Pretty excited to launch next week and we already have a few users. We’re seeing a 60-80% token savings from our prompting engine compared to traditional vibe coding and the code / output quality is great. Mind you this isn’t a wrapper, the idea is this is a workspace you use with the LLM’s you already use, and you can use any LLM or a combination if you choose.
Here’s me blabbing on about the build phase / dynamic prompts in a demo video if you’re curious. https://supercut.ai/share/launchchair/yawrk-drkkqU-Vx686DBXm
We also have a full agent API / MCP so an agent can use the app end to end if you’re into that sort of thing.