What Chat is
Chat is the interactive interface with the platform's AI models. It behaves like a familiar assistant chat, but every conversation runs inside your IntraLLM environment, on models your administrator approved, with optional access to your documents, knowledge bases, tools and the web (when enabled).
Use chat for writing, analysis, coding, translation, planning, summarisation, and question-and-answer over enterprise content.
The layout
| Element | What it does |
|---|---|
| Module sidebar | New Chat, Search, Settings, and your recent conversations grouped as Today, Previous 7 days, Previous 30 days. |
| Model selector | Choose which model answers this conversation. Only approved models appear. |
| Message input | Type or dictate your message, attach files, and use the input menu for extra actions (such as camera capture). |
| Chat controls | Per-conversation options: capabilities (Vision, Web Search, Code Interpreter, Image Generation, Video Generation, Citations, File Upload), model parameters (Valves), knowledge, and tools — depending on what your administrator enables. |
| Response actions | Hover a response for copy, regenerate, branch/edit, like/dislike and read-aloud controls. |
Start a conversation
- Select New Chat in the module sidebar (or just send a message from the landing input).
- Optionally choose a model.
- Ask your question and send it. The response streams in progressively.
- Titles are generated automatically so the conversation is easy to find later.
Note:
Chat history visibility (Today / Previous 7 days / Previous 30 days) can be controlled from My Settings → Interface, and administrators can set defaults for the workspace.
Go beyond plain questions
- Attach files — ask questions grounded in PDFs, spreadsheets, images, audio or video. See Chat Basics.
- Use a knowledge base — answer from a curated collection of enterprise documents. See Knowledge Bases & RAG.
- Use capabilities — enable web search, image analysis, code execution or media generation per conversation. See Chat Controls & Options.
- Use a template — start from an assistant that already knows a job (email triage, report drafting…). See Using Templates.
- Compare models — ask the same question of several models at once. See Model Comparison.
- Reuse prompts — insert saved prompts from your library. See Prompt Library.
Managing conversations
Rename, search, clone, share, archive or delete chats from the conversation menu in the sidebar. See Managing Conversations.
Notes and limits
- Available models, capabilities and tools depend on your administrator's configuration and your group permissions.
- Very large attachments or long conversations can hit token limits; the platform warns you when a message exceeds the current limit so you can shorten it or start a new conversation.
- Generated content can be wrong. Review important outputs — especially anything that feeds business decisions. See FAQ for expectations.