Add fact-checking to any workflow.
Lenz connects to the tools your team already uses. Use it to screen AI-written content before it goes out, flag claims in an automated workflow, or verify a fact while working in Claude or Cursor — without switching tabs. Pick the connection that fits your process.
Use Lenz from the chat you already have open. Ask it to check a claim and the verdict comes back in the thread, with the sources behind it. Neither needs a key. Claude connects in one click, and one sign-in covers the desktop app, claude.ai and mobile. ChatGPT takes a few more steps, through OpenAI’s developer mode, and is in beta.
Lenz is a remote MCP server at lenz.io/mcp. Point a coding
tool at it with a Bearer header and your API key — one command in
Claude Code and Codex, a config block anywhere else.
The Lenz node is available on n8n.io. Add a fact-check step to any workflow — no HTTP configuration, just your API key. The node returns a structured verdict, score, and citations.
A native Zapier app — same structured output as n8n, no code required. One trigger and four actions: verify, assess, extract claims, and grounded follow-up.
The same four API primitives — extract, assess, verify, ask —
available as shell commands. pipx install "lenz-io[cli]",
paste your key once, done. Add --json to pipe into jq.
Send a claim, a headline, or a link to the Lenz number and the verdict comes back in the thread — sources included. Link it to your account from your profile and claims and quota are shared across web and WhatsApp.
All of these on one screen, with your key already in place — Open the setup screen →
Building a direct integration?
Lenz is a REST API with typed SDKs. Four primitives — extract, assess, verify, ask — cover everything from bulk triage to a full audit trail. Prefer raw HTTP? The endpoints are in the developer docs →
Typed client. client.assess(claims=[...]) for a fast synchronous pass over a list of up to 20,
client.verify_and_wait(claim) when you want the citations
and the full trace.
Same four primitives, fully typed. client.assess({ claims }) returns one
row per claim, in order, ready to branch on before anything ships.
Resources
Add fact-checking to any workflow.
Pick your tool and we'll write the setup steps. Free tier, no card required.