A year ago, half the marketing world told us "AI search" was overhyped. The other half was shipping "ChatGPT SEO checklists" in a week.

We ignored both.

Instead we did one boring thing: we scraped LLM citations. Every day. Across ChatGPT, Perplexity, Gemini, Claude. For hundreds of brands. And we asked one question when AI recommends a product, where does that recommendation actually come from?

Here's what we found that nobody was talking about:

I said "never again" to hardware.

I meant it. I had done it before. I know what hardware costs, not just in money, but in decisions you make at 2am about components that may or may not arrive, about inventory that ties up capital for months before a single unit ships. When I moved into SaaS, the relief was real. Software scales. Software does not sit in a warehouse. 

Hey ProductHunt!

Trophy is now powering over 24M streaks which is kind of crazy to think about considering we only launched 1.0 here in January this year.
One of the parts I find most interesting about building horizontal infrastructure is that as you scale and power more and more products you get to see insights that most teams building in isolation will only see a part of, and you can use those insights to make the the infrastructure better for everyone.
For example, because we power streaks for so many users, Trophy can tell that 25% of all streaks are lost on a Friday, closely followed by Saturday (19%) and then Wednesday (18%).

Hey everyone

ClawOffice was a bit of a gimmick - a 3D virtual office for AI agents. It was fun to build and got some attention on launch, but let's be real: it didn't take off. People thought it was cool for a minute, then moved on. No real retention, no real problem being solved.

Here's what I actually learned from it:

• Novelty value. A cool concept gets you a launch day. It doesn't get you users who come back on day 30.

• I was building for the demo, not the workflow. ClawOffice looked great in a screenshot. It didn't solve anything measurable for anyone.

• "What gets tracked gets improved" is real. The founders I talked to afterward all had the same pain - they were shipping features and running experiments with no clue what was actually driving revenue.

no one asked actually but to be honest i think i had to say something cuz i kinda feel bad that all the support i got here just went away...
right now if you visited the domain finkitty.com you will find out that its listed for sale, i took this decision after a very long sitting with myself and ended up deciding that since im not having any users in this app i might just kill it and shift my focus into something else (working on templateson.com now)
yet im still holding it inside cuz i do like the name of this app and i feel like it has very good potential and i just cant see it..
so... if you have any great idea for an app named "FinKitty" please let me know
thanks

Hey Product Hunt community! I'm Sankalp, founder of Ordrpro. We're live on Product Hunt today and I wanted to share our story.

THE PROBLEM: Most restaurant owners have no idea how much revenue they're silently losing every month. We're talking 15-30% of potential revenue gone to mispriced menu items, customers who ordered once and never came back, peak demand hours not being utilized, and operational blind spots draining margins.

Hey all,

One of the most important and challenging experiences you ll have as a founder is fundraising.

I genuinely love listening to podcasts. It's one of the best ways I've found to stay on top of new trends, pick up strategies I wouldn't have discovered otherwise, and come across founders and operators I'd never stumble on through regular reading.

So I'm always on the lookout for new ones worth adding to the rotation.

For 20 years, SEO was a human game.
You wrote for people, optimized for Google's crawlers, and built backlinks by convincing other humans to link to you.
The inputs were human. The outputs were human.

GEO is different. You're optimizing for language models that extract and synthesize. The inputs are structured data, schema markup, comparison tables. The outputs are citations, not clicks.

So where does the human fit now?

What the data says about AI's performance: