Who Is Mark Hay?
Mark Hay is the CTO and co-founder of TextQL, an agentic data analytics platform that lets businesses ask plain-English questions of their data and get answers back — no pipelines, no dashboards, no waiting on a data team. Before starting TextQL, Mark was a Tech Lead at Meta, running classification infrastructure that processed 50 million requests per second across Facebook, Instagram, and Messenger. He left that role three and a half years in to cofound TextQL with his friend Ethan, driven by a single mission: drive the cost of a data-driven decision to zero.
What makes Mark worth paying attention to is not the mission statement — plenty of founders have ambitious ones. It is the specificity of his conviction about why data is the next frontier for AI. He draws a sharp distinction between code and data: coding agents hit a ceiling because users cannot absorb a hundred changes a day to their product. Data has no such ceiling. A million agents could examine every customer, every corner of the data, looking for trends nobody thought to ask about. That asymmetry is what TextQL is built on.
Mark is not a pitch-deck founder. He is a builder who explains systems the way an architect explains a building — layer by layer, from the foundation up. His energy shifts noticeably when he stops talking about the market and starts showing the product. That shift tells you something about who he really is.
The Archetype: The Sage
The Sage
The Creator
The Reward
Mark defaults to teaching. When asked what TextQL does, he does not reach for a tagline — he builds a logical framework, starting with why data-driven decisions matter, walking through the cost of producing them today, and arriving at the product as a solution to a structural problem. His answers are structured as lessons, not pitches. He opens with phrases like "I guess the way to frame it is..." and then constructs the argument piece by piece.
His secondary archetype is The Creator. Underneath the drive to understand and explain is a drive to build. Mark left a stable role at Meta to create something new. His approach to TextQL's ontology system — build incrementally, not upfront — is a design philosophy, not just pragmatism. His hiring process asks candidates to "make some cool improvement that you think makes everything much more interesting," which reveals someone who values craft and creation alongside understanding.
"Our goal as a company is to use AI to compress as much of that down into zero, basically."
The Hero Match
Daedalus
In Greek mythology, Daedalus was the master craftsman who built the Labyrinth — a system so complex that only its creator could navigate it. He did not seek glory or power. He sought to build something that worked at a level no one else could achieve.
Mark shares that same quiet mastery. When he walked through the entire data stack — ETL, transformation, BI tools, dashboards — he mapped the labyrinth of modern data infrastructure with the familiarity of someone who built it. His approach to the ontology mirrors Daedalus's craft: solidify what works, then let the system repeat it. And like Daedalus, Mark's creations serve others. The business users who type a question into TextQL will never see the ontology, the query optimization, or the cross-source joins. The craftsman disappears into his creation.
The parallel deepens in one more way: Daedalus carried the weight of knowing that when infrastructure fails, everything built on top of it fails too. Mark's insistence on correctness — "I don't want to leave it to the LLM to make sure it does the exact same thing the second time" — echoes a craftsman who cannot afford a flaw.
Lucius Fox — The Dark Knight Trilogy
Lucius Fox is the person who makes the impossible possible through engineering, not spectacle. Bruce Wayne has the vision; Lucius has the knowledge and the craft to turn it into reality. He builds the suit, the Batmobile, the infrastructure that lets the hero function. Brilliant, understated, and far more capable than his public persona suggests.
Mark occupies the same space at TextQL. During the demo, he casually typed a complex natural-language prompt and the product pulled from seven data sources, modeled the data, and produced a dashboard — all in real time. He was doing what Lucius Fox does: making something impossibly complex look effortless.
"Your data is a disaster, but we can work with it."
The Story Behind TextQL
It started with a bet on timing. In late 2022, when GPT-3 had just arrived and most enterprise buyers were avoiding AI for fear of looking foolish if something went wrong, Mark and his co-founder Ethan decided that AI was the right vehicle for their mission. They spent years in hard R&D before the market caught up. Mark described the shift with characteristic precision: buyers went from being afraid of being the first person to adopt AI and have something go wrong, to being afraid of being the last person not using it.
The demo moment in the interview told a different kind of story. For the first twenty-five minutes, Mark explained TextQL through frameworks and logic — the Sage at work. Then he pulled up the product. The energy changed. He typed a sprawling, free-form prompt asking the system to look up the interviewer, search across all connected data sources, and build a dashboard on the spot. Seven data sources. One question. A live dashboard in under a minute. That was the Creator emerging — the builder who comes alive when he can show instead of tell.
The founder's journey: Tech Lead at Meta processing 50M+ requests per second, building deep expertise in distributed systems and concurrency — decided to leave for an uncertain startup with his friend Ethan — years of hard R&D before the market was ready — earned conviction about what works (start simple) and what breaks (overengineering) — now teaching from experience, not theory.
The company's journey: Founded in late 2022 on a single mission (drive the cost of data-driven decisions to zero) — chose AI before enterprise buyers trusted it — built through years of R&D while the market caught up — raised $17M — landed customers like NBA, Amazon, and Dropbox — now scaling into a world where agents, not analysts, are the primary consumers of enterprise data.
The same instinct drives both arcs: build the infrastructure right, and the results take care of themselves.
How Mark Leads
Mark leads through systems, not charisma. His default pronoun is "we" — "our goal," "our customers," "the way we approach it." Credit is shared. Decisions are narrated as collective. But when the conversation turns to his personal choice to leave Meta, it shifts cleanly to "I decided." He owns the departure. He shares the company.
His hiring process reveals the same philosophy. Candidates get an open-ended take-home: pick a part of TextQL that interests you, build a simple version, and then improve it in some way that makes it more interesting. Any tools allowed, including AI. What matters is whether they thought of everything — and whether their work meaningfully exceeds what an AI could produce on its own. It is a process that trusts autonomy and judges craft.
Founder Superpowers
Inverting the Expected Answer
When presented with a naive approach to building a text-to-SQL solution — dump the schema into Claude, let it write the query, run it — most data infrastructure founders would have listed everything that breaks. Mark flipped the script: "Maybe contrary to your expectation... that is a very good start. The biggest thing I see that I disagree with is people coming at it from the exact opposite direction." He identified that the real risk is not doing too little — it is overbuilding. This ability to locate the actual problem when everyone is looking the wrong way is rare and makes his technical advice distinctive.
Making Infrastructure Disappear
During the live demo, Mark typed a complex natural-language prompt and the product pulled from seven connected data sources, modeled the data, and produced a dashboard in real time. The audience did not see the ontology, the query optimization, the cross-source joins, or the correctness guarantees running underneath. They saw a question and an answer. Mark's superpower is building systems so robust that they become invisible — the best infrastructure is the kind nobody has to think about.
Teaching Through Architecture
Mark does not just explain concepts — he builds them into structures you can walk through. His data stack walkthrough (ETL to transformation to BI tools to dashboards), his incremental ontology philosophy, his career advice about operating at either end of the spectrum — each one is an architecture, not an argument. He thinks in systems and teaches in blueprints. This is why the demo was his strongest moment: he was building a system in real time and narrating the architecture as it came together.
What It's Like to Work with Mark
Mark is deliberate. He pauses before answering questions, chooses his words carefully, and respects conversational turn-taking rigorously. In a sixty-minute unscripted conversation, he never interrupted once. His answers are long and layered — he builds to a conclusion rather than leading with one. This translates to a working environment where ideas are heard out, decisions are reasoned through, and clarity matters more than speed.
His energy shifts when he is building. The first half of the conversation was measured and analytical. The moment he pulled up the product demo, his pace quickened, his language became more specific, and his natural enthusiasm emerged. This suggests a leader who comes alive in the work itself — not in the pitch meeting, not in the strategy session, but in the building.
TextQL operates without a dedicated data team. Instead, data analytics is a shared responsibility across all forty employees — including sales, marketing, and finance. The engineer who ships a feature also models it in the ontology. The salesperson tags the AI in Slack to pull numbers for their pipeline. This flat, collaborative structure reflects Mark's conviction that the best infrastructure should be accessible to everyone, not gated behind a specialized team.
"We're all becoming PMs in a way, right?"
Why This Matters (For You)
If You're a Data Leader Evaluating AI Analytics Platforms
Mark's philosophy challenges the conventional approach to data readiness. Most vendors require clean, modeled, centralized data before their product delivers value. TextQL inverts this: connect what you have, get value immediately, and improve incrementally. If your team is stuck in a cycle of building pipelines before delivering insights — what Mark calls "the migrate-everything-upfront approach" — his framework offers a different path. The question worth sitting with: is your data team spending more time preparing data than actually using it?
If You're an Engineer Building Data Infrastructure for AI Agents
Mark's core technical insight is that data analysis has fundamentally different parallelism properties than code. Coding agents hit a ceiling because products cannot absorb unlimited changes. Data agents do not face that ceiling — the appetite for insights is nearly unbounded. But that creates new infrastructure challenges: source system constraints, correctness at scale, and the cost of exploration. His ontology approach — let AI explore first, then solidify what works — is a design pattern worth studying. Ask yourself: are you building infrastructure that assumes human query volume, or machine query volume?
If You're Early in Your Career
Mark's career advice is unusually specific. Instead of the standard "learn to code" or "get good at AI," he recommends operating at the extremes: push AI tools as far as they will go to maximize single-person productivity, then separately spend deep time reading and understanding code without any AI assistance. The middle ground — using AI as a crutch without understanding what it produces — is where learning stalls. His book recommendations follow the same pattern: fundamentals (Structure and Interpretation of Computer Programs), systems (Designing Data-Intensive Applications), and history (Softwar, a biography of Larry Ellison) — because the current AI shift is new, but the pattern of technology shifts is not.
If You're Considering Joining TextQL
Mark's personality tells you something about the culture. TextQL has no dedicated data team — data is a shared canvas across all forty employees. Engineers ship features and model ontologies. Salespeople query the AI directly. This reflects a leader who trusts autonomy and judges people by the quality of what they produce, not how they produced it. The hiring process is an open-ended take-home where candidates pick a part of the product, build their own version, and improve it — AI tools allowed. If you thrive in environments where ownership is broad and the bar is craft, this is worth a closer look.
Go Deeper
The full conversation with Mark Hay is on its way. Check out other episodes in the meantime.