We are building the layer that reads the market for you
Vates exists because the gap between what is knowable about a market and what any one desk can actually get through keeps widening. We think that gap is a machine problem — and that closing it should make analysts better, not redundant.
Three positions the product is built on
01 — The problem
The market produces more than anyone can read
Every session generates more filings, prints, transcripts, and commentary than a desk can get through — and the volume grows every year while the number of hours does not. The limit on most research processes is no longer access to information. It is attention.
- Coverage silently narrows to whatever is already in the news
- The quiet name that is becoming interesting gets noticed late
- By the time a pattern is obvious enough to spot manually, it is priced
02 — The position
Machines should read. People should decide.
We think the useful division of labour is unambiguous: continuous reading at scale is a machine problem, and judgement under uncertainty is a human one. Products that blur that line tend to produce confident nonsense that someone still has to check.
- The model reads everything, consistently, without getting tired or bored
- It reports what changed, against what baseline, with what confidence
- The thesis, the sizing, and the call remain with the person accountable for them
03 — The standard
A signal you cannot interrogate is a signal you will ignore
Every analyst has muted an alerting system that cried wolf. We hold ourselves to a harder standard than accuracy alone: every signal has to be traceable back to the evidence that produced it, and the system has to say when it is unsure.
- Show the inputs, not just the conclusion
- State confidence explicitly, including when it is low
- Keep the full history — including the misses — auditable after the fact
Built for data you cannot afford to leak
Market positions and research are among the most sensitive data a firm holds. Vates is built so that access is explicit, activity is reviewable, and your research never becomes someone else's training data.
- Role-based access control, with an audit-friendly history across data, research, and alerts
- Encryption in transit and at rest, and isolation between tenants
- Your proprietary research and positions are never used to train shared models
- Ongoing monitoring and a formal risk management process
The fastest way to judge this is to see it
Bring your instruments and your actual workflow. We will show you what Vates surfaces against them, including where it would not have helped.