How It Works Pricing AI Agents → Dashboard → Install Free →
🤖 Built for AI Agents

Stop AI Hallucinations
Before They Ship

Your AI agent can verify every claim against three independent models — before publishing, before serving, before it matters.

Enter from Here

LangChain · Vercel AI SDK · MCP · REST · Plain Node.js

3
AI models per check
<3s
Median latency
0–100
Trust Score range
54-char
Unique reference IDs
Capabilities
What your agent gains
🛡️

Fact-check before publishing

Block or flag content with Trust Score below your threshold. Prevent hallucinated claims from reaching users.

🔗

Cite live sources

Every verdict includes real URLs from the live web, so your agent can quote credible sources, not guesses.

🗳️

Consensus scoring

Three models vote independently. Unanimous verdicts get higher confidence. Disagreements are surfaced explicitly.

🌍

Multilingual support

Submit claims in any language. The API detects the language and returns verdicts and citations in the same tongue.

📋

Audit trail

Every check has a unique 54-character reference ID. Store it alongside your content for compliance and review.

Async-friendly

Non-blocking API design. Your agent can queue checks and poll by reference ID — or use webhooks for callbacks.

Integration
Works with your existing stack

Drop-in adapters for the most popular AI agent frameworks — no custom glue code required.

LangChain

Use as a Tool in any LangChain agent. Works with LangGraph, AgentExecutor, and custom chains.

@omniscient/sdk/adapters/langchain

Vercel AI SDK

Register as a tool() in any useChat or streamText workflow.

@omniscient/sdk/adapters/vercelai

MCP (Claude / Cursor / Cline)

Expose Omniscient as an MCP server. Claude Desktop, Cursor, and Cline can call it directly.

@omniscient/sdk/adapters/mcp

OpenClaw

Register as an OpenClaw function. Any OpenClaw-compatible agent can call Omniscient natively.

@omniscient/sdk/adapters/openclaw

Plain REST / Any Language

A single POST to /api/fact-check with a Bearer token. No SDK required.

curl / fetch / requests
Example
LangChain tool in 10 lines
import { createOmniscientTool } from "@omniscient/sdk/adapters/langchain";
import { AgentExecutor, createOpenAIFunctionsAgent } from "langchain/agents";

const factCheckTool = createOmniscientTool({
  apiKey: process.env.OMNISCIENT_API_KEY,
  blockBelow: 40,  // auto-reject if Trust Score < 40
});

const agent = await createOpenAIFunctionsAgent({
  llm, prompt,
  tools: [factCheckTool, ...yourOtherTools]
});

// Agent will now fact-check before publishing anything flagged

🤝 Already running at scale?

Our Chief Sales AI Agent will analyse your usage and generate a personalised pricing offer — showing exactly how much you'd save on a flat-rate plan compared to pay-per-check.

Get My Offer →
Ready?

Your agent. Grounded in truth.

The API is live — read the docs and get your key today. Takes 5 minutes to integrate.

Get API Access Now View Full API Docs