A private AI assistant for the work you don't want indexed
A private AI assistant handles everyday knowledge work — drafting, summarising, reviewing, explaining — inside a workspace where only your account can read the conversation history. ChatOTL adds multi-model switching, branching and file uploads on top of that boundary.
An assistant becomes useful once you stop censoring what you tell it. That only happens when you trust where the conversation lands.
ChatOTL is built for that daily-driver role: fast streaming replies, a model picker you can change mid-thread, and file attachments — all inside a workspace scoped to one account.
What the assistant handles well
- Turning rough notes into a clear draft, then tightening tone and length.
- Summarising a long document you attach and answering follow-up questions about it.
- Explaining unfamiliar code, error messages or contracts in plain language.
- Working through a decision out loud, with branches for each option you want to test.
Switch models without restarting the thread
Different models are good at different things. Draft with one, review with another, and keep the same context — the model picker applies to the next turn, not the whole conversation.
Branch instead of scrolling
When an answer is close but heading the wrong way, fork it. The original thread stays intact and the branch carries the context forward, so you can compare two directions side by side later.
Example prompts
- "Rewrite this update for a non-technical exec, keep it under 150 words."
- "Here's a 40-page PDF — what are the obligations on the supplier side?"
- "Explain why this migration deadlocks and give me two safer orderings."
Limitations
- No live web access — the assistant reasons over what you give it and what the model already knows.
- It does not act on external systems for you; agent clients connect through MCP with your explicit consent.
Frequently asked questions
How is this different from a general AI chatbot?
The interaction is similar. The difference is ownership: your conversations, files and keys are scoped to your account, and you choose which provider processes each message.
Can I use my own API keys?
Yes. Keys are stored per account, and you can point at any OpenAI-compatible endpoint.
Can other tools use my assistant?
Yes — ChatOTL exposes an OAuth-protected MCP server, so Claude, Cursor or ChatGPT can read and search your threads as you after you approve consent.
Does it remember previous conversations?
Each thread keeps its own history, and you can search across your messages. There is no hidden cross-thread profile of you.
Put the assistant to work
Bring a document, a draft or a bug and see how it handles a real task.
Open ChatOTL