Imagine you have to make a little change in all your microservices, like changing a line in Dockerfile, but you have 50 microservices, ouch right! The PRs, pipelines, reviews... I have created a dashboard with a chatbot where you can simply ask "Update all repositories with alpine base image version X to Y", and then click next a bunch of time and see a list of PRs with their pipeline status.

Hi friends I'm Daniel, the creator of Infra as AI. I was laid off from my SRE role a couple of months ago, and decided to build something that could always help me at work. My job was to keep the lights on; For example: Make sure the pipelines are consistant across the org, but this was hard, I had to open each repo, make a little change, and make a PR with all the enterprise seremonies that comes after, and then I had to do this for many repos and many teams; It was hard, I could not remember which team owns what. Also, sometimes the pipelines fail cuz some repo had a deprecated function, so I had to read the changelogs; And I would know it fails after opening 200 tabs of CI files.

At work, I solved this the way Kernel does, with git patches. Our workaround was to generate that change in one repo, create patch files, and then `git apply` on each repo using workflows. Now with the help of HR, I got time to hack around something I always needed. I just ask for a change, and then let the agent read the docs, find out the deprecation, make those PRs, and show me a list of them with their pipeline status next to it. For this, coding agents where perfect, they could read the changelogs, documentations, readmes and all that and also work in parallel!

## What to expect

We first index a repository, which means we extract its important characteristics of after cloining it, and then save it in a vector database for future similarity searches. Then user asks for a change in the chatbot, our platform suggests candidate repositories for approval, and then runs Claude Code on them in parallel. This is followed by a list of PRs filled with details ready for the user to review and create.

## Tech

LangGraph: The agent is built on top of it GPT5: Finds relevant repositories Sonnet: Your coding agent pinecone: Holds repositories vectors context7: Getting docs

TL;DR: I created a platform where you can run a query against your codebase, and create a massive number of PRs to make simple changes in your microservices. Please have a look and let me know your feedbacks <3

PS: I'm giving away trials with 2 euros credits so you can index and query your repositories, but please watch for my pocket too ^^