The Boring Stuff Wins: What Building an AI-Native Company Actually Looks Like

Most organizations are not behind on AI. They are behind on the thinking required to use it.

I say some version of that at the start of every episode, and I keep saying it because the evidence keeps piling up. Companies have access to the same models. They can all buy the same tools. What separates the ones that pull ahead isn't the technology — it's the quality of thought they bring to it. The systems they design around it. The discipline they impose on it.

Which is why my conversation with Aidan McConnell was one I'd been looking forward to. Aidan is the founder and CEO of QuantLink AI, and before that he spent about five years at Sezzle, starting as a data science intern in 2019 and leaving as Head of AI/ML, having built and personally hired a globally distributed team of around fifteen people. He's now building what he describes as a vertically integrated data and AI platform for investing — premium market data with the quant workflows, analytics, and institutional-grade reporting layered on top, made accessible enough that, in his words, his mom could use it.

We spent the hour walking his company through the five-stage framework I write about in Future Proof Your Business: find your nest, diversify, become a platform, build a team with an ownership mindset, and invest in R&D while shaping your external environment. What struck me most wasn't how neatly QuantLink mapped onto that structure — plenty of good companies do. It was how many of the decisive moves turned out to be the unglamorous ones.

Finding the niche in a saturated market

The obvious objection to a company like QuantLink is that the space is crowded. Bloomberg, Merrill, Schwab — the incumbents are entrenched, and they have decades of institutional credibility that a young company simply can't buy.

Aidan's insight was to look at where that incumbency was actually failing people. A Bloomberg terminal runs north of twenty thousand dollars a year, and for a huge swath of investors and analysts, that's massive overkill for the analytics and reporting they actually need. On the other end sits the retail world of "trust the signal, just buy or sell." Aidan noticed that both extremes take control away from the user. So QuantLink was built to flip the game: instead of handing people signals, give them the tools to construct their own — pull the data they want, transform it their way, and decide for themselves.

That's what finding your nest looks like in practice. Not a brand-new market, but a gap inside a familiar one that the big players have structural reasons to ignore.

100% AI-written — and the ground truth that keeps it honest

Here's the line that made me stop and dig in: every bit of QuantLink's software is written by AI.

My immediate question was the one any operator would ask. How does that not collapse into slop? In a platform people trust with investment decisions, a hallucinated calculation isn't a bug — it's a breach of trust.

Aidan's answer is the most transferable idea in the whole episode. Early on, they hit exactly the problems you'd expect: models hallucinating, ignoring guardrails, producing inconsistent output. Their fix was to build an ontology — a graph that links everything together, from the data layer to the front-end components, with strict, deterministic pass/fail validation at every step. The generative model gives you flexibility; the ontology gives you a ground truth to check that flexibility against before anything reaches the user or the codebase.

He pointed to the SEC's XBRL system as a primitive version of the same idea: click a single number in a filing and it expands into pages of documentation defining exactly what that number is. QuantLink does that for its data, then does it again for its interface — components defined in structured JSON and hyperlinked, so the agent, the user, and the validation systems are all operating off the same source of truth.

The counterintuitive part is that this doesn't slow them down. It speeds them up. "If they're not constrained by something that's a ground truth," Aidan told me, "they will hallucinate and create slop." Constraint, done right, is what makes the acceleration safe.

This is the thinking I mean. The AI is the easy part now. The engineering discipline around it is the hard part, and it's the part that compounds.

The surprise: the boring stuff won

If there's one moment from this conversation I'd want every founder to sit with, it's this.

When Aidan started QuantLink, he was convinced the data science and AI would be the draw. That's what would pull customers in. He was wrong — and he's refreshingly honest about it. What people actually reached for was the charts. A Tableau-like layer that let them import their own data and build branded, pixel-perfect visuals — the kind you'd put in a 10-K presentation. Then the reports: the PDFs, the decks, the "boring" word-processing output that turns analysis into something you can send a client. Then the collaboration features — a shared file system like Google Drive, a Slack-style messaging hub, the approval and compliance workflows that investment teams currently grind through over email and WhatsApp.

None of that is exciting. None of it gets clicks. But it's where the real demand was. "Doing the boring stuff really well," he called it.

This is diversification in its truest form — not chasing adjacent shiny objects, but noticing which unglamorous capabilities customers keep double-clicking on and building those out properly. The differentiated, exciting feature gets you in the door. The reliable, boring one makes you indispensable.

From product to platform

The natural next question, once you've found your nest and started to diversify, is how you keep a bigger competitor from simply copying you. Aidan's answer is to stop being a closed product and become an open one.

The vision he's chasing — and he was candid that this is short-term, something the team has talked about "every single day for the last month" — is a marketplace. Bring your own API keys and data warehouse. Sell your own algorithms. Build your own chart templates and agent ecosystems. He reached for Figma and Snowflake as models: platforms whose real leverage comes from other people building and selling on top of them. QuantLink provides the storage layer and the tools; the community provides the rest.

That's the third stage of the framework, and it's the one that turns a good product into durable infrastructure.

People, ownership, and the long bet

On the team side, Aidan is building the ownership mindset in the most direct way possible: equity for everyone. He got in early at a startup that scaled into the billions, so he knows the risk-reward math of the people who take that bet with you. Add to that a genuine 20% time culture — interns who build a demo one week and see it become a core product feature the next — and you get a team that contributes to the roadmap rather than just executing it.

And on R&D, he made a bet worth chewing on. He thinks the future of AI, at least in software, isn't models generating code. It's models generating deterministic graphs — structured chains of steps that call code that already exists — validated against a real-world outcome, like whether an investment beat the risk-free rate. If he's right, the winning systems will be fine-tuned models a fraction of the size, cost, and latency of today's trillion-parameter behemoths. Whether or not that specific prediction lands, the instinct behind it — that the frontier is structure, not scale — is exactly the kind of forward thinking the fifth stage of my framework is about.

What comes next

What I appreciated most about this conversation is that Aidan didn't hide behind the AI. He treated it as one layer in a well-engineered system, and he was clear-eyed about where the actual value lived — often in the least glamorous corners of the product.

That's the lesson I keep coming back to. The organizations that win in an intelligence-driven economy won't be the ones with the best models. Everyone has the models. They'll be the ones that did the thinking: the ground truth, the validation, the boring features done exceptionally well, the systems that let AI accelerate them instead of embarrassing them.

Build the architecture, and the advantage follows.

🎧 Listen to the full episode of What Comes Next with Aidan McConnell: Building an AI-Native Platform to Take On the Bloomberg Terminal — with Aidan McConnell

📘 The five-stage framework in this piece comes from my book, Future Proof Your Business — available on Amazon.

Connect with Aidan McConnell at quantlink.ai or on LinkedIn.