Skip to content
Methodology

How Kairo produces a research brief

Every claim in a Kairo report is traceable to a source, a timestamp, and a confidence score.

1. Evidence Engine

We fetch DexScreener, GeckoTerminal, and the project website in parallel. Each fact — price, liquidity, socials, deploy date — is wrapped in an Evidence Object with provider, capture time, and availability. Nothing downstream sees raw provider blobs.

2. Deterministic metrics

Numeric facts (24h change, volume/liquidity ratio, market age, socials count) are computed in code, not by an LLM. They are the ground truth the rest of the pipeline reasons over.

3. Specialist analysts

Five specialists — Market, Narrative, Developer, Wallet, Risk — read their own evidence bundle and emit findings tied to evidence IDs. If evidence is missing, the specialist declares 'insufficient-evidence' rather than inventing content.

4. Lead analyst + conviction

A lead analyst composes the summary from specialist findings only. Conviction is scored from signal weights and confidence bands, not raw prose.

5. Provenance you can audit

Every section shows source badges, freshness age, and confidence. The Evidence panel exposes the raw objects so you can verify the chain from claim to source.

Not financial advice

Kairo is a research tool, not investment advice. Reports may be partial when providers are unreachable or when a token is new. Always verify claims against on-chain data before acting.