CELF AI

An intelligent conversation system that combines recent context with selective memory.

• Cumulative project memory — every time you come back with a question, the system "remembers" your project's context without you having to re-explain it from scratch.
• Client-owned local data — the memory lives in your own browser (IndexedDB), not on our servers, giving you higher privacy and full ownership of your data.
• Selective retrieval, not full history — it pulls only the actually relevant information instead of stuffing every request with the entire history, for faster and more focused responses.
• Handles long texts and code in one place, useful for writing documentation and code snippets together.
• No buildup inside the context window — no need to resend the full history every time, so token count doesn't keep growing, giving a good rate of savings.
• The picture of your project stays complete and accurate within the same session no matter how long it runs — the longer the session goes, the more project knowledge the system accumulates, without the number of tokens sent growing at the same rate.
• Local data belongs to this device and browser and may be lost if browser data is cleared.