AI is accelerating many conversations around speed and quality but a fully-automated development process is just emerging yet. What about experiments then?
.jpg)
I've been testing many different tools, from features to entire platforms, to test and understand how far is going quality when we look for tooling support to run a full-development cycle (from UX to cyber security) through automation in any sense, and just got amazed by Replit team last year.
Anyway, since then (~june 2025), I didn't have a chance to use it for a professional demand until some months ago. A small but yet relevant pain point emerged from some teams at a startup I was collaborating with: agile retrospectives were needed as much as a truly free whiteboard to conduct this kind of meeting. And happily, no constraints around AI tools were made for this bucket, so I was ready to start.
From there it borns Retrozin (https://retrozin.replit.app/) - a small and specific tool to allow different people who are in charge to conduct one or more retro meetings with one or more groups.
It was my first professional tool entirely built with AI support for all steps - and Replit supported all milestones after the Discovery phase (this one made with multi-agents across ChatGPT, Claude, Gemini, Manus, DeepSeek, and Perplexity). I have a light background on development but never felt stronger enough to build something heavier than a contact from with post/get at a website.
So, I was really amazed by how Replit structure made possible to decide and take care of architecture and development decisions very easily, allowing not only non-trivial tasks such as database decisions, connections and configuration, but also going deeper to security checks and even integrations such as payment gates (I couldn't believe I was doing this by myself). It naturally presents things such as a "homologation" view before going to Production, with a friendly interface to guarantee the user will keep in mind some important aspects before just deploying something cool. And a lot of interesting or useful features are there to be considered or not (such as multi-agents to work at the same project, different artifacts related to the same project context, among others).
So, this is a small product experiment running live and checking one more step about how tooling is evolving and allowing professionals in general to use their repertoire potential more easily - that's how I see my design and development backgrounds being merged at these times, finally.
Please sign in to leave a comment.
No comments yet. Be the first to comment!