Fact-checking for AI-drafted text.
Lenz is a fact-checking API. It identifies factual statements in a draft, a memo, a model’s answer or a page of copy that it can check against independent public sources, and returns a result for each.
Example: a paragraph written by a model
“GPT-4 was released in March 2023 with a context window of 128,000 tokens. OpenAI was founded in 2015 as a non-profit research laboratory.”
Full verification
- OpenAI released GPT-4 in March 2023. True high confidence 10/10 go.nature.comreuters.comapnews.com+9 more
- GPT-4 had a context window of 128,000 tokens at its release in March 2023. False high confidence 1/10 openai.comweb.archive.orgdevelopers.openai.com+10 more The March 2023 release did not provide a 128,000-token context window. OpenAI documented an 8,192-token window for GPT-4 and limited access to a 32,768-token variant; the 128K window was announced for GPT-4 Turbo in November 2023. The claim conflates the original model with that later release.
- OpenAI was founded in 2015 as a nonprofit research laboratory. True high confidence 10/10 storage.courtlistener.com990s.foundationcenter.orgtechcrunch.com+16 more
Checked 7 September 2026. Open a row for the sources it used and the reasoning that weighed them.
Use Lenz in your workflow.
You hand over the text — a draft, a memo, an answer, a page of copy. Lenz reads it and picks out the factual statements it can check against public sources, so you are not listing them yourself. Each of those statements then gets a check.
There are two kinds, and you choose per statement. An assessment is a judgement: a panel of models reads the statement and returns a verdict and a confidence level in about 10 seconds, which is how you sort a long draft and find the statements worth looking at. A full verification is an investigation: Lenz searches independent public sources, has models argue both sides on what it found and a panel review the arguments, and returns the verdict with those sources and the reasoning, in about 90 seconds per statement.
Inside Claude, with one sign-in and no key; in Cursor and Claude Code, with one command and your key. From a script in Python or TypeScript. Use the REST API, n8n or Zapier to automate checks.
What teams check.
- Newsletters, briefs and posts Check the numbers, dates and names in a draft before it goes out
- Tax and legal memos Forms, deadlines, thresholds, what a court held
- Model-generated answers The factual statements in an answer, before a customer sees it
- Market and competitor claims Market figures and competitor prices checked against public sources
Full verification, in order.
Framing → Research → Debate → Panel Review → Conclusion
Lenz makes the statement precise enough to test, then searches public sources. Models from multiple vendors argue both sides using that evidence. Three automated reviewers assess the arguments before Lenz returns the verdict, the score, the sources and the reasoning trace. That is what a full verification does, and it takes about 90 seconds per statement.
The example above is a paragraph Lenz wrote and checked; every row links to the published result.
Public sources only. Text only. Lenz does not judge whether the text was machine-written.
Pricing
Free tier, no card: 100 assessments or 10 full verifications a month, in any combination. Developer is $99 a month for 5,000 assessments or 500 full verifications, the same way, and Scale is $399 a month for production volumes. Every plan includes 1,000 extractions a day, an extraction being one call however long the text.