Today we're announcing that we've raised a $20 million Series A to accelerate our mission of building the intelligence layer for the markets. With it, we're doubling down on a platform that reads everything the market publishes, judges what actually matters, and hands a desk the three headlines worth their morning instead of the four hundred that aren't.
The round was led by the partner who wrote our first institutional cheque, with participation from everyone who has backed us since the seed. We're deliberately not turning this post into a list of logos. The people involved know who they are, and what they bought is the subject of the rest of this page.
Rise of the judgement desk
Something has shifted in who our users are. The classic buyer of market data wanted more of it, faster, and measured vendors in milliseconds. The people signing up for Forecite are measuring something else: how much of their morning survives contact with the tape.
They run concentrated books at single-manager funds. They cover industrials at long/short shops where a third of the reading arrives as a forward from a colleague, not a row from a feed. They build systematic strategies where a backtest is only as honest as the data's memory of what was known when. What they have in common is that information stopped being their scarce resource years ago. Judgement is, and until recently nobody sold it.
That is the desk we build for. Not a faster firehose for people who already drown, but a layer that reads everything so a person can afford to read almost nothing.
Built on judgement primitives
Underneath the product is a small set of primitives we have been assembling since the first commit, and this round exists to deepen every one of them:
- Coverage at the source. Edge connectors colocated with the publishers they read, across the wires, the filings systems, the exchanges and the financial press. Fifty-four sources today, growing most months.
- The Verdict Engine. Our in-house machine learning, built and evaluated against millions of headlines and the real price behaviour that followed each one. Not a sentiment widget. A judgement about whether a document can move a price.
- Decomposed scores. Every verdict breaks into named sub-scores on published scales, because a number you can interrogate is worth ten you have to trust.
- Live delivery. A feed where a row arrives already judged, with server-side filtering so software and people alike narrow before they read.
- History that holds still. Replayable, point-in-time honest, because research that can't trust its data's memory isn't research.
- Surfaces for every reader. A dashboard for humans, an API and SDKs for systems, Warren and an MCP server for the agents that increasingly sit between the two.
"I stopped keeping the tab open all day. I keep it open for two hours and then I close it, and I know I haven't missed the thing."
Portfolio manager, single-manager equity fund, ~$180m
That quote is the product working. The tab closing is the feature.
What comes next
The capital goes to three places, in order of how long they take.
Coverage first. Wider source reach across more markets and, by this time next year, judgement in a language other than English. This is the slowest and least demonstrable thing money can buy, and it converts into nothing a customer can see for months. It is also the moat.
Calibration second. The engine holds one global bar for what counts as actionable today. A desk covering a name three analysts follow deserves the same quality of judgement as one covering a mega-cap, and getting there is careful, unglamorous research work we can now fund properly.
People third. More of the team on the surfaces where someone actually meets a score, because a good judgement shown badly is a wasted one. We are hiring across engineering, research and product, and the careers page is finally worthy of the word.
Building the intelligence layer
We think the next decade of market data looks less like terminals and more like judgement: infrastructure that doesn't just move information but takes a position on it, in public, with the possibility of being wrong. Being checkably wrong is the entire discipline of this company. Every verdict we publish is a claim someone can hold against what the market did next.
A Series A doesn't change that. It just means we get to do it wider, deeper, and in more languages, for the desks that decided judgement was worth paying for before it was obvious. To them, and to the partners who saw the same thing: thank you. Now back to work.