TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Ask HN: What GEN AI tools you built

3 pointsby 2bor-2nabout 1 year ago
I'm curious to learn about the innovative tools or products you've either built for yourselves or discovered online, powered by Generative AI. Whether it's a personal project to streamline your workflow, a creative tool that sparks joy, or a product that has significantly impacted your daily life, I'd love to hear your stories.

2 comments

ogouabout 1 year ago
Art Review Generator. 4 years ago I released a small book of images made with an early GAN tool. I thought it would be interesting to have an AI write the introduction to the book based on analysing the images. My plan was to do some basic image sentiment and classification with AWS and feed those tags into a LLM for a few rounds of generation. Back then, all that was much more primitive and I ended up building a bunch of bridge code to make it happen. The real trick was getting something that sounded like actual art language. That didn&#x27;t exist, it was much too general, so I made a tool. I scraped 50+ years of ArtForum Magazine reviews and trained a model myself. After using it on my own project, I thought it might be interesting to build a free site for others to experiment with it. So, artreviewgenerator.com was born. My big achievement is not using GPUs for the generation. All old school multi-core server muscle. It gets a fair amount of traffic and I&#x27;m now on various mailing lists as an AI company owner, which I find hilarious.<p><a href="https:&#x2F;&#x2F;artreviewgenerator.com&#x2F;" rel="nofollow">https:&#x2F;&#x2F;artreviewgenerator.com&#x2F;</a>
standfestabout 1 year ago
atm mostly RAG setups for corporations, and a little bit in the compliance domain. not super exciting. last year at iccv paris, we did something correlated to HouseGAN and presented a new dataset to train such setups: <a href="https:&#x2F;&#x2F;data.4tu.nl&#x2F;datasets&#x2F;e1d89cb5-6872-48fc-be63-aadd687ee6f9" rel="nofollow">https:&#x2F;&#x2F;data.4tu.nl&#x2F;datasets&#x2F;e1d89cb5-6872-48fc-be63-aadd687...</a> and <a href="https:&#x2F;&#x2F;zenodo.org&#x2F;records&#x2F;7788422" rel="nofollow">https:&#x2F;&#x2F;zenodo.org&#x2F;records&#x2F;7788422</a>