Launched this week
Lettertrace
Track your AI visibility for free (using your own API keys!)
627 followers
Track your AI visibility for free (using your own API keys!)
627 followers
Lettertrace measures how often Claude, ChatGPT, and Gemini mention your company. But there's a catch: it's free, developer-first, bring-your-own-key, and open source.
Lettertrace
Hey Product Hunt!
Mathew here. I'm the founder of The Letter Company.
Today, we're debuting one of my favorite things that we've ever built. You've likely heard of AEO. Or GEO. Or AI Search. It's a popular space rife with tools. But there's been a big problem: it's pay-to-play. Even to find out how you're doing on AI Search, you need to fork over something crazy like $250/mo.
And I beg.... for what? AI search is effectively measured by a crafty cron job that orchestrates a bunch of model provider calls.
No more $250/mo fees. Now, it's free. Lettertrace is a BYOK telemetry tool for measuring how you're doing on AI search. It's open source. MIT license. Free even on our hosted version. There are no paid tiers. None. Nada. You'll never put down a credit card. We're moving beyond that. It's time we get telemetry on our GEO, just like we had on our SEO, without having to pay for it.
Lettertrace is super easy to use. Like SO easy. Why? Cause it's got an amazing CLI. Just install it, give Claude Code or Codex access, and your coding agent will go to town. It'll add the topics your buyers actually ask about ("best CRM for startups"). It'll generate realistic prompt variations and run them against Claude, ChatGPT, Gemini, and other major LLMs. It'll track visibility, share of voice, prominence, and sentiment.
npm install lettertrace. claude code. prompt. That's it. Or, if you're non-technical, sign-up traditionally.
And it's actually open source. Not open core, none of those gimmicks. A clean MIT license. Bring your own key. You pay the model providers directly, no markup on top. You'll spend ... maybe $3? If you're just paying to measure AI search, you're getting ripped off and we're putting an end to that.
Three things I'd genuinely like feedback on. First, which answer engines matter enough to add next. Second, whether the generated prompt variations read like questions real people ask, because that's the piece I'm least sure about. Third, did the video effectively explain our purpose as a company?
Code: github.com/letterstory/lettertrace
@shabb_katoch Thanks a lot
Lettertrace
@shabb_katoch Thank you Shabnam!
@pregasen Outstanding launch, excited to see Lettertrace quickly become the go-to standard for measuring brand visibility across LLMs, qq is there support for scheduled cron triggers built into the CLI, or is it intended to be hooked into existing GitHub Actions / CI pipelines?
Lettertrace
@priya_kushwaha1 There is support! Doesn't need to be part of your CI Pipeline at all actually. It runs on any arbitrary clock.
@pregasen Congrats on the launch Mathew. Great vid....lead with top 5 hopefully. Good on you for open sourcing.
Lettertrace
@zolani_matebese thank you!
@pregasen Congratulations on the launch! Free + open source is a real gift for founders.
One thing I'd love your take on: since it's BYOK, non-technical folks still have to go set up their own Anthropic/OpenAI/Google accounts before they can run it - that's a small ask for a developer, but a real wall for the small business owners you mentioned. Any plans for a managed-key path down the line? I can see this tool being very helpful for SBO.
Inkle
Following this closely since I track something similar for my own site. One methodology question that would actually change how I read the numbers: how do you tell the difference between a page getting pulled in as a citation or source versus the AI actually naming the brand in the answer text itself? I have watched my own citation count jump sharply on one engine in a single day while actual brand mention count stayed flat at zero the whole time, so those two clearly do not move together for me. Curious whether you track them as separate metrics or treat citation as a proxy for visibility.
Lettertrace
@abdullah_javaid3 Yep, these are diferrent things. We can detect it semantically. How often it's cited, how often it's mentioned, and the sentiment on the mention itself!
@pregasen That answers it, thanks. One more layer I am curious about: when a citation happens without the brand explicitly named in the answer text, does that count toward the citation metric on its own, or does the sentiment score only attach once there is an actual named mention to hang it on? Trying to work out whether a flat mention count with a rising citation trend means invisible but trusted, or just not being read yet.
Progress AI Observability
Ooh, love love love that this is dev-first and oss! Some insight on what you asked - "which answer engines matter enough to add next" - you've hit the main ones we track at Progress, but Grok could be good and for certain niches also Meta AI
Lettertrace
@denitsapenchevavaltchanova Awesome Grok and Meta AI is on next week's roadmap. We'll add!!
The question I have not seen asked yet is what makes the number stable enough to trend. Ask the same model the same thing twice and you get different answers, and a silent model update moves your baseline without telling you. So a line going up is mixing your own work with drift you cannot see.
Cheap fix, and I think your architecture already allows it. Measure a control set in the same run, a few competitors or unrelated brands you are not touching. Then drift shows up as a common shift across all of them and you can subtract it out. Being BYOK and open source also means you can pin model versions and re-run history, which the 250 a month tools cannot really offer.
Bhava
Congrats on the Lettertrace launch, Mathew. The open-source + BYOK framing is very clean, and I loved that the website scan produces editable questions before the first run.
One launch-day thought from trying it: when a first run returns 0%, a new user needs one obvious next move. Is it prompts, competitors, model coverage, or simply a brand that is not yet visible? Turning that zero into a small diagnosis could make the first run feel much more actionable.
Lettertrace
@riya_jawandhiya the rest of our stack helps with that! https://letter.company
Chromatic Lens
Would love to have a single docker run command so I can test it on my own infra in one click. (The current github repo doesnt have the Dockerfile / image either.)
Anyway Thanks for building it, love opensource.
Lettertrace
@dhanushreddy29 We could make that happen. @cmillstein could you help with that?
Lettertrace
@dhanushreddy29 @pregasen absolutely! I’ll get this up and running today and then follow up here :)
Lettertrace
@dhanushreddy29 @pregasen Added! There's a Dockerfile and a published image now. Setup's in the README: https://github.com/letterstory/lettertrace#run-it-with-docker
Congrats on the launch! Really like the open-source + BYOK approach.
How do you decide which prompts are most relevant to track for a company?