Subscribe
Sign in
GPT Pilot — First ever REAL AI developer inside an IDE!
First ever AI developer inside an IDE is here! It can create development plan, write code, run commands, test app, debug and much more... Try it out and see how future of software development will look like.
81
Replies
@codewithimanshu it can work with any programming language you just have to specify it. For now it doesn't adapt to different coding styles if you use gpt4 in background (which we recommend) but you can use your own custom LLM that is trained on your codebase and then it will write in your code style
@codewithimanshu that depends on your LLM, but it shouldn't be too hard. We don't have interface for that so it would have to be changed in code but I can assist you. You would have to change endpoint you are hitting to your own LLM and that is basically it. If you are interested in trying it ping me on discord and we can hop on a call
I will always admire and support these kinds of projects.
Open Source dev is pretty challenging, so it's good to see projects that have got such great traction. Hats off to the Makers and contributors that made it possible! You have my respect.
Great work and good luck with this launch!
Sweet! Looks like it can really take care of a ton of boilerplate code when jumpstarting a new project.
Congrats, guys! I'm considering using it to rewrite parts of the product we have maintained for about ten years. If we can apply it there, it will lift a big burden off our backs.
@palagruza thank you so much for support. We will be happy to tackle your issues for you!
Sweet! When you say it writes production-ready apps, do you mean it writes the tests too?
@velusamy_subramaniam we are actually going in different direction. We are not code generation tool, integrated chatGPT or AI assistant like most tools. We created AI developer. GPT Pilot can create project from scratch to end, it can run commands, read output from them, test app, interact with developer, debug any issue and also add features to the project
@vedran_prica for now we support only VS code or you can use it as CLI https://github.com/Pythagora-io/... but we do plan to expand that in future
@vedran_prica Unfortunately not yet although JetBrain plugin is our next step after the VS Code extension.
Very cool!
I didn't quite get the repo comparison tool, how does that fit into the whole story?
As for code generation - do you see it as a pro or as a con that it doesn't use any high level framework? Can I get it to use one?
What about helping me later, can it help me continue developing my app?
What is the precision, how many mistakes does it make? What are some examples of the most complex thing is can build?
@martin_sosic Thanks for the support 🙏 Yes, seems like it's not clear - the repo comparison tool is the app that we've created with GPT Pilot. Currently, the most complex apps are around 1500 lines of code like the example one.
Regarding precision, it's hard to quantify how often does it make mistakes because it depends very much on the problem it is solving.
Regarding upgrading of the app - yes, definitely - once you start an app with GPT Pilot, you can always tell it to add more features. This is what I did in the video. After it created the full authentication with mail verification and password reset, I told it to add the repo comparison logic.
@martin_sosic repo comparison tool is just an app we created as a demo to show what can be done using GPT Pilot. Feel free to try any other app and let us know how it went, I would be really interested to hear feedback!
Regarding frameworks there are pros and cons, but we wanted to make sure that core of GPT Pilot work really good and then, later on, we can add option to pick specific frameworks or out of the box solutions for part of project.
Yes, after you are done with development, you can load project using gpt pilot and add new features or fix bugs you found and so on...
Congratulations on the launch! Awesome to see such a cool, AI-powered developer tool now in the market.
Looks great, congratz on your traction so far. Is this also suitable for data science tasks in Python or R? What is a typical use case for this product?
@josip_lebesgue Currently, the use case that works best are web apps because of the testability but it can be used for DS as well - GPT Pilot core tech is language agnostic.
Looks very interesting. Have you thought about how to differentiate from similar products, like GitHub's Co-Pilot? In any case, I support you and vote for you.
@alex_egorov yeah, we are actually going in different direction. We are not code generation tool, integrated chatGPT or AI assistant like most tools. We created AI developer. GPT Pilot can create project from scratch to end, it can run commands, read output from them, test app, interact with developer, debug any issue and also add features to the project
Pythagora