
Initially I found floot very powerful and quick to work with. Like other uses have noted, the AI develops a worrying habit of building things that were not asked for, and spends a lot of unrecoverable time doing so, and uses a lot of tokens and time to unpick errors especially. (£600 last month) It's own errors and forgotten context eat your tokens. Whole sections of the app can become unusable. The absolute killer for any sort of commercial app is not AI related, its conceptually poor thinking... that the Sandbox and the Live app SHARE the database.... and the there is no staging environment - so like yesterday when the app created a buggy cleanup SQL statement run by a cron-job, the live app was affected by clients losing data....!!! So create a duplicate perhaps? NO! duplicate apps share the live and sandbox database... and so for every duplicate you create another potential headache. The UI AI said to the back end AI - Apparently I built this... after one error needed rectifying: It loses context in unpredictable ways: I would consider this for development only if you can redeploy the Code to Vercel and build a professional workflow from your Git and hook up to Xano or Postgres outside of Floot. It's not a finished thing. Be careful.
What's great
fast performance (5)no-code support (10)
What needs improvement
AI context issues (2)built-in tech stack (1)
Report
14 views
