MultipleChat AI Desktop docs.

Getting started

MultipleChat AI Desktop is a self-contained app for macOS and Windows. There is no server to set up and no account to create — the installer contains everything the app needs.

Install on macOS

1. Download the .dmg installer (Apple Silicon and Intel are both supported; macOS 12 or newer).
2. Open the .dmg and drag MultipleChat AI into your Applications folder.
3. Launch it from Applications. If macOS asks for confirmation the first time, choose Open.

Install on Windows

1. Download the .exe installer (64-bit Windows 10 or 11).
2. Run the installer and follow the prompts. If SmartScreen appears, choose More info → Run anyway.
3. Launch MultipleChat AI from the Start menu.

Using cloud providers, the app is very light. If you plan to run local models on your own hardware, 16 GB RAM or more is recommended.

Connect an AI provider

The app is BYOK — bring your own key. On first launch you'll be asked to pick an AI provider; you can change it, or mix several, at any time in Settings → AI Providers → LLM.

Cloud providers

1. Create an API key with your provider (for example OpenAI, Anthropic, Google Gemini, Mistral, Groq, OpenRouter, Perplexity, Cohere, or Azure OpenAI).
2. In Settings → AI Providers → LLM, select the provider, paste the key, and pick a default model.
3. Save. The key is stored only on your computer and is sent only to that provider when you chat.

Local models (fully offline)

Prefer no cloud at all? Install Ollama or LM Studio, download a model there, then select the matching provider in the LLM settings. Chats then never touch the internet.

Three levels of model settings

• System LLM — the default model used everywhere.
• Workspace LLM — overrides the default for one workspace (e.g. a coding workspace on one model, a writing workspace on another).
• Agent LLM — the model used for @agent runs; pick one that supports tool calling.

Workspaces

A workspace is a self-contained project: its own chat history, its own documents, and optionally its own model. Documents and conversations in one workspace are invisible to the others, which keeps contexts clean and answers focused.

1. Click New Workspace in the sidebar and give it a name.
2. Set a workspace-specific model under the workspace settings if you don't want the system default.
3. Fine-tune behavior per workspace: system prompt, chat mode (pure chat vs. document-grounded query), and how many past messages are kept in context.

Good pattern: one workspace per project or client — "Contracts", "Thesis", "Support macros" — instead of one giant chat.

Chat with your documents

Every workspace can hold documents. The app splits them into passages and indexes them with a local embedding model into a local vector database — nothing is uploaded anywhere. When you ask a question, the most relevant passages are retrieved and given to the model alongside your prompt.

1. Open a workspace and click the upload / documents icon.
2. Drag in files — PDF, Word, text, Markdown, CSV, code files and more — or point it at a folder.
3. Move the files into the workspace and click Save and Embed. Indexing runs locally on your machine.
4. Ask questions in chat. Answers are grounded in your files, and you can open the citations to see which passages were used.

Tips for better answers

• Prefer text-based PDFs over scans; run OCR first for scanned documents.
• Ask specific questions ("What does §7.2 say about termination?") rather than "summarize everything".
• Pin a document in the workspace if it must always be in context regardless of the question.

AI agents

Agents let the model do things instead of just answering. Start any message with @agent and the model plans the steps and picks the right tools automatically.

Built-in skills

• Web search & scraping — look up live information and read pages.
• Document search (RAG) — search across the workspace's embedded documents.
• Charts — generate charts from data in the conversation or your files.
• Files — write results to files and save them on your computer.
• Summarization — condense long documents on demand.

Enable, disable, and configure skills under Settings → Agent Skills. Set the agent's model in the same place — it should be a model with solid tool-calling (most current frontier models qualify).

Example: @agent find the three most recent EU AI Act summaries online, compare them with the draft PDF in this workspace, and save the differences as a file.

Where your data lives

Everything the app knows is stored in a local application folder on your computer: chat history, workspace settings, uploaded documents, the vector index, and your API keys.

• macOS: ~/Library/Application Support/ in the app's own folder.
• Windows: %APPDATA% in the app's own folder.

When you send a message, the request goes directly from your computer to the AI provider you configured, under that provider's privacy terms. There is no MultipleChat server in between — we never see, collect, store, or train on your chats or documents. Telemetry is disabled in our builds; the app sends no usage analytics to us or anyone else.

Deleting the app's data folder (or uninstalling and removing app data) removes everything permanently.

Updates & troubleshooting

Updates

We ship updated builds regularly, merging improvements and security updates from the upstream open-source project. Install a new version over the old one — your workspaces, chats and settings are kept.

Common fixes

• "Invalid API key" — re-paste the key without spaces and check that it's active (and funded, where the provider requires credit) in the provider's console.
• Answers ignore my documents — confirm the files are embedded in this workspace and the workspace is in query/document mode, then re-ask with a more specific question.
• Agent does nothing — make sure the Agent LLM is set to a tool-calling model in Settings → Agent Skills.
• App won't open after download (macOS) — right-click the app → Open once, to pass Gatekeeper's first-launch check.

Still stuck? Email [email protected] — include your OS and app version.