Connect with the Product Hunt community

Hello, hello

With a new, shorter iteration to Prosaic, two pesky bugs have been fixed:

• Creating a file no longer overwrites existing files; shows error instead.

• Entering .md extension in filename no longer creates double extension.

Two days ago, I shared the 10k mistake product owners make with AI search. The response was overwhelming.

Since then, we've more than doubled our dataset at Rankfender. And many found they were invisible.

But here's what scared me more:

Of those who WERE visible, 43% had incorrect information in AI answers.

Tobira launched yesterday and we ended up #1 with 568+ upvotes on a Monday. I still can't quite believe it.

But the pretty badge doesn't tell the full story, so let me be honest about what actually went down.

Hey Product Hunters

We're back. And this time, with a much sharper focus.

When we first launched @Curatora , it was a content discovery and publishing tool. Over time, one thing became very clear from user conversations.

Publishing was not the hard part.
Finding strong, timely content ideas was.

Hey everyone!
We just shipped TurboQuant into OpenCut AI, and this one changes what hardware you need to run the full AI stack.
The problem we had
OpenCut AI runs everything locally -- LLM, transcription, voice cloning, image generation. That's great for privacy, but brutal on memory. Running the full stack needed 35+ GB RAM. Most of our users have 8-16 GB laptops, so they were stuck with tiny 1B models that gave mediocre scripts, slow commands, and limited context.
What TurboQuant does
TurboQuant implements two algorithms from Google Research paper PolarQuant and QJL. That compress the KV cache (the biggest memory bottleneck during AI inference) by up to 6x with mathematically proven quality preservation.
In plain terms: your AI models now use a fraction of the memory without getting dumber.
Before vs After
On a 16 GB machine:
- Before: Llama 3.2 1B + Whisper Base + TTS = barely fits, mediocre quality
- After: Llama 3.1 8B + Whisper Medium + TTS = runs comfortably, dramatically better output
On an 8 GB machine:
- Before: Could only run the 1B model alone
- After: Runs a 3B model + Whisper Base + TTS together
Full stack memory:
- Before: 35 GB for everything
- After: 15 GB for everything
What this means for editing
- Better AI commands "remove the intro" actually works now because Mistral 7B understands context far better than a 1B model
- Better transcription Whisper Medium fits where only Whisper Base could before, so captions are more accurate
- Longer content: Process hour-long podcast transcripts without running out of memory. The 6x KV cache reduction means 6x longer input context
One-click setup in Settings
We added a new AI Optimization panel in Settings. It auto-detects your hardware and recommends the best configuration:
- Performance Tier: Lite (4-8 GB), Standard (8-16 GB), or Pro (16-32 GB). Each tier is tagged with "Best for your hardware" based on your actual RAM.
- KV Cache Compression: Pick 4-bit (near-lossless), 3-bit (5x compression), or 2-bit (aggressive). Recommended level highlighted based on your system.
- Memory Budget: Set once, and the system optimizes everything to fit.

Would love to hear, what's your RAM situation, and does this make local AI editing viable for you?

What are the alternatives to Stripe?

Paddle

got rejected

Hey everyone,

As much as I use Gemini day-to-day, the chat interface is starting to feel a bit cluttered as my history grows. I've been thinking about the quality-of-life improvements that would make the experience much smoother, but I'm curious to know what the community thinks is the highest priority.

PinchBench is the leading @OpenClaw benchmark.

The team is looking for contributors to make it even better.

Hi everybody, have a great Friday

Just wanted to thank you all for the support over the launch of Toone this week couple days ago!

This meme made me laugh because it is painfully real.

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:

While building a product, I ve also been trying to run content on social media to bring in more traffic. I experimented with creating AI-generated characters and producing UGC-style videos around them.

During this process, I realized something interesting: there are hundreds of tools that can generate virtual characters and UGC-style videos. But what actually makes a video engaging isn t the tool - it s the authenticity of the person creating the content.

If you have 20 minutes to spare today, I highly recommend you watch this talk by Rory Sutherland.

I was completely engrossed for the full 20 minutes because he delivers so many interesting insights.

Hey Everyone!

TabDog new version 3.0.0 is just released!

shipped a bunch of updates today:

featured projects with usdc on base

• you can now promote your project to the homepage carousel using usdc on base chain

• packages: 1 day ($0.50), 3 days ($1), 7 days ($2), 30 days ($5), lifetime ($15)

• integrated privy wallet connect supports rabby, metamask, coinbase wallet, rainbow, walletconnect, and email login

• no stripe, no paypal, no country restrictions works for builders everywhere

mandatory socials

Been iterating on Room Service quite a bit recently and wanted to share a quick update.

With tools like Claude Code, we re entering a world where products can be built end-to-end inside AI workflows.

But there s a problem I ve been thinking about: As building gets faster, user research risks getting left behind.

At Velozity, we ve been working on solving this by launching an MCP server that brings user research directly into these environments.

Now inside your coding workflow, you can:

Most developer portfolios are static you add your projects once and that's it. No reason to come back.

We just shipped a Live Feed on VibeTalent that changes that. It's a unified activity stream showing everything happening across the builder community:

• GitHub commits & PRs merged

• New projects shipped (with tech stack and live links)

• Daily coding streaks (30+ day milestones get highlighted)

• New builders joining the platform

A few things that make it interesting:

We re launching Glam AI today: an AI platform for creating trendy image and video content.

To celebrate the launch, our team created a fun video with the Product Hunt team (@rajiv_ayyangar and @rrhoover). You can see it on X.

When you re-record a demo, it's rarely because the capture is broken. It's usually something small, maybe a line you didn't like, a weird pause, or the webcam overlay blocking part of your demo.

And at that point, you're either re-recording the whole thing or dragging the file into a separate editor just to fix one moment.

Hello Product Hunt!

Charlie here, co-founder of Trophy. Since we launched Trophy 1.0 back in Jan we've been continuing to ship new features to Trophy, and March was no exception. One of the biggest updates was the addition of points levels so I thought I'd take the time to share that with you here.

While building RepoLens, one thing became very clear to me:

understanding a repository is one problem, but understanding what changed and what actually matters is a different one.

That s the direction behind RepoLens Version 2.

I m trying to learn which signals would be most valuable for real engineering teams when repositories are moving quickly.

Road to 1,000,000 Votap users Day 60 | Current: 1294
Over 100 users have already chosen their fate. Will you be the next?
That s how many people have already generated their username through the new update. Now onto step two:
I m currently working on making conversations actually feel like conversations.
You ll be able to:
tag the person you re replying to
get notified when someone responds
go back and forth in proper threads
Not just random one-off comments anymore.
Download Votap from the App Store if you want to try it when it comes out.
More tomorrow.

Hey Product Hunt!

We ve all been there. You design a beautiful, complex layout in Figma, only to have it die a slow death during the "handover." Either the developer says it s impossible, or you spend three days fighting with nested tables and inline CSS to get it to look half as good in an inbox.

Launched last week, open-source frontier model @MiniMax M2.7 scores 56.2% on SWE Bench Pro, converging towards the best proprietary models like @Claude by Anthropic Opus 4.6.