Your AI is confident.
Should it act?

One function call. Returns ACT or ESCALATE. Every decision logged with a full audit trail. Zero dependencies. Under 1ms.

Try the demo ↓ GitHub → Docs →
Works with: LangGraph CrewAI AutoGen any Python
trust_demo.py 
from cognilateral_trust import evaluate_trust

result = evaluate_trust(0.85)
result.verdict        # → "ACT"
result.should_proceed # → True
result.record_id      # → "a1b2c3d4..."

result = evaluate_trust(0.42)
result.verdict        # → "ESCALATE"
result.should_proceed # → False
Live demo

Drag the slider. Watch the verdict.

Real trust engine. Real-time verdict.

Confidence 0.70
How sure is the AI? Drag to simulate.
0.00 guessing1.00 certain
Can you undo it?
Calls API, writes DB, affects users?
Drag the slider...
Pricing

Free forever. Scale when ready.

Open source library + hosted API.

Open Source
$0 forever
Full trust engine, runs locally
  • evaluate_trust() full pipeline
  • @trust_gate decorator
  • LangGraph TrustNode + CrewAI TrustTool
  • JSONL local audit trail
  • 24 modules, zero dependencies
  • Apache 2.0
pip install cognilateral-trust
Hosted API
$1 /10K evals
1,000 evals/day free, then metered
  • REST API — no infra to manage
  • Accountability record per evaluation
  • Nutrition label on every response
  • Tier-based rate limits
Get API Key
Why this exists

Built for the person at the other end of the decision.

Nobody is building for the person who gets affected when the AI is wrong and doesn't know it. We are.

Open source because trust should not be a premium feature.

Get started

Build trust into your stack.

Private beta. Every claim backed by automated tests. Trust should not be a premium feature — the library is free forever.