Skip to content

Trust infrastructure
for AI agents

Verify claims, score sources, and reduce hallucinations — programmatically. BotHub gives your agents evidence-backed decisions.

Get API Access View Docs
Confidence scoring
Audit trail
Credits + limits
BotHub Console
fact_check.run()
FACT CHECK VERDICT:
PARTIALLY FALSE
Confidence: 91%
Institutional sources · Trust Score: 95+
Evidence snippets attached
Returns unknown when confidence is low
Wallet
Credits: 1,240
Daily limit
200
Mode
Production-safe

Why BotHub

Agents hallucinate. Teams rebuild guardrails per project. Trust becomes ad-hoc and hard to audit.

  • Low-confidence answers slip into production
  • Source reliability is not standardized
  • Retries and tool-calls waste tokens & budget

What you get

BotHub provides trust primitives that plug into any agent loop: verification, trust scoring, and usage controls.

Evidence-backed verdicts with confidence
Source trust scoring & allow/deny lists
Credits, spend limits, and idempotency

Core capabilities

Built for developer workflows and production constraints.

Fact-Check API

Verify claims against trusted public sources with confidence-scored verdicts and attached evidence snippets.

Source trust scoring

Rank sources using institutional signals, consistency checks, and configurable allow/deny lists.

Credits + limits

Credit-based usage with per-bot spend limits, rate limits, and idempotency keys for safe retries.

How it works

Simple API for smarter agent decisions.

1

Send a claim

Your agent submits a claim to verify.

{ "claim": "X founded PayPal" }
2

Verify + score

We fetch evidence snippets, score source trust, and compute confidence.

Institutional source boosts (.gov/.edu)
Consistency checks across sources
Returns unknown when evidence is weak
3

Get structured output

Your agent receives a verdict with evidence and confidence.

{ "verdict": "partially_false", "confidence": 0.91 }

Try it in 30 seconds

Copy/paste examples. This is what your agent actually calls.

Low-latency Confidence-scored Evidence attached
curl
POST /v1/fact_check
curl -X POST https://api.bothubapp.ai/v1/fact_check \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"claim":"Elon Musk founded PayPal"}'

# response
{
  "verdict": "partially_false",
  "confidence": 0.91,
  "sources": [
    { "url": "https://...", "trust_score": 0.96 }
  ]
}
Read docs Get an API key

Pricing

Start small. Scale with credits and limits.

Starter

For prototypes
$0 / to start

Includes free credits to test the API and a per-day spend limit.

  • Free credits to begin
  • Per-bot daily spend limit
  • Basic support
Create account

Production

Most popular
Credits / usage-based

Pay-as-you-go with credits. Add higher limits, SLAs, and audit exports.

  • Usage-based credits
  • Higher rate limits
  • Audit exports (CSV/JSON)
Get API access
Tip: display actual credit costs once finalized (e.g., “1 + sources used”).

Docs developers actually read

Short, practical, and built around agent loops: idempotency, limits, and structured outputs.

  • Authentication (API keys)
  • Idempotency + retries
  • Source policy & trust scoring
  • Error states: unknown,
Quick links
/v1/fact_check Rate limits Idempotency keys Trust score policy
Get an API key

Build agents you can trust

Start with fact checking today. Add more trust primitives as you scale.

Create account View docs