Managing Machines vs. Managing Humans. What's The Difference?

Large model is still less random compared to human. Large model listen to you, like follows your instruction, the human are not. So it's still easier to organize.

We know that LLMs are nondeterministic. Wouldn't so many agents cause error compounding just throwing human bodies at a problem may not be the right solution.

100 agents than 100 people.

Tell me about Apodex. You moved from academia to the industry. What kind of product is it?

We're not just building a chatbot which can have some nice conversation with you. What we really want to do is to solve very hard problems, scientific problem, economic problem, something that can create a lot of value. Usually it is like a unsolved problem. So we are trying to use our heavy solver to solve the hardest problem in developing large model. They are you know creating a loop. We are self evolving to become a stronger and stronger model.

There are a lot of fresh graduate or people who wants to transition their careers. They're very confused. How do you think prepare us towards a positive outcome?

AI can accelerate this work 10x or 100x. If you have some open source project, especially agent, it's very hard to get resource to train a large model. But for writing agent framework, you only need APIs. You can still build a very comprehensive strong agent framework just using API. You don't need that many people to do the same kind of work.

Today I have the chief scientist from Apodex, Simon Du. He was a computer science professor at University of Washington and now he's building deep reasoning models and research agents that can beat the frontier labs. Simon, welcome.

Hi. Hello everyone. Very happy to be here.

When I was doing research about your background and your company my research said that you're one of the most decorated young theorists in machine learning and the honors you stack include Sloan Fellowship, Samsung AI Researcher of the Year, NSF CAREER Award, Pioneer Award, Intel Rising Star, MIT Tech Review Innovators Under 35 and AI2050 Fellow and so many you know, this is so amazing. But I'm curious, how do you introduce yourself to a new friends?

Well, I'm just like a faculty in University of Washington. So, I like to work with young people, especially smart young people to work on hard problems. That's what I'm doing right now.

I love it. Tell me about Apodex. You're now you moved from academia to the industry and you're the chief scientist at Apodex. What kind of product is it?

Yeah. So our vision is to build heavy duty solver. So we are building a large language model but kind of different from like normal large language model for example GPT, Gemini or like Claude. What we are building is like we are very focused on solving hard problems. That's why we call our model heavy duty solver. So we're not just building for example a chatbot which can you know have some nice conversation with you but what we really want to do is to solve very hard problems like scientific problem or economic problem or something that can create a lot of value it can be usually it is like a unsolved problem so that really require a very strong solver to solve it so that's a general vision of our model and I would say generally speaking to solve a hard problem. The first step is to do so research. So our current product or the first product we just released really focus on deep research.

So deep research is a mode in actually several large language model products like ChatGPT Gemini they all have their deep research mode but this is our fourth focus. So we are building we will say even better deep research product which can give you a more complete and accurate research report than those like deep research output from GPT or Gemini. So that's the first product we have. Yeah. So this is a general introduction of what we have.

What do you think is the problem this model that you're building is really solving?

So we are trying to so we are building a general heavy duty solver. So we are not exactly targeting one spec very specific domain but we do have some target domain that we are considering. One is the I would say drug discovery drug repurpose healthcare this kind of medical domains. So we believe there's a lot of unsolved problem there of course and also it can create a lot of values from there. So that's one targeting domain we are seriously investigating. So in our team or we have a sub team which we call heavy duty discovery.

So this team tries to identify the hardest and challenging and most valuable problem that we need to solve using the strongest AI model and for now they are researching the hardest problem we see in this biomedical domain. So that's the particular domain we are very interested in. There are also another domain is basically the large language model or AI development itself. That's why we always focusing on self evolving which is one of the goal of our company. So we are trying to use deep research or our heavy solver to solve the hardest problem in developing large model. So in this in this way we are you know creating a loop. So we are self evolving to become a stronger and stronger model.

So we hope we can you know exponentially expediate the development process throughout you know the of the AI development. Yeah. And so those are two I think we are domain that we will definitely spend a lot of resource on but we are also considering other domains. For example material science we're also considering making new materials. Also generally we are trying to predict the future which most relevant there is as is like finance for example you can predict the stock price of course you can create value yeah so that's yeah that's another domain that we are interested in

Yeah that's amazing so you have you basically have different you focus on different domains and these are deep research heavy domains like finance and biotech material science and you also have you also are building the model in such a way that It's self-evolving. So maybe let's separate these two. First let's talk about domains, right? What do you mean by you know you're focusing on deep research in specific domains. Does it mean that your models are specific?

Not exactly. So we are building a general heavy duty solver. For now it's a general deep research agent. So not only about the model itself also there's a agent system around it I may talk about later. But in training this model we are not targeting on specific domain. I mean we do have for example training data from all kinds of domain including biology, physics, chemistry, engineering, finance and so on but we are not say you're not like say 50% from b not like that it's kind of evenly distributed data across all domains but also about the knowledge.

So that's important but a more important I'll say capability of this model for deep research there are okay there are several key capabilities one is planning so one is you want to do some research you first need to make a research plan so that's called plan planning step so we do have specific data for this planning step like how do you decompose the problem into sub problems and you can attack each sub problem individually or separately and then so how to decompose a very hard problem into sub problem and even you can decompose sub problem into sub sub problem so that's is a planning and or decomposition capability

So is your planning phase you mentioned you have data for you know specifically planning and decomposing a harder problem into small pieces is your data about specifically doing scientific research

So actually there are common patterns across domains once you want to decompose a hard problem. It's like once you so first of all for example you ask so first of all it will consider whether this problem is a specific problem or general openended problem and based on this judgment you will decompose problem whether you will first decide whether I need to search more information and then based on information further make plan or I already have enough knowledge in my model I can already make some decision to decompose problem. So it's really like how scientists attacks a problem. So we do gather this kind of data to help the model learn how serious scientists attack this problem or decompose this problem. So that's the first step and yeah we kind of gather those data to change the model's behavior to think like the a scientist.

That makes sense.

Yeah. Yeah. Yeah. And yeah so that's the first key capability. The second key capability is really about search.

So search is you know nowaday we think search is a very common you know everyone use it but actually it's still a hard problem once you want to search for some hard to identify information for example is the query you use in Google search matters a lot because you know different query Google will return different answers right or different order of the pages and you hope that your query is good enough to gather the most useful information in the first page That's how human when you do search it's like so we are we train our model to you know to do more efficient and thorough search to write better query so that we can get the most important useful information we want often times is about you know maybe some of the key information from some paper from some report so you need to first so it's called the multi-hop problem basically you need to first search for some keyword to gather the information, gather the report and then maybe you'll further do some search on that report or from that web page you further do some search to search for the next key information.

So this is a general search capability and there are some benchmarks to measure this and on those benchmark we can get probably the strongest result. So this measure just a key we call it atomic capability which is search to gather the hardest or hard to find information. So that's the second key capability. Yeah.

When you say when you say for search you beat the your the top of the benchmark that did you mean that you actually beat the deep research across the board like OpenAI or Claude you know those those deep

Yeah. Yeah. So there's a benchmark there are several benchmark one is called BrowseComp. It's developed by OpenAI. So basically they have I think 1,000ish problems. Those are very hard to find. Basically asking you to find some information that is really hard just by purely say prompting a model or purely do Google research.

Can you give an example like what is a hard search problem?

For example, something like find a book name by 1990s and with the first word start with zy something like this.

That doesn't sound like it can be easily solved by Google search.

Yeah, exactly. So you need to you need to have a multiple step usually and so you may first search for books by some missing some time range and then missing time you do another search. So it's like usually it's a multi-step search. We call it agentic search. So it's not like a single Google query search problem. We are building a system to solve to improve the search capability. So on this benchmark there are around 1,000ish problems. There are easy problems and harder problems and then harder problems sometime we can solve we can find the answer but they cannot so there I think and overall score we are we are better so that's why we are we say you know we are we are the state of the art or SOTA we have the SOTA performance so the reason we can get this strong performance is really about so there are two things one is about the data like we are building certain data set so that we can improve the general cap capability of search and secondly we are building an agent system around our model.

So we call it agent team. So it's a multi-agent system to do not only about search but overall this deep the whole deep research process from the planning to search and then there's also a last step which is called summary. Basically you summary all the information you gathered. So for this system it's a multi-agent system which is a whole system or agent system right now because now I think every LLM company they have something like this like a multi-agent system basically a single agent is what we called ReAct mode reasoning and then act that's called it ReAct mode it has its limitation because one of the hardest or like bottleneck of a single large model is context length no now people often say you know we have a context window of 1 million or maybe slightly more.

However, 1 million is even is still not enough. So you need you know multiple agents so that they can process more information. That's why you need this multi-agent system. Okay. But this is not like you know super novel in the you know mid of 2026 because it's like maybe it's novel maybe half year ago but it's now I think now every company has a version of it. What differentiate from us is like we also in this multi-agent system building we also have agent to do verification. Okay. So if you think about it when you do search you know now know you can write anything on the website right so some of the result can be you know reliable but some are not for example generally speaking you know the information from wiki is like you know someone at least check it so it's more reliable than a random post from say Reddit so we also have specific verification called verification agent to judge whether this information source source is reliable or not. So but this verification concept is quite general. It's not only about judging a source is like reliable or not.

But this is very important of course but also about for example whether the reasoning from the source to a claim is correct or not. So this like judging the another agent's reasoning process is correct or not. Okay. So that's another kind of verification. There are other kind of verification which also we also do. For example, if it's about some coding task, you can actually write some tests to verify this program you write is correct. So we but what we emphasize here is that we do have an explicit verification agent. It can be you know judging the source, judging the reasoning process, judging the code you write is correct or not. But we have an explicit sub verification agent to do verification which actually is very important to improve the accuracy of the results we get either from the search either for search or for final deep research report.

But what you're saying is you have different ways of judging whether something is correct or not.

The most common one is we call it LLM as a judge. Basically we use basically just using our own model as a verifier to look at another or the output of another agent but the agent itself is still our model. It's just you know how we the role are different one for example one agents to process all the information gathered and then generates a claim and another agent is a verifier to judge this reasoning process is correct or not. The difference that their input are different. For example, the first agent reasoning agent input is like the search result. For example, another agent, this verifier agent is entire not only about the search result, but also the reasoning process and final claim the first agent generated. So the second agent need kind of needs more information so that they can judge whether the whole process is correct or not. So that's what basically what differentiate two kinds of agents but you know you do need the second agent so that you can have this verification step and the final outcome will be more accurate.

How do you handle the large amount of context?

The current I would say technology of larger model one of the key bottleneck is that a single agent or a single large model cannot hold a very very long context say more than 1 million. Okay. But if you basically do divide and conquer then you can decompose very hard problem into subtask. Okay. So that's why it's go it goes back to the first point I make like the decomposition or the planning is very very important. So once you decompose problem you don't need a very long context for a single sub agent. Okay. So for example the original problem it may take a very long time context or a lot of tokens more than say 10 million tokens to solve. Okay. However what you actually need as a input to those sub agent to a search agent it might be much simpler. Okay. For example, a single query to a sub agent for some search problem might just be find a book or I'll say find a math book in 1980 about calculus. So this very concrete one something like this.

So for this specific problem you don't need a very very long context. It's like it's a reasonably context should be you should doing a pretty good job. Okay. So basically the whole system is about decomposing a hard complex problem into small sub problem but each individual problem only require a relatively small context. So that as a whole you still have you still need maybe more than 10 million tokens but each a single agent doesn't need a very long context to solve the sub problem it need to solve.

We know that LLMs are nondeterministic. Wouldn't the sheer volume of you know so many agents cause error compounding or having or you know having friction when coordinating them? Because if we if we use the analogy of thinking about human organizations, right? Often times if you're solving a problem within a human organization, just throwing human bodies at a problem may not be the right solution. Right.

Yeah. Exactly. Exactly. So I think there are two things that we develop or we emphasize that are really important for this kind of you know how do you basically how do you organize a large group of agent or people first of all I would say that I think large language model is still less random compared to human their outcome is more it's more predictable than general yeah yeah I mean generally speaking you know the language model listen to you know it follows your instruction at least that's what they're trained, but the human are not. So I think it's still easier to organize like for example 100 agents than 100 people. Yeah. Okay. Okay. So there are two things I think those are important. First of all is verification. Okay. Maybe I can talk bit more detail. So verification agent one single verification agent it may more than one I'll say reasoning agent or search agents output. Okay.

So basically although you may assign a lot of tasks to different agents but we have several intermediate steps so that at certain step basically it's like the aligning or the sync it's kind of aligning meeting or sync meeting just like you know when you company but those verification agent basically acting as organizer of this alignment meeting in a company basically it gather all the information and then it will verify which one is correct and maybe discard those wrong ones and then it will kind of reassign the task for the next step. Right? So we just like how you organize a company you know while you have weekly meeting because you need to sync all the information from this week's output. We also have some similar basics like this verification agent will verify all the information and then we'll reassign some tasks to next step. So that's one very important step like verification is not only about verifying the correctness but also about gathering all the information and then you can decide what to do in the next step. Okay. And secondly it's about redundancy. Okay.

So in a in a company or human organization usually we will not assign the same task to for example two or more people although I know some company do to make some comp to make some competition you know but we are not doing like this so or in spirit maybe it's similar but you know LLM agent is still cheaper than real human right so we can let same agent or sorry the same we can assign same task to multiple agents and because the LLM they have some randomness in it. So it will have more coverage about the same task. We'll gather kind of more information than a single agent. Okay, especially for complex more open-ended problems. If you even if you use the same prompt to same prompt to the same model because of the randomness within large model, we'll still generate maybe different answers. And this kind of diversity is also important because you may gather more complete answers or coverage or complete information about the same problem.

So well it's it sounds to me like statistical sampling.

Yeah, exactly. You sample more. Exactly. You sample more. So it's not really like the you know about competition. It's really about I would say cooperation in some sense. Even though their tasks are the same but even you assign the same task to the same sorry different agent although those different agent are in essence they are the same model but they can still generate more complete diverse answers or information to you. Okay. So this redundancy is not new in computer science you know in system design cluster you know you often have this kind of redundancy but here is it's more about gathering more diverse data or diverse information so that you can process you can get a more complete picture about the same problem. So yeah so basically these two things one is verification one is this redundancy these two things are what make you know this you can get very complete and accurate answer to your deep research question

Are you using this teams of agents for your team's work?

Yeah I'm using this model of course kind of every day I think this especially when we do some deep research problem I think it's very useful and it's very I think it's super useful actually especially now we are is It's a quite generic problem. You can insert it for you know how you organize your team this kind of problem or some what's the right research problem or for example what's the right biomedical problem we should study and you can ask follow-up question to it as well and usually our Apodex model can give you a very complete comprehensive answer. So yeah we are using it every day actually.

Can you share an example question that you use for your day-to-day building a larger model?

One important aspect is about data right but how do you process the data? How do you label data or how do you add some tags is actually what kind of tags you need to add is a research problem like how do you or basically how do you organize data in the right way. So we often search this kind of problems. So this is just one example and also you know you can also do some research and building large model architecture, optimizer you know you can also do this kind of research as well.

So if I were to compare this research question and answer it from a human perspective, a human only equipped with ChatGPT or Claude versus a human researcher equipped with Apodex how's the how's the workflow different or you know I have one question I write a prompt I pass it to either ChatGPT, Claude and Apodex and then what's what's going to happen next is Apodex running longer uh

It depends on the problem. Some problem we do run longer some problem maybe run as quicker time but usually I would say if you have a complex problem okay then we will our system will assign more sub agent right to make sure that the information we get is correct so we'll cross check a lot of evidences claims so for that because our this whole verification and redundance agent with redundancy system sometime especially for long complex problem we can run longer the interface very I would say it's very similar to deep research model of ChatGPT and Gemini basically you ask some question it will generate a report so this interface is very similar it's just that in the process we emphasize more on we do a lot more verification search or some research with some redundancy there

It's like a science fiction right you're asking basically an AI to predict the future

People always want to predict future either for money reason or whatever right so predicting future I would say it's not the entire especially in finance it's definitely not a new problem so I also work in a quant company before you know quantitative finance where they are basically trying to predict the future like every even every second if you are doing high frequency trading so but it's very different from what we do okay so for finance when you try to predict future usually the most informative information is about is from the market data basically the price and volume the bar chart you know the chart you see on those finan like Robinhood whatever yeah so but those are very informative I is it does have predictive power for example you can predict the next second next minute next hour or next day you know the longer this horizon is the less accurate it is okay so this is like I would say the more or less a traditional or yeah or classical

Quant research is about we're trying to use the market data to predict the future price and you can predict you know better than 50% accuracy then you can generally speaking you can make some money okay yeah so but what we are doing is different we are more our difference is more like the fundamental analysis of in finance okay where you analyze this company's outcome based on for example it's upstream downstream sector or company industries and so you have a like holistic view of these companies for example next months next year's earning whatever so that's what you are trying to predict so this process is really so now you know different financial institute they all have their their researchers and you know fund managers doing this kind of research trying to gather more information about the company about the sector about the whole industry so that can predict in the future what's it like what this companies will you know sell price go up or go down.

Okay. So our basically if you think about it this is exactly what deep research is about. You are trying to gather all the information about this company and then you're trying to predict the future of this company. Yeah. So this is so I would say you know deep this is predicting future in for but this kind of predict is really about deep research. So but this you know goes beyond this finance you can also predict for example the outcome of a match of a world cup right because you can predict the outcome of a sport game is also about first gathering all the relevant information for even about for example whether the player is strong or not whether how good their cooperation how good is the coach whether some players are injured or not you know you need to gather all the information to make the right bet

When you say you have data do you mean that you actually have the data how you know the experts knowledge?

It's very different for quant research. Quant model is more like a classical machine learning model. Basically you are predicting a number in quant generally you predict the return of a stock okay for one day or something like that. And the input is usually or the most important input is market data is like price volume. This kind of data. Quant companies sit down what I heard is like they definitely use some news data as well but they first transform the natural language news into some features as still numerical. Yeah. Yeah. Exactly. Into a numerical numbers and then use those as a input to this model as itself. That's how you know traditional quant predict the future or the stock price. But here large model it's a generative model. It generates a say natural language with reasons. Okay.

So that's more like a you know fundamental analyst you know those researcher to do to analyze a stock right basically you gather all the information may not be even quantitative like you it's can be some kind of news some words from the CEO things like that and you make some judge you know human judgment you that's how the I think non-quant financial institute analyze stock right so basically large model is more like this because larger model now it can reason it can think so it just behaves like a researcher in this kind of non-quant institute, financial institute to analyze a stock however you know large model it can I would say just in terms of search capability is much stronger than human right now because it can like can do multi-hop search things like that and you know it can run 24 hour so it can analyze a lot of information than than the human.

So I would say this kind of deep research capability is generally stronger than a single human but of course you need to train to make it happen. So I think there's a that's how the main difference from the quant like it's more like the more traditional financial institute how they analyze stock. Yeah the data for quant and non-quant quant and non-quant model are or we say the quant model and the larger model are very different.

So for quant model usually it's very structured you structure you try to structure everything into a numerical format like a what you call a feature vector things like that but here you know large model just can deal with the natural language directly so that's very like what the normal human does to process this kind of information so that's the main difference and of course you need to train how you train it is like you need to train the model to have for large model, you need to train the model to have this judgment, reasoning, analysis capability and you need to have good either using for example good analysis data either from usually like from real humans you know expert how they analyze so they give this kind of data. So you can mimic their behavior or you can use say reinforcement learning so that you can when you do correct prediction you get some reward. So will somehow change the model to get better prediction capability. Yeah so this there are several training methodology to enhance this kind of capability.

And which ones are you using?

So we use all like we use all kind of training methodology. Yeah. Yeah. So yeah. So I think that's standard. And I'm not claiming this is something super novel. It's like standard people use supervised fine tuning and reinforce learning. Although the details are very different but at the high level those are standard steps.

Who should be using Apodex?

I think as long as you want to do some serious research on anything you should use like what have we have discussed for example if you are a scientist you want to research something about your own domains problem like biology physic whatever you can use it. You can do very thorough research to give you very it's basically can generate a paper you know about what's the question you want to ask. Okay. Or you know like another example we talk about like about finance or just like you want to predict some future event you should use Apodex.

Do you think a domain expert team should think about using a commercial solution like Apodex or should consider building a system themselves like multi-agent system themselves?

So first of all I think there are common patterns about reasoning and deep research they can generalize with different domain. Okay we specifically gather this kind of data. So our I would say search deep research verification capability is super strong compared to other model and we also support you upload your own data just like as a information source so we can you can also our model can also use this information source as well and so generally I say I think it's highly non-trivial to build agent system around like a raw model so I mean if they have super super strong scientist and engineering team and resources like GPU I mean they can build their strong model but you know that will take a very long time usually it will take a very long time and a lot of resource I think the easiest way is to use Apodex

How do you think that can affect our lives in the future if we outsource all these research to agents uh

If you look at the deep research process right there's a lot of like if you ask a human to do the kind of deep research there's a lot of labor task like do some search, you do some editing, gather information. So the basically our deep research agent can automate this all together and because we have verification the outcome is very accurate. So now it basically give you more freedom to think about what is the most important problem and to do like what kind of what problem you need to do research. So this is still up to at least for I think for the for now it's still up to human maybe in the future like AI can also tell you should do some research that's more like a sci-fi situation for now it's still based on human judgment to do what kind of research what kind of question you want to answer right I think this is still a more like subjective thing because you know because different people may have different opinions there's no single answer that you can only rely on yeah at least for Yeah. Yeah.

So I think it's give you people more freedom to explore more to make think more about what is the right question instead of doing all this kind of laboring work during this deep research process.

Is there a part of that world that would worry us or worry you?

I think different people may have different opinion. I'm generally optimistic. I think every new technology will create more more opportunity, more kinds of work people can do. It maybe beyond our imagination. So yeah. So I'm still optimistic about you know people can do more creative interesting work using AI.

CEO Eric was like booed by the students when he was talking about AI. There seems like a little bit sentiment about like people are confused about the future. How do you prepare us towards a positive outcome?

I think we are at a transitioning transition point right because now the work we have still the work we had we like before large model came out for example software engineering like front end back end many kind of office work things like that those are I mean those are you know you know AI can accelerate this work quite a bit like 10x or 100x so yeah I mean we should admit like you don't need that many people to do the same kind of work for this kind of work but I believe there are new work or new kind of jobs people will create it's maybe even beyond my imagination but generally speaking for new grad or undergrad or high school students I think the fundamentals are always important like the reasoning like human reasoning like either you know the common sense those are always important because there are always a step you know human can make some judgment and for people who want to transit or for example software engineers who want to transit to large

Models I would say so nowaday markets like this I don't think lo hard you require a lot of IQ intelligence I would say it's really about you should be very responsible and you need to be hardworking so those are more important characteristic of a good candidate we are we are looking for however it's it's very hard to identify those signals So if you look at all these frontier lab how they hire referral is probably the most common way to get a job in those frontier lab or any place that are building the large models because you know whether you are hardworking whether you are responsible whether you are keen to detail those are characteristic that are hard to be identified on your CV so you so the most reliable way is just asking you know some referrals however I think there's one way that is maybe is more common now is like if you have some open source project especially on agent well so why I say agent because it's very hard to get resource to train a large model but for writing some agent framework you only need API so you can still build a pretty actually very comprehensive strong agent framework just using API so I think that's probably the near-term kind of direction people can do to if you want to transit to some doing some larger model related work

But you are building something to replace this human reasoning process right how's it different

So I to me is like human is always like a verifier for large model because you ask language model problem questions and then we generate some answer interact with you in the past you may need to do reasoning our own but now you know you let the larger model generate answer with it only but you need to be able to reason in order to verify or look at the answers from large model whether it is correct or not. I think so see that's why I mean because you need to make the final call or you need to judge whether the response from large model is good or not you to have this judgment capability you still need to learn to reason unless you put everything you rely everything on large model then sure you don't need reason but as long as you are still interact with large model you still need the reasoning the judgment to even more if to use model more effectively.

You know this it's really interesting. I have a you know this is a thought process I'm thinking the how do someone like get trained if they don't get exposed to such opportunities to train that human judgment by doing it you know going through the process of the of quant finance analysts the whole process themselves because what I can do is this okay you have 150 agents who did all analysis for me and I'm going to be the final verify as a person what I'm going to do is I'm going to put it the results to another large language model and then have it to judge it for me. I will I will have a result. I mean, if that's the end outcome, right, I'm looking for. But it didn't pass through my brain at all. I've done it myself. I literally, you know, we probably all have done it, right? From one AI to another AI, right? I think we still grew up in the in an era that we still get trained from scratch. But I don't know about the next generation, right? What are you going to do?

Hard to predict. That's a hard problem like because I think they from when they started school they already have this all the AI tools and you know the all the reasoning process is done by AI then I don't know what their brain will be like after they graduate I generally believe learn by doing something even though with AI's help you still learn the whole process I think that's still very helpful and really can improve your understanding about larger model itself also the reasoning or general system building capability and generally speaking you know just like you work with some people if you know these people better then you can ask right question you can cooperate better so I think you know maybe the next generation students are trying to have a better you know how to better interact with those so this is a very important training subject

Are you concerned at all about OpenAI, Claude, their deep research progress.

Not really. I would say you know they are for-profit companies and I know I mean they are focusing on coding agent. So I say actually the deep research is not their focus right now and as far as I know actually we put even more resource you know people or compute maybe compute as well to for deep research and them so I'm I'm not too worried about they will catch up.

What's on your roadmap?

For now we are still a deep research agent. So we are building a heavy duty solver.

You're saying heavy duty solver?

Yeah. Yeah. Yeah. So heavy duty solver is our the vision of our company, right? So heavy duty solver is not only about deep research, although deep research might be the first important step. So after you do this research you may have a concrete plan to execute but you still need to execute for some problem execution means coding but here maybe coding is not about like software coding it's about scientific coding for example you need to use some biology tools just like computation biology or bioinformatics those kind of tools and you write some system to solve a computation biology problem.

So we still need to you know coding is very important of course like you need to have coding to really execute something also so that's one way another thing is after we have some deep research plan we'll also need to cooper collaborate with the math labs to do real experiments yeah so that's another thing we plan to do in the future so yeah so by heavy duty solver we really mean end to end solving this problem not only about planning which is basically what deep research does but we were trying to really solve the problem.

If you were to teach me one thing to help me become a better founder or a better AI user or researcher what would that be?

Yeah. So I think nowadays this system level thinking is very important like we call our agent system AgentOS operating system. It's really use the idea from operating system. This is like a classical you know it's a classic in computer science like Windows Mac there are operating system but how if you understand how this operating system works how was the structure how they organized that's really helpful nowadays to understand all this the whole large model system I think yeah I think it become more and more important nowadays to think about in a s system way

Could you recommend a book to our audience maybe maybe fresh graduate you younger version of yourself.

I think the operating system textbook might be the most useful one I would say. Yeah. Yeah. I think this I mean system I think operating system is not only about computer science you can think of more broadly. It's about organizing a system although operating system is like how basically instantiating this idea into a computer system but I say it goes beyond computer science. So I think learning the structure how it works is generally helpful.

Yeah, thank you so much, Simon.