Lenz ads

🚀 𝗟𝗲𝗻𝘇 𝗳𝗼𝗿 𝗚𝗼𝗼𝗴𝗹𝗲 𝗗𝗼𝗰𝘀 𝗶𝘀 𝗻𝗼𝘄 𝗹𝗶𝘃𝗲 𝗼𝗻 𝘁𝗵𝗲 𝗚𝗼𝗼𝗴𝗹𝗲 𝗪𝗼𝗿𝗸𝘀𝗽𝗮𝗰𝗲 𝗠𝗮𝗿𝗸𝗲𝘁𝗽𝗹𝗮𝗰𝗲! We've brought Lenz's claim verification and citation checking directly into Google Docs, making it easier for writers, editors, and content teams to research and fact-check their work without leaving the document. Good content takes research. And checking facts, figures and cited sources is often one of the most time-consuming parts of getting a piece ready to publish. With the new extension, Lenz reviews claims and cited sources across a draft, flags what needs attention and brings the findings, supporting evidence and suggested corrections right next to the relevant text. It's not about replacing the writer's judgment or the editor's review. It's about giving them better tools to do the work they already do, with less manual research and back-and-forth. We ran a 𝗕𝟮𝗕 𝗺𝗮𝗿𝗸𝗲𝘁𝗶𝗻𝗴 𝗱𝗿𝗮𝗳𝘁 through the extension and found a couple of interesting examples of what Lenz can catch. Sharing those in the comments below. 👇 🙌 𝗘𝘅𝗽𝗹𝗼𝗿𝗲 𝗟𝗲𝗻𝘇 𝗳𝗼𝗿 𝗚𝗼𝗼𝗴𝗹𝗲 𝗗𝗼𝗰𝘀: https://lnkd.in/d5zewNUb 📥 Install from Google Workspace Marketplace: https://lnkd.in/dvWTC-Qm #ContentMarketing #FactChecking #GoogleDocs #AI

When the same model writes an answer and then grades it, what you get back is one model's opinion twice over. Lenz checks the claim against independent published sources instead, so the answer comes from somewhere other than the thing that produced it. A quick check gives you a verdict and how confident it is in about ten seconds. When something really needs to hold up, the deep check investigates the claim: it researches independent sources, has two models argue both sides, and has three reviewers score the evidence. You get back a score out of 10 and the sources it rests on.

We gave five frontier models the same 997 real-world fact-checks, and they failed to agree on nearly two thirds of them. Whichever model you've got sitting in your stack is making that call on its own, and a different one would often have called it differently. Lenz runs the claim through framing, research, debate, panel review and conclusion, then gives you one verdict along with the sources it rests on.

The person who wrote a claim is badly placed to check it, and the model that generated one is worse still. Lenz checks factual claims against independent published sources and returns a verdict, a confidence score and the citations it rests on. A quick check takes about ten seconds. When something really needs to hold up, the deep check researches the claim, has two models argue both sides, and has three reviewers score the evidence.

Verifying a figure before it goes out takes seconds. Correcting it afterwards takes a retraction, an awkward email, and whatever credibility went with it. Lenz pulls the checkable claims out of a draft and gives you a verdict on each one, so you are reviewing what got flagged rather than reading the whole thing again with fresh suspicion.

We gave five frontier models the same 997 real-world fact-checks. They failed to agree on nearly two thirds of them. Whichever model sits in your stack is making that call on its own, and a different one would often have called it differently. Lenz isn't affiliated with any model vendor: it checks the claim against independent published sources and returns one verdict, with the evidence behind it.