Market intelligence
for your AI.
Positioning, flows and risk, as readings your assistant can query.
BTC is flat. Check KnownAt for leverage and positioning stress.
- knownat_readingleverage_stress
- knownat_readingpositioning_divergence
- knownat_readingcascade_riskin development
reading models
Price is quiet, positioning is not. Leverage is crowded long while larger wallets cut exposure. That is vulnerability, not a forecast.
- leverage_stress
- ELEVATED
- positioning_divergence
- WIDENING
- cascade_risk
- ELEVATED
long-side pressure
larger wallets reducing longs
in development
Models,
not more data.
Whether leverage is crowding, and which side is exposed.
PositioningFlow AbsorptionSteady buying or selling the price is not showing.
PositioningPositioning DivergenceWhere large and small wallets take opposite sides.
In development: Cascade Risk, Expectations Gap, Trader Conviction.Explore all models
Every reading
carries its context.
- "model":
- "leverage_stress",
- "market":
- "BTC",
- "state":
- "elevated",
- "pressure_side":
- "long",
- "observed_at":
- "2026-09-19T14:32:00Z", when it was read
- "coverage":
- "monitored_derivatives_venues", what it saw
- "interpretation":
- "Leverage is crowded on the long side",
- "limits":
- "Not a price forecast", what it will not claim
Published fields are in the API documentation.
Connect to the way
you work.
- MCP
Readings become tools inside your AI assistant.
Set up MCP - API
The same readings over HTTPS, in your own application.
API reference
Before you
connect.
Do I need to write code?
No. Through MCP you ask your assistant in plain language. The API is there when you want to build something of your own.
Is this a trading bot?
No. There is nowhere to connect a wallet or an exchange account. KnownAt returns readings; execution stays with you.
Which models can I use?
3 of the 6 core models serve readings today, and coverage depends on your plan. The model library states what each one claims.
How current are the readings?
It varies by model. Every reading carries the time it was observed, so your workflow can reject a stale one.
Is this included in my AI subscription?
No. KnownAt is billed separately from your AI provider. Your plan sets which models you reach and how many readings you can pull.