Lenz

Overview

Lenz is a fact-checking API for AI products, content workflows, and document pipelines. It is designed for teams that need to verify factual claims before an answer, report, article, memo, or generated document is shipped to users.

The product is built around four primitives: extract verifiable claims from text, assess claims quickly, verify higher-risk claims through a deeper pipeline, and ask follow-up questions grounded in a completed verification. Lenz returns structured verdicts, citations, confidence signals, and audit details rather than a single model's unsupported answer.

For teams building AI agent or AI content workflows, Lenz fills a narrow but important gap: it can sit after generation and before publication, escalation, customer delivery, or compliance review. It is especially relevant when wrong claims create legal, medical, financial, brand, or customer-trust risk.

Key Features

• Claim extraction - Pull atomic, verifiable claims out of model output, drafts, reports, or other text so verification can run on specific statements.
• Fast assessment - Use /assess for a synchronous multi-model panel verdict in roughly the time range Lenz documents for interactive or bulk-screening workflows.
• Deep verification - Use /verify when a claim needs citations, stronger evidence gathering, adjudication, and a fuller audit trail.
• Grounded follow-up - Ask questions against an existing verification with /ask instead of starting a new ungrounded chat.
• Developer SDKs and CLI - Integrate through REST, Python, TypeScript, or the Lenz CLI, with documented API keys, typed errors, idempotency, and webhook handling.
• Workflow integrations - Connect Lenz through MCP, n8n, Zapier, WhatsApp, and automation workflows so factual checks can run where teams already work.

Integration Guide

Lenz works best when verification depth is matched to risk. A practical integration usually starts with /extract, sends the full text or selected claims through /assess, and escalates low-confidence or high-impact claims to /verify.

For production workflows, the official docs recommend webhook delivery instead of polling for deep verification. The API supports idempotency keys on write operations, so retrying a request after a network failure does not create duplicate verification tasks or double-spend quota when the body is unchanged.

Common integration patterns include:

• Screening generated marketing or SEO drafts before they publish
• Checking claims in AI research, due diligence, or legal-memo workflows
• Running regression tests on a golden set of factual claims before an AI product deploys
• Routing customer-reported bad answers into evidence-backed incident triage
• Giving Claude, Cursor, or other MCP clients access to a fact-checking tool through lenz.io/mcp

This makes Lenz complementary to AI writing assistants and research tools rather than a replacement for them. It does not generate the whole document; it checks whether the factual claims deserve to ship.

Pricing & Plans

Lenz has a free tier and self-serve paid plans. The official developer page lists a free plan with daily /extract quota and limited monthly /assess, /verify, and /ask quota. Paid self-serve pricing starts with Developer at $99 per month, followed by Scale at $399 per month. Enterprise is custom.

The important cost question is not only monthly price. Teams should estimate how many claims each document produces, how often fast assessment is enough, how many claims need deep verification, and whether webhooks or batch verification are required for throughput.

How It Compares

Generic search APIs, web scraping tools, and retrieval systems can gather source material, but they do not necessarily produce a structured verdict with adjudication, confidence, citations, and a reusable audit trail. Exa and Perplexity AI can help teams discover or answer with sources; Lenz is more narrowly focused on verifying claims as a pipeline step.

Plagiarism and AI-detection tools such as Copyleaks answer a different question. They can help with originality or authorship risk, while Lenz evaluates whether factual claims are true, false, mixed, or uncertain according to gathered evidence.

That focus makes Lenz useful when a team already has generation, search, and review tools but lacks a formal factual-verification gate before publication.

Best For

• AI product teams that need a factual gate before answers, reports, or generated documents reach customers
• Content teams publishing high-volume AI-assisted articles, briefs, or research summaries
• Developers adding verification to AI productivity or document workflows through an API, SDK, CLI, or webhook
• Teams that need MCP-accessible fact checking inside coding or research assistants
• Compliance-sensitive organizations that want a stored citation trail for why a claim passed or failed review

FAQ