Is the engineering team at Replit using... Replit?

In the spirit of dogfooding, I was wondering if Replit’s workspace is built on… Replit?

We’re seeing so many promises by all AI platforms (i.e. “It’s like having an entire team of software engineers on demand, ready to build what you need” Agent - Replit ), I’m just wondering if anyone is actually integrating such tooling into their workflows for (moderately) complex projects?

In particular, any luck integrating AI coding into existing projects, or is the agent coding approach better/only suited to new projects, where all architecture decisions are left to the AI?

Here is an example from We eat our own dog food at Replit. Here's how we actually use our own product every day: Marketing builds replit.com Our non-technical marketers update pages, optimize SEO keywords and backlinks… | Andrew Verrilli

Engineering builds Replit Agent… with Replit Agent Our eng team uses Replit Agent to code Replit Agent. They’ve built custom trace tools that analyze telemetry, debug prompts, and track sub-agent behavior. It’s turtles all the way down.

You can also follow the engineering team on LinkedIn/X etc. and many of them also talk about this.

I haven’t seen it directly, however, from what I see - they are doing this in some aspects.