Send a message
Be specific about the source material, desired output, and constraints. For example, ask AI to summarize a document, produce a structured outline, revise a paragraph in an academic tone, or update a project file.Built-in examples
If you are unsure where to start, the input offers examples. They are prompts you can use as a starting point and then adapt to your work.Choose a model
Select an enabled model from the model selector at the top of chat. The selection affects messages sent afterward and does not interrupt a response already in progress. If the list is empty, enable a model in Model services.Research and tools
When a task needs more than text generation, Otterflow can show tool calls as grouped steps in the conversation. This makes the work visible rather than opaque: you can see actions such as reading files, searching, or changing project content.Quote content
Select text in the workspace and use Quick quote to attach it to the next message. A quote chip appears above the input so you can confirm what context will be sent. To provide a complete file instead of a selection, right-click it in the file manager and choose Add to current conversation. See File management.New conversations and history
Use New conversation to start fresh with an empty context. Use History to reopen an earlier thread and continue it.Queue versus guide
When AI is still replying, a follow-up message can use one of two modes:- Queue waits until the current response finishes. Queued messages can be edited or removed.
- Guide inserts your message at the next safe boundary to steer the ongoing generation.
Tool-call visibility, citations, and navigation
Tool calls show which actions AI takes. Citation cards let you inspect referenced material and jump back to its original file or location. This keeps the relationship between a response and its source context visible.Copy, retry, edit, stop, and continue
Use message actions to copy a result, retry a response, or edit a previous prompt when you need a better direction. You can stop a long response and continue the task later with an additional message.Memory and remaining quota
Memory lets AI retain useful preferences across interactions when the feature is available. Remaining quota is visible in the chat experience and account settings. For quota details and usage trends, see Account and billing.Tips
- State the target audience, language, tone, and file constraints in the request.
- Quote the exact passage you want AI to use when precision matters.
- Review grouped tool calls and changed files before treating a task as complete.