Meet Analyst One
What have we been doing behind the scenes, and where are going?
How We Got Here
I once spent an evening I’d promised to get dinner with a friend who was moving out of London, hunting for a single CO2 emissions table, buried somewhere around page 223 of an annual report[1]. I found it eventually. I’ve never quite forgiven the report for it. That evening — and a few hundred like it — is roughly why Felix Research exists, and why, after a heads-down first half of the year, I’m finally writing to you as founder.
It’s been a productive six months. We built out our core infrastructure, spent our waking hours trawling through research reports, model cards and infrastructure configs, and in March we put our first demo — Amuse Bouche — in front of a small group of testers and partners. Their feedback was generous, sharp, and more encouraging than we had any right to expect. It also told us exactly where to push. Since then we’ve been heads-down on one thing: turning that demo into a platform worth launching. It’s nearly there, and this is me finally telling you what it looks like.
What We’re Doing
We started Felix with a straightforward goal: faster, more accurate financial research. We know that problem because we’ve lived it — the cancelled dinners, the annual reports read at midnight, the crucial line missed because it sat on page 223 and not page 12. Over the years, the volume of information that we process for investment decisions has only grown, as the landscape of available data and information continues to proliferate. Yet the tools the industry actually uses haven’t made that much easier. If anything, they’ve made it easier to miss things.
We aren’t the first to notice that AI might help here — and I hope this isn’t the first you’re hearing of the idea either. But from day one our focus has been different. Most of what’s on offer is a chatbot, or a thin wrapper sitting on top of databases and documents that were never designed to work with it. We didn’t want to add another one. We wanted to ask what applied AI should look like when it’s built for real professionals, from the ground up.
I was never the best analyst on my teams. What I lacked in patience I made up for in building workflows, and that instinct underpins everything we do: make information discovery faster and lower-friction, in real use cases. So we built the best OCR and extraction infrastructure we could — and then went further. Information is only worth anything if you can trust it. We can’t promise AI that never errs, but we’ve built a harness around it that cites every claim back to its source. At each step, a citation marker keeps the original document one click away. No black box — just an answer you can check.
Name Games
Felix Research was never meant to be just another B2B AI company. The goal isn’t to ship one good product and call it done; it’s to keep finding better ways to help people comprehend more, faster, and trust what they find. For now that work lives in financial markets, the domain we know best. But the technology underneath it reaches much further — into tailored and in-house workflows, and research domains well beyond finance. Everyone deserves better research, and in the age of AI, being able to trust your information isn’t a nice-to-have. It’s the whole game.
The product has worn a few names on the way here. Amuse Bouche was the first — a taste of what was coming. That became FelixOne. But we wanted a name that anchored the tool in its job and its industry, rather than in us.[2] So FelixOne became Analyst One: the last analyst you’ll ever have to hire. The “One” is deliberate. It’s our first product, not our last.
Meet the Analyst
So what is it, and what can it do?
Analyst One is a buy-side analysis workspace built to make document comprehension and analysis a single, fast workflow. The idea underneath is simple: take static data and turn it into information you can reuse and trust.
When you upload a file or paste a link, a few things happen. The document is securely extracted on our servers — we don’t retain your data — and turned, in full, into a structured dataset. That dataset is then run through our proprietary analysis engine, which surfaces the key figures, relationships and details inside it. When it’s done, you get a ready-to-work document alongside a tailored report that ties what we found back into the analysis you’re actually doing. Each document is read on its own terms and as part of the larger question you’re asking, then presented the way we’ve found most useful to look at it.
From there, it’s yours. Our chat assistant stays alongside you the whole time — and since Dimitri and I couldn’t agree on which side of the screen it belonged, it’s on both. The agent can help you dig deeper, consider alternative viewpoints, and draft output, right in the platform. In the great irony of life, of course, all of the output can be exported not only as Markdown or Excel for tables, but as nicely formatted PDFs that are ready to be shared.
What’s Next?
Analyst One launches by the end of the month, just in time for the European summer break. Our timing is, as ever, impeccable. It’s nearly ready; we’re just making sure it’s got the goods.
You can join the waitlist now at felixresearch.com/waitlist to be first through the door — and to catch a discount before the free tier fills up. If you find it useful, share it; every time someone passes the link on, I smile a little. And if you’re not convinced, by all means send it to your competitors, so they can suffer such a terrible product.
Joking aside: this is a first release. It isn’t finished. Between you reading this and the doors opening, we’ll have added features, sped things up, and almost certainly broken and fixed a few things along the way. Put it through its paces and tell us what you find — the good, the bad and the ugly. There’s a conveniently placed feedback button waiting for exactly that.
Pricing comes soon. It’ll be more expensive than you’d like and cheaper than we’d like — and still better value than anything else of its kind. That much I’m confident saying.
And if you’re building or investing in applied AI for industry, I’d like to talk.
— Ben