Evidence
Evidence links claims to proof. Every assertion in an Echo traces back to sources, with clear provenance and confidence scoring.
Why evidence matters
When an AI agent says a venue has an "intimate atmosphere," that claim needs backing. Evidence provides the chain: where the data came from, how it was processed, and how confident the system is in the claim.
What evidence contains
Each piece of evidence connects a claim to its supporting data:
| Component | What it captures |
|---|---|
| Claim | The assertion being supported |
| Source | Where the data comes from |
| Provenance | The chain from original data to processed claim |
| Confidence | How strongly the evidence supports the claim |
Confidence scoring
Confidence reflects how well-supported a claim is. Higher confidence comes from more data, stronger agreement between sources, fresher information, and more reliable source types.
A claim backed by many recent, agreeing sources scores higher than one with sparse or conflicting information.
What agents can do with evidence
When a guest asks "why this hotel?", an agent with access to evidence can give specifics rather than assertions. It can explain what sources support a claim and how confident the system is.
This is transparent reasoning rather than a black box.