects. Sometimes the room is lit so you can plan ahead. with AI, it's mostly a dark room. No one's done it before in that context. So you have a lantern in your hand and you need to walk the room. Until you've walked the room, you have no idea where the walls are, where the table is, where the door is to exit the door. And you're looking for exit. So you got to walk the room and you don't know how long it's going to take. You have an idea how big the room is, but you have no idea how to walk it. So you may have to hit a couple of walls before you enter each end of the room. That's basically the way to look at these AI projects. You need to allocate a certain time and try to keep iterating really fast until you get to the endpoint. And at the end of each iteration or sprint, you do need to have some deliverable to show the business. So to give them more confidence whether to invest in further sprints or not.

How many days or hours are you engaging with this specific company for what you're doing?

So basically the way you want to structure this is you start with low risk and gradually increase the risk. So you start with low F low commitments where because most of this is a new area for most common most engineering companies. You start with training and awareness, but then gradually you shift into discovery and you want to just basically keep exploring and make sure that it's not a lot of this cost effective for them because they're just experimenting.

Mm-hmm.

But once they start seeing return on investment, they start seeing having more clarity around the steps ahead, that's where the risk reduces because of that clarity. And that will also allow their senior leadership teams to essentially allocate more budget because of that clarity.

So. So you're hand holding the team. You're not developing for them, right?

⁓ In the early stages, it's more like building with them. It's more like experimenting and discovering with them. But in the later stages, it could be building for them. So it could be doing it for them. So it's a two-stage model, essentially. Yeah. It's where you build for them. And some of these don't need the first two stages. So if a company is completely new to it, they do need the first two stages. They do need the building awareness and understanding, and number two, mapping out use cases.

got it. Okay, it could be long-term project. Mm-hmm.

the discovery phase, but some companies are already done that stage, either through other trainings or through other approaches or they are dead or the senior leadership team are more well-versed. With those teams, they do just want to get started building or piloting. So with those stages that you can just basically engage them and start building straight away pilots and then experiment and make sure those pilots are successful, identify if those successful and then scale.

Mm.

beyond that into actual production grade.

Yeah, this type of consulting is hard to scale though, right? You can take on how many clients per year for doing this.

⁓ So with any consulting model, essentially it's divided into multiple services, depending on which client is at, at each stage of service. Any consulting model normally scales either by providing software for that page, for that service. So that could be one way to scale through software and technology or through a head count, through hiring, hiring staff. is a new role called forward deployed engineers.

Mm-hmm. Yep.

So I've seen a couple of VC backed firms, especially coming from like Andrew Horowitz and A16Z. Those startups, they also operate in these sectors. What they do is they hire a lot of FDEs or forward deployed engineers. These are essentially engineers who are good well versed with AI and they can build AI tools. These engineers get embedded in the teams of the client organizations and they essentially help them build pilots quite quickly, iterate and scale their successful pilots. That's how you scale. with a business model like this.

Which one is your your past your plan to pass? Are you planning hiring? Are you?

Ideally, want to ⁓ scale through software first. So ideally, I think where the market is at, especially in the UK, you do need to increase the market understanding and awareness of AI. I know all of them want AI. There is a demand there. Helping them understand and have clarity on the AI and then gradually move them towards the pilot stage. Once there is a lot of demand for the pilot stage and it's becoming a bottleneck, use software. to release capacity. So you can provide them with software, which is these days easy to make, as long as you understand the requirements for the product, and then use forward deployed engineers to scale.

And yeah, and speaking of product building product building software, I think you already have a prototype that you can talk about. Sure. Tell me about it. Like you already have something right?

Yes. Yes, I do. So how should this work? we share screen?

Yeah, you can share screen and tell me about it. Right. And I'm zooming you. You came up with this idea. It's from your training work, your advising work.

⁓ Yes. So let me see if I can. Can you see my screen?

Yes, it's coming up. me one sec. Yeah, can. Yeah. UK real, there's a real requirements management system. Is this specific to the real, the real industry?

So I'm hoping it down to DRAIL for now because I have a lot of experience in DRAIL. And essentially ⁓ what this tool does is it provides engineers and systems engineers in the rail industry when they're managing massive complicated systems like railway systems. So the kind of systems I'm talking about is like building a railway from location A to location B. You need to think about the trains, the stations, the electrification system, drainage system, the F works, the tunnels. Can you see like the highways level crossings? There are so many things to think about the overhead line equipment, the signaling system, all of those subsystems can break down into lots of many systems. So what this tool essentially does, it allows you to get an idea of the complexity. So. It's AI assisted, but AI doesn't do the work for you. What AI does, essentially allows you to understand and analyze what you have and to help you expand on it. So you can start with essentially defining your system breakdown and see how your system breaks down into subsystems and how they break down into components. And then later on, go and start writing down the requirements for the project. and understand how these requirements break down into sub-requirements and sub-sub-requirements or component requirements.

Is this related to one specific one real use case a real real real way company are using this right now or or are these mock tasks? ⁓

These are just mock. These are just mock dating. This is more like at the moment at the mock level validating the product requirements. So building the product is one thing, but essentially the most important aspect is identifying the features and the needs that the tool needs to solve. Right.

Who are your customers for this product?

The customers will be engineering organizations and consultancies who are essentially delivering ⁓ the project. So for instance, an engineering organization might be designing the entire system for a client, for a large client. So let's say the project is to build a railway and their client or the sponsor of the project will provide, will give that contract to a design consultancy.

Mm-hmm.

to an engineering consultancy to design all of their subsystems for them. And these design consultancy have like different teams, like a electrification team, signaling team, rolling stock team that design all of the various components. And there is a team that manages all the requirements. We call them requirements manager or systems engineers. And they essentially help by talking to lots of experts within the company. They will identify the requirements for the system. So they help break down

Okay, and then they will come up with something like this.

bigger requirements into smaller requirements. So for instance, if a requirement says the train system shall maintain safe operation of 200 meters, maybe the smaller requirements will be the positioning shall be accurate to within two meters, the automatic braking shall activate within one second. So breaking these things down and all of these requirements, you need to look to make sure that you don't have any gaps, you don't have any conflicts, the requirement is specific enough, that it has good quality of writing quality and that you can, when it comes to later down the line, when you want to start building things, you need to make sure that all the designs are verified against them. So you don't have any requirement. And at the moment, there are big tools like IBM Doors, Dynamo-Graphic Oriented Requirements System out there, that are very old. These are very old tools that have been used quite a lot. But all of these clients, they want to move away from that into more modern tools, especially provide AI assistance. ⁓

Mm.

So that's essentially what this tool allows you to do. It allows you to go in, ⁓ have AI help you generate acceptance criteria, help you understand the confidence score and the quality of the model, help you check the quality of the requirements in terms of clarity, completeness, correctness, verifiability, all of those things. Identify which subsystems have quite a thing, create baselines against a set.

Okay.

So as your requirements change, understand, for instance, how it impacts the entire project.

Okay, so this is essentially assistance for a project manager, right?

Yes, it's for a systems engineer, but the project manager can also use it as well. So a project manager, like a product manager, let's say on a software project, instance, can also use it as well. So for instance, if you have a very complex software, let's say you're doing software for banks, regulated environment, very complicated software, lots of requirements, and you need to make sure that all the requirements are met. And as your system keeps evolving and requirements keep changing, understanding how

Right.

the rest of the system gets impacted.

Right. So currently you're saying these, what's the output of this whole thing?

One of the things for instance that it can produce is first of all on this helping you understand for instance where you have gaps in your requirements like in this view for instance telling you that you're missing this set of requirements Helping you check compliance against standards that you need to comply with Helping you understand conflicts where one requirement says you need to be a and another one that says that you need to be B Like make sure that you're not conflicting and the other main major bit is assurance. So for instance One of the things that this tool can help, for instance, is to assign stakeholders to certain requirements. So it suggests to you that this person is electrification expert. This person needs to provide documentation that says we've complied with this requirement, electrification requirement. And all that person do is once they log in, they only see their own requirements. So like this person, like this is Emily Brown's requirements.

Mm-hmm.

And they need to submit evidence that they have complied with this requirement. So when they click, press Submit, they upload their designs and documentation. And AI essentially helps them with making sure that that requirement is closed and complete. So it helps with compliance checks.

I see. Yeah. This is. I see. So, so, so here's my understanding. So let's, let's say the system engineer or in the product product or project manager are working together to realize, to build something that's quite complex. And then they have to have this basically a map or the blueprints of what they're building. And then, right. And in order to, and this whole thing that coming out as a guideline or a blueprint for building something really complex. can be, let's say a thousand pages big. I'm exaggerating. don't know. I don't know how long these things are, but it should detail all the specs of every component, sub-component that needs to be built. And in order to get the information, the system engineer manager or whoever this is, need to interview the stakeholders and get the requirements.

Yes?

So

Yeah?

they know what needs to be built and then they are compiling this whole thing together. And every detail and specs need to be checked against standards, policies, regulations, right? Industry standards or the requirements and meet the stakeholder needs as well. And this is a huge collaborative effort that needs to happen before things are built basically.

Exactly. exam. Yes. Exactly. And most of these workflows are done in Excel spreadsheets and emails at the moment with only one tool, which is called Doors. ⁓ Basically just giving you a view that looks like this, just your list of the comments in a spreadsheet, but nothing else. It just allows you to record the comments in a spreadsheet, add details in it. And that's about it. ⁓ And normally what they do is every time that they want to make sure that a very requirement has been met, they create an export or create baselines. They sent an export, is an Excel spreadsheet and emails. Those emails go back and forth. There's a stakeholder provides some compliance statement or evidence. Then that goes back and forth a couple of times because it's before it's finalized. And when the requirements change or new requirements get added, that whole process needs to repeat. And it's all done in and spreadsheets and emails. So this is more, this becomes more like a single standard source of truth for everything. So it allows you to. check how many of your comments have been met, gives you a dashboard high level view, check which ones are at risk, which ones has no evidence, which one has pending review. It allows you to check for gaps and conflicts in requirements, check their quality exposure of your coverage analysis. It allows you to check the requirement quality across all your requirements. And it also allows you to essentially check, for instance, the traceability health. Like for instance, have you under specified or over specified certain subsystems ⁓ in different views? And essentially the main purpose is assurance. So at every stage gate, you want to press a button, it produces you a pack that you can then use for your meetings. So for instance, we are going to go into construction now. Are we done with detailed design? So press a button, produce a pack, and it will provide you all the documentation you need to sit down and actually make sure that the stage get can pass. And if not, then you can go and check like where you're missing things, whether you need to break down your system a bit more, define requirements. Like for instance, in here, you can see that we have one requirement for traction braking, but no requirements for traction motors or emergency brakes. And three requirements on power supply. So you can see like where your system is specified well, then check where the... ⁓

Mm.

experts have uploaded the right evidence and go from there. The other thing that they need to do is risk register. So make sure that all your risks have been essentially added, especially when you're performing a safety case and you need to follow certain standards for that. And those risks need to be allocated to requirements as well or for a subsystem. So again, checking gaps in there, checking that you have enough mitigations in place, making sure that there is no orphan risks. and also checking that all of your risks are going from high all the way towards low when you have applied the mitigation. Otherwise, they don't let you build the system.

How big is the market?

This. So ⁓ in terms of the market is mostly public sector. So public sector companies like Network Rail, like big they running big engineering infrastructure companies ⁓ and also small medium enterprises where they provide systems engineering capability towards those public sector companies. So private sector for small medium businesses or larger consultancies that are engineering consultancies that provide systems engineering ⁓ expertise to those clients. And some large public sector client organizations also have this. Five processes. Right now systems engineering practices where you are managing requirements are very big in aerospace, transport and defense. These are where it's used significantly and gradually manufacturing as well. But For the start, for the early stages, I'm focusing on your rail because rail right now is very mature in terms of using systems engineering. And I know from the market that a lot of big, big, big clients and big organizations want to move away from doors.

That's what they're using now. Basically, they're really behind. They're using.

Yeah, they're using a very old tool. They're using this tool right now called Doors, ⁓ IBM Doors, which ⁓ is used for commerce engineering. And it's a very old tool. It has some AI features added in, but most of the people who use it hate it in terms of how easy is it to use or not. And the other problem it has is that ⁓ a lot of people cannot give a view of the requirements to the stakeholders because as soon as stakeholders log in, they're getting on data with so many features that they just don't want to deal with it. So they rather do stuff in spreadsheets. So it has a usability problem.

Mm. How are you validating this idea? And what are you hearing so far?

So what I've done is essentially talking to a lot of systems engineers in my network. I was five years a systems engineer as well. So a couple of years in my early, my career, I was systems engineer. So I've been exposed to all these processes. I know like the pain points. So one is myself. The other one is I talked to 10 to 20 systems engineers in my network, my old managers and understanding which features they care about the most. which is at the moment, assurance is the biggest piece at the moment where they really like, where AI can really help with.

they're willing to pay?

Yeah, they essentially the way that this works is a bit more slow. You need to go through a procurement process with these organizations. So you start with a pilot, you go through procurement and through that procurement, you gradually get integrated in. they basically providing five to 10 licenses to engineers to can essentially try it out on project. with this type of software, because it's an enterprise software, you don't You don't really like making money by subscriptions where people can just come in and sign up. But in terms of the main point is you can provide two ways of working for small businesses. You provide subscriptions where they can sign up, let's say, a couple of hundred pounds a month for a user or with big companies. can essentially provide a package, an enterprise package. where you enroll their organization into it through a procurement process. But normally a sale cycle could be anything from a couple of months to six months or so.

Yeah, yeah. Are you looking for funding? I mean, I'm connected to with with investors here in Silicon Valley. can introduce you to some someone.

⁓ At the moment, my strategy is to bootstrap it if I can. So that the reason I'm running my business, first of all, as a consulting training business is because it allows me to basically have unlimited time and money. So all I have to do is keep building on the side. It will slow down the build, it allowed, but with the use of AI, it allows me to speed up. So AI can speed me up a bit more because I can plan the architecture and they do the work and I can get the AI to do bits and I can fix it to a hundred percent. But essentially I can, AI will be more like my additional software engineers, but because I have the idea of like,

Mm-hmm.

how everything needs to fit together. I can plan it on paper and I can get AI to do the actual coding and the grant work and I can make sure that the code is good quality that goes into code base and speed things up. And then if I'm running out of money, I can always do training and consulting on the site.

How many hours are you working per week?

⁓ I do my normal hours, like 40 to 50 hours a week.

No, not, not, not, no, not too scary.

Yeah, no, no, no. I make sure I just follow the good process and I don't kill myself 80 hours because ⁓ one of the things I did consider with this idea is that first of all, anyone can come in and copy software these days. It's very easy to copy, right? What they don't have is distribution. Most people don't have the distribution I have, right? And most people don't understand the system well enough to know what features need to go in. Right? So...

Yeah. Right. No. Well, first of all, you're going to release it. Right, right, right.

It's a very special software, right? So there are certain modes you can have for your business. One is having the right credentials and compliance with the regulations. That's one mode. Distribution is another mode, like knowing how to sell it and knowing the people to sell it to. Number three is having an understanding of what needs to go inside the software, the features that are relevant. And number four is essentially how fast you can get to the market and the other stuff. related to that and vendor, vendor lock-in. Like the other thing with this is because their sales cycle is really high, like it's not easy to sell enterprise to big companies. It's going to be very easy as well. you're the procurement, you're in something, they're not going to switch it up easily. You're in the rain and that's going to be another mode. So all of these modes added up. This means that someone from India or some other country cannot easily come in and vibe code something like that. mean, this is.

Yep. Yep. But it's also going to be very sticky, you know, once you sell it, I'm not going to be. Yeah. Right, exactly.

This UI and stuff is vibe coded, obviously, because I ⁓ basically been just getting the idea, building the back end and building the actual system to make it production grade, that stuff you can't vibe code. UI you can vibe code, but the back end is not easily vibe. You need to sit down and do actual software engineering to build it up.

Yeah, I agree. Most importantly, don't think anybody can just come in and sell to a real manufacturing company. I don't know what to say to them. What am going to say to them?

Yeah. You need to have credentials and need to have, you need to be ⁓ essentially have the right standards and compliance in place because they're not going to buy from you. You need to become an approved supplier first. And all of those things are modes. Now in terms of investment, I think I may look at, look at that a bit down the line once I have the product built fully up. So I have validated product. It's just building the product fully up and ⁓ like scaling basically when it comes to scaling.

Yeah.

But I think in the early stages, strategy is to bootstrap it by just training and consulting. And once the product starts becoming more, ⁓ what's it adoption, and it's generating enough revenue, then switching to just product, focusing on that.

Mm-hmm. Yeah, yeah, sounds like a good plan. Good luck with this whole journey. I know it's very hard, but it's worth it.

Yeah. Well, sounds good.

Yeah. Okay. I have one more question. I'm starting a closing question. I probably surprised you a little bit because I didn't tell you beforehand. So try your best. If you could recommend a book to, let's say, yourself when you were 20 year old, just graduated from college or yourself today, like building products, building this company and yourself, let's say 10, 20 years from now, like what would it be? You don't have to tell me three books. One or two is great.

So a book for a 20 year old who wants to build products, did you say?

Now the 20 year old you, someone who just graduated, let's say, because I think new grads have a tremendous confusion last year and this year because they don't know what to do, right? Even for software engineers. I've heard of software engineer grads couldn't find a job this year, last year. I think people are lost. And ⁓ what about yourself today? What kind of book would you read now? And what kind of book you would read in? 20 years.

So right now, a lot of the books that I got are more higher level books in terms of like, I don't remember the names of them, books around understanding how to balance your life and how to set life vision, those kinds of books that are more high level, rather than just technical books, related to business or technical books related to programming. It's more about like, Harder questions like this stuff that you need to sit down on like vision statements or write down your values and figuring out like how Like what are the five different types of wealth you want to have like time money? Social relationships that kind of stuff. So things that are make your life more balanced is basically the kind of things I'm reading now like Books on philosophy those kind of bits so high level But when you're a bit younger ⁓ What I would recommend reading is help things that would help you understand how to build a really good career So for instance, Daniel Chrisley's books are really good. So and revolution ⁓ then key person of influence over subscribed and 24 assets these these four books will essentially help you understand how business works ⁓ No, but I understand I've not fully read them back-to-back

I know. Did you read all of them?

But I know about all the models in them. So I know, for instance, the ascending transaction model when it comes to marketing or creating products. Or I know ⁓ Entrepreneur Revolution is about a key person of influence model, KPI model, which is about how do you make yourself such an invaluable person in your industry that you don't have to worry about jobs anymore, or you don't have to worry about opportunities anymore. People come to you. You become that person that everyone knows of. like this guy is the AI guy in engineering. People just come to you for that. And the importance of writing books, the importance of like publishing, like content and writing books. Like these people, like if people are new in their career and they start doing this, they create way more opportunities for themselves and they don't have to, they don't have to like apply for jobs. Jobs come to them. They don't have to worry about getting jobs. or even creating businesses for themselves. They just decide to create their own job instead of create their dream job, instead of trying to find one and trying to accept any low paying graduate position that comes around. And just learn business by doing business. And that becomes their MBA, essentially. That's what I would recommend reading. And for people who are engineers, I would recommend

It's really hard. Yeah, yeah.

⁓ essentially reading as many technical books as they can, especially these days, like on AI. So, ⁓ obviously read my book. If you want to build with fast API and AI services, but read AI engineering book by Chip Huyan, read the machine learning system book by Chip Huyan. those books are time lens. Read the, ⁓ read the, ⁓ machine learning, hands on machine learning with PyTorch and Scikit-learn. Like these are foundational tech books.

I read that.

that really helps you become good AI engineers.

Do you think those are still needed? I'm going to challenge you on that foundation books versus just adding the AI.

Yeah, absolutely. Yeah, because AI makes mistakes. How do you make sure that it's not made a mistake? It's by knowing the expertise. Like it's not going to replace expertise and craft. Like for instance, if AI tomorrow can basically, ⁓ good, really good paintings, do you still need to learn to paint?

I want to see a human paint. I care less about AI-generated painting these days. I don't care.

Well, paintings, yeah, with paintings in the future, handcrafted artifacts will have more value than any factory-produced or AI-produced artifacts. So any skill is still going to be valuable because if you handcraft something, like one of the things I would do with my websites, underneath the website, I say handcrafted by, ⁓ under the website, would say handcrafted or hand coded.

100. Created by human.

Yeah, created by human. Like even some businesses, like for instance, was it Ministry of Defense? don't remember which website was it or Royal Navy, RAF, Royal Air Force or Royal Navy in the UK. You go to their website, they say you will talk to humans. Like explicitly say there is no, we're not going to be using customer AI agents.

I love it.

And some businesses have adopted that strategy. Even though a lot of businesses are doing customer service bots these days, people actually want to talk to humans. Even myself, when I go and I ended up having to talk to a customer bot, I always say, talk to agent or talk to human. Straight away, the first message. And it bypassed entire agent process.

It's almost a backlash on like the sentiment of like it's not about how good your your child bot is. It's about I don't know to expect.

Yeah, so. No, not necessarily. AI, the purpose of AI would be to bypass bits that you don't want to spend time on, not necessarily to bypass the entire craft. The other thing is your mental health and your meaning of life will be based on the things you produce by hand yourself or practicing a craft. the enjoyment of your career and your work comes from the craft. I always, for me, the most energizing thing is coming up with models, business models, or coming up with actually writing code that builds a software. Like I don't need to build this software that I showed you right now. I don't need to build it. I could just sit down and just do consulting and training all day. That could just be my thing. And the consulting doesn't need to do coding. I could just do advisory and coaching and training. That could just be my business. But I love coding. I make sure that even ⁓ if I had to get a project, I'll do the coding.

and playing.

Or even if I hire, I would still be involved in coding. That was the case with my previous employer as well. The founders were coding all the time. They were involved in the coding process, even though if they swipe coded bits of it, they would always wipe code. So they would always code themselves. They would always be involved in the engineering part of it because that's where you have the most fun. They're designing the developing part and building part. That's the craft. And even though these days AI can do stuff, I think AI... I don't like AI to do my paintings for me. I want to paint for myself. That's the enjoyment I get. The writing process and the craft, the painting process, you get the enjoyment. When I wrote my book, I didn't use AI to write it.

Is it is Is it just joy or is it also pride?

It's identity as well. It's identity and joy. Like you want to feel like it's yours. Like you don't want to feel like a fraud. You want to feel like you did it. You want to be proud of it. You want to feel like you own it. And you want to feel like, ⁓ yeah, you want to feel a satisfaction from the pain of doing it.

So you take trade, right? Right. from the pain. So it's joy, pride and pain altogether that discrete this dopamine hit in our brain that's truly satisfying.

Yeah, it gives you the meaning. Gives you the meaning. Without the pain, all the happiness would feel meaningless. Without the pain.

This conversation is so deep. Yeah. was fascinating. Yeah. You want to turn off the screen sharing? Now we can close it out. Yeah. This is amazing. So I can see the whole you before. Yeah. Okay. Ali, thank you so much for coming back this time. mean, last time you gave us the playbook of building production AI systems. This time we went really deep. I feel right.

Yeah, yeah. There you go. Right? Yeah, we went just about wipe coding this time. went all the way down to velocity of using AI.

Yeah, I know. Philosophical stuff, right? And you showed us like you shared your stack, how you automate your work, how you train companies, how you build a product on the side. And I really appreciate you share your journey and all these philosophical nuggets that you're sharing. feel like this is very meaningful, you know, for all my audience and for average people, whether they're engineers or not. Right? Use your brain I feel like use your brain. Okay, what's your last ending note?

My last ending note would be to the technology that the AI and everything that's coming, it's going to come here, it's going to stay where it is. We are at a new revolution. Previously was industrial revolution. We now have AI information revolution. There is no reason to fight it, but essentially you need to accept that it's here to stay and it's going to change the entire workforce and the industries forever. The best thing to do is to essentially plan ahead and to adopt it, to start understanding the limitations, the capabilities, the benefits of using it and create your own mind about how you want to use it, how you want to ride the wave with the wave that's coming. There is no point sitting when waiting for the wave to hit you and get ground. You want to plan ahead and surf the wave. And that's how you get the most benefit for your careers and for your life from this and start like becoming more essentially becoming a more capable human because of that technology.

Yeah. Yeah. Yeah. And it's balanced. We have to be also balanced with using your own brain as well. Right.

Everything is about balance. Life is about balance. We want to have a balance about even this. So I said about the bad stuff, even though I'm teaching AI and stuff and promoting AI, I promote the limitations and the bad things about it more than the good things. I always tell people, if you can do it without AI, do it without AI.

Yeah, yeah. Thank you so much. It's a fascinating conversation. love this. Well, don't hand up yet.

Thank you very much. That was really good.