AI Research Assistant — Verify Any Claim, Find Real Sources
Paste any statement. Get a sourced verdict in about a minute — independent investigation, two-sided AI debate, cited references you can click through.
Why researchers use Lenz
- Real sources you can click and read yourself. Every verdict comes with the actual evidence Lenz used to reach it — click straight through to the studies, datasets, and primary documents.
- Evidence traced back to the underlying study. Lenz follows claims past Wikipedia and secondary summaries to the source material that actually backs them.
- The cross-check done in about a minute. Independent investigation across multiple sources, with a verdict that surfaces what the evidence agrees on and where it disagrees.
- A sourced verdict, not a confident guess. Two AI advocates argue both sides; three independent expert models score them. You get the verdict and the receipts.
How Lenz researches a claim
- Paste the claim — A study finding, a statistic, an assertion you read somewhere. Anything specific you can verify against evidence.
- Lenz scans the open web — Across studies, news, datasets, primary documents, and reputable secondary sources. Evidence collected on both sides.
- AI models debate — Two AI advocates argue opposing sides. Three independent expert models then score the strength of the evidence.
- Cite the sources — A 1–10 confidence score, a sourced conclusion, and the full list of cited references you can use in your own work.
The whole process typically takes about a minute or two. Read more about how Lenz researches claims.
Pre-check before you
Before you cite
A statistic, study finding, or assertion you’re about to commit to in a paper or essay. Pre-check it first — the cited references double as your starting bibliography, ready to read in the originals.
Before you publish
Journalism, briefs, content destined for an audience. Run the underlying claim through Lenz before it ships. Walk away with both the verdict and the source list.
Before you trust the AI
ChatGPT, Gemini, and Claude sound equally confident whether they’re right or wrong. Run the claim through Lenz, see both sides debated, and decide on evidence rather than tone.
Recently researched
Cloud workflow insights released by an unspecified organization reported that 98% of nearly 3,000 monitored organizations across the United States, the United Kingdom, and Australia received a throughput alert from a supplier domain during a 7-day window in February 2021.
Mercuric chloride (HgCl₂), historically known as corrosive sublimate, is now primarily restricted to laboratory research and industrial catalysis due to its extreme toxicity.
Drinking coffee late in the day can disrupt sleep for many people.
Browse the full library of verified claims.
Frequently asked questions
Is the AI research assistant free?
Yes. You get 10 free research queries per month, no credit card
required. Plus and Pro plans unlock more if you need them.
How many sources does Lenz check?
Lenz scans the open web for the strongest available evidence on
the claim — across studies, news, datasets, primary documents,
and reputable secondary sources — and surfaces the most
relevant findings in the verdict. Every cited source is scored as
supporting, refuting, or neutral, with a relevance score so you
can prioritize.
Can I cite Lenz in a research paper?
Cite the underlying sources Lenz found, not Lenz itself. Every
verification gives you a real list of references you can use
directly in your bibliography. Lenz is the research assistant, the
cited evidence is the actual scholarship.
Does Lenz verify whether a citation exists or accurately represents a paper?
No. Lenz verifies claims, not sources. It takes a
statement and evaluates whether it holds against the broader
evidence. It does not validate that a particular paper exists or
that someone summarized it correctly. For that, use a citation-
checking tool.
How is this different from Perplexity or ChatGPT?
Lenz pre-checks one specific claim — independent
investigation, two-sided debate among multiple AI models, sourced
verdict on a 1–10 scale, full reasoning trail. Different
question from “give me an overview”, different output.
Use them together: get the overview from Perplexity or ChatGPT,
pre-check the specific claims with Lenz. Read about
why Lenz exists.
What about contested claims?
When the expert models disagree, Lenz shows the disagreement
instead of papering over it. The disagreement is the transparency.
You see the individual scores, the arguments, and decide what to
make of them.
Does it handle non-English sources?
The pipeline runs in English, but the underlying search can surface
non-English sources when they’re relevant.
Research smarter. Cite real sources.
Paste any claim. Get a sourced verdict and a real reference list in about a minute.
Start researching a claim