Inspiration
AI is no longer limited by what it can do, it's limited by discoverability. Thousands of AI tools, plugins, and specialized assistants already exist, but most people never find the ones that fit their workflow. Instead, they rely on ChatGPT for everything, manually stitching together repetitive processes because they don't know a better solution exists. Intervene AI tackles this distribution problem by understanding how people actually work around ChatGPT and surfacing the right tools at the exact moment they're needed.
What it does
Intervene AI watches how you use ChatGPT, spots the tabs you keep bouncing between, figures out what you're actually trying to get done, and suggests plugins that eliminate the busywork. Because if you've copied the same thing into ChatGPT twelve times, there's probably a better way.
How we built it
Everything was built with the help of Codex and ChatGPT.
Challenges we ran into
This is the very first time I've built something slightly technical, so everything was a challenge!! The original scope was to analyse the user's workflow with rrweb, and an LLM would match the user with an AI-powered app from a wider directory, but I've hit some issues with scraping wider directories, so I decided to build a minimal MVP focused on recommending ChatGPT integrations and custom GPTs.
Accomplishments that we're proud of
Being able to build something that functions, for the first time ever, and in under 3 days!
What we learned
What's next for Intervene AI
Built With
- chrome
- codex
- gpt
- manifest
Log in or sign up for Devpost to join the conversation.