Spent years in Amazon KDP self-publishing and always hated writing book descriptions from scratch.

So I built PublishAI you enter your book topic, pick a genre, add keywords, and get a full book description, title ideas, and Amazon keywords in seconds.

It's completely free right now. No signup required.

Would love feedback from anyone here who writes, publishes, or knows authors who struggle with this. What would you add?

Hi everyone!

Hey Product Hunt,

I'm the builder behind Threelane, shipping it from Dubai.

Why I built it

I make a lot of product demos and tech reels. Every smooth multi-cam tool I tried was either subscription-locked, cloud-locked, or both. Loom is fine for talking heads but breaks once you want a second camera. Riverside is great but pricey and online-only. ScreenStudio nailed cursor zooms but it's Mac-only and paid. Nothing was free, local, AND multi-cam. I got tired of waiting for someone to build it, so I did.

Hey Product Hunt community!

I m the maker behind SimuScore.com. As a developer and a massive football fan, I ve always found traditional tournament brackets a bit too rigid especially now that we are entering the historic, chaotic era of the first-ever 48-team FIFA World Cup with 12 complex groups!

That s why I built SimuScore. I wanted to create a fast, dynamic, and data-driven platform where anyone can instantly simulate every group dynamic, project those massive under-dog upsets, and map out the entire knockout path all the way to the final.

What makes SimuScore different?

Hey everyone, quick self-promo but I d genuinely love feedback.

I ve been working on an AI Vocal Remover for creators who want to separate vocals from songs for remix ideas, karaoke practice, cover practice, short-form video audio, or cleaning up demo material.

The tool is here:

https://freemusiccreator.ai/ai-v...

Recently, I had the opportunity to work on VDOMA.AI - a startup where I was responsible for designing the MVP from the ground up. For me, this project became much more than another design task. It was a chance to join the product at its earliest stage, when there were no predefined flows or ready-made solutions. Together with the team, we were shaping how the product should work before designing the first screens.

About the project. VDOMA.AI is a startup focused on simplifying apartment search and rentals through AI. The idea was to reduce the time people spend searching for housing by combining AI-powered search, personalized recommendations, and tools that help landlords automate routine communication.
The Problem. Finding a place to rent is still a frustrating experience. Users often browse dozens of outdated or duplicate listings, communicate with multiple landlords, and spend hours filtering apartments that don t actually match their needs. On the other side, landlords answer the same questions repeatedly and struggle to manage listings efficiently. We wanted to rethink this experience instead of simply creating another rental marketplace.
The Solution. Our goal was to build an MVP where AI becomes a real assistant rather than just another feature.

Introducing Rescript: an open-source text-based video editor.
Edit video and audio files by simply editing the transcript text.
Drop in a video and it is transcribed locally with per-word timestamps and speaker labels. Delete words in the transcript and the corresponding clip is cut from the video.

Runs fully in the browser: Local, free, offline and open source.

Disclosure: I work with the LynavoDrive team.

We open-sourced the local core of a computer-first file tool, then wrote down the product boundary we kept seeing shortened in discussions:

- initial setup needs internet and the first connection happens on the same LAN;

- after setup, LAN file access can work without public internet;

Working with PDF documents online is often a frustrating experience. Popular online document utilities frequently force users through subscription paywalls, restrict free usage to 2 tasks per day, or slap intrusive watermarks across confidential files.
To solve this, we engineered TheyLovePDF https://www.theylovepdf.com a fast, 100% free online PDF suite built for high-speed processing directly in your browser.
Architectural Focus: Client-Side Speed & Privacy

Traditional PDF tools process every document on external backend servers, creating queue delays and privacy risks. TheyLovePDF utilizes modern browser technologies to execute heavy operations client-side whenever possible:
1. Unlimited Merge & Split PDF: Drag and drop multiple PDF files to combine them or extract specific page ranges instantly with zero watermarks.
2. Target Size Compression: Compress large PDF files precisely under 100KB, 200KB, and 1MB for portal uploads and email attachments.
3. Multi-Format Document Conversion: Convert between PDF and Word (DOCX), Excel (XLSX), PowerPoint (PPTX), and high-resolution JPG images.
4. Digital Signatures & Editing: Sign agreements online, fill out PDF forms, rotate pages, and apply password protection.
5. AI-Powered OCR & Chat: Extract text from scanned bank statements and interact with complex PDF documents using natural language AI.
Zero-Trust Privacy Model
Document security is our top priority. Transferred files use 256-bit SSL encryption, with automated server deletion within 2 hours. We do not store, track, or monetise user document data.
Try It Free
If you are tired of daily processing limits or watermarked exports, check out the web app:
https://www.theylovepdf.com
We would love to hear feedback and feature requests from the DEV community!

I spent months job hunting this year, and what wore me down wasn't rejection. It was not knowing which ads were real.

So I built WorkstationAI. Paste a job ad and get an A to F "should I apply?" score with a scam and ghost-job check, then CV tailoring, company research, and interview prep built from live research on that company. Free tier, pay per use, launching here on 4 August.

The hard part has turned out to be calibration, not the model. Telling someone "this is a D, don't apply" only works if they trust it. Right now it reads the ad text only, with no live check on whether the company is actually hiring. Flagging a genuine ghost job without crying wolf on every vaguely written ad is the thing I keep rebuilding.

Two questions for anyone who's been through it: