Build task-oriented custom agents for your codebase that perform engineering tasks with high precision powered by intelligence and context from your data. Build agents for use cases like system design, debugging, integration testing, onboarding etc.
Aditi here, and I couldn’t be more excited to finally launch Potpie—our open-source platform for building custom AI agents to automate engineering workflows across the SDLC.
Dhiren and I have been friends for a decade, going all the way back to our engineering college days. From endless hackathons to countless late-night coding sessions, our friendship was built on a shared obsession: solving real-world engineering problems.
Dhiren’s journey has always revolved around developers. Whether at Cisco or his stint at a fast-paced startup, his passion has consistently been to build internal tools that make engineers’ lives easier. I’ve navigated product and engineering roles at Zoho and Razorpay, focused on building dev tools that have directly saved companies hundreds of thousands of dollars annually.
Potpie wasn’t originally planned—it emerged organically from one of our side projects. Initially, we just wanted to automate integration testing with AI agents. But as we dived deeper, we realized the true potential of these agents extended far beyond testing. What if developers had a platform to effortlessly spin up custom AI agents tailored for any engineering task?
And so, Potpie was born.
We strongly believe that an AI agent’s performance is entirely shaped by its circumstances—the tooling it uses, the context it understands, and the knowledge it has access to. That’s why Potpie deeply understands your codebase, provides precise tooling like semantic search and code retrieval, and autonomously executes tasks for you.
Making Potpie open source was a no-brainer for us. Tech, we believe, should always be open, accessible, and community-driven.
We’re thrilled to finally share Potpie with the world and can’t wait to see how it empowers you.
Excited for your feedback and to hear your stories!
> Hey POTPIE AI I’ve been following your tool — it’s super interesting!
I’m a UGC creator specializing in short-form video for tech tools and startups. I’m currently offering 2 free videos to select brands to build my portfolio.
I’d love to make something engaging for [POTPIE AI] that shows off your product in a clean, relatable way.
To ensure my content quality you can visit @the_tech_gagan my insta page. I'm sure you'll not get disappointed.
If you’re open to it, I can send over a draft script or get started this week.
Let me know — would love to collaborate!
Congrats on the launch! 🎉 Really loving how clean and snappy the UI looks — it’s the kind of experience that makes you want to keep exploring. Looks like a solid product!
Congratulations folks.. This looks super interesting and I will give it a shot soon!
@aditi_kothari For teams working across evolving codebases, how does the platform handle drift between an agent’s initial training and later code changes? If a critical dependency gets refactored, does the agent self-correct using semantic search, or does it require manual context updates to stay reliable?
Just came across potpie and tried it on my ongoing project. Tbh this is feels like I have a team of engineers to whom I can offload the boring parts of the development.
Btw do you folks plan on brining in an agent specialized in handling deployments and infra?
nice product, congrats on the launch! curious though, as a non-tech person who actually has a tech co-founder - would there be a use case for me? simple coding stuff I could do that would work well within the larger ecosystem?
All the best @aditi_kothari and Dhiren! Product looks slick
@aditi_kothari and team has done a great job, making life of a builder simpler.
Congrats on the launch guys! Open-source FTW! Btw, are there any guardrails to prevent "rogue agents" in the ecosystem?
Congrats on the launch, Aditi! Potpie sounds like a valuable tool for enhancing engineering workflows. Looking forward to seeing how it evolves!
