AI agent: Quick start

The ONLYOFFICE AI agent allows you to use AI features directly in ONLYOFFICE editors to speed up everyday tasks such as text generation, rewriting, formatting, translation, and content analysis.

AI tools are built into the editors, so you don't need to install any plugins. However, for security reasons, no AI model is connected by default. To start using AI features, connect a model of your choice as described below.

If the AI Tools tab is not displayed on the top toolbar, go to the Plugins tab, click Background Plugins, and make sure the AI Tools switch is turned on.

Quick setup

Step 1. Connect an AI provider

Go to the AI Tools tab and click Settings. The AI Settings window will open on the AI Models tab.
In the AI Models tab, click Add Model.
Select an AI provider from the Provider dropdown (for example, OpenAI, Anthropic, Google Gemini, or OpenRouter). Click Import Custom Provider to connect a provider that isn't listed (see Adding a custom AI provider).
Enter your API key. If you use a local provider such as LM Studio or Ollama, enter the Base URL instead.
Select a Model, enter a Profile name to identify the connection, and click Add Model.
In the Model Assignment tab, assign different models to specific tasks — AI Chat, Code, Summarization, Translation, Text Analysis, Image generation, OCR, and Vision. Any task set to Default model uses the model selected above.
In the MCP Servers tab (optional), click Edit configuration and enter your server configuration in JSON format under mcpServers, then click Save to make the server's tools available in AI chat.
In the Web search tab (optional), select a Web Search Engine (for example, Exa) and enter its API key to allow the AI agent to retrieve real-time information from the web.

You can change providers, assigned models, MCP servers, or web search settings at any time from AI Tools → Settings.

If the API key is invalid, an error message appears under the API key field, and the Add Model button remains unavailable. Check the key and enter it again.

Adding an AI model in the AI Settings window

Step 2. Use the AI agent

To open the AI agent input panel, use the keyboard shortcut for your operating system. Type your prompt and press Enter to receive a response. No additional configuration is required.

Keyboard shortcuts:

Open the AI agent panel	Windows and Linux	`Ctrl`+`/`
Open the AI agent panel	macOS	`Cmd`+`/`
Close the AI agent panel	All platforms	`Esc`

What you can do with the AI agent

Once configured, the AI agent can help you:

Generate, rewrite, and expand text.
Summarize documents and extract key information.
Translate content and improve grammar.
Analyze text for clarity, structure, and tone.
Automate simple file and content workflows using AI agents.

Troubleshooting

Expand all

Collapse all

The AI Tools tab does not appear: Make sure the AI Tools plugin is enabled in the Background plugins. If you have just installed or updated the plugin, restart the editor.

The AI agent opens but does not respond: Check that your API key is valid and that the selected AI provider is available. Network restrictions, firewall rules, or proxy settings may prevent the editor from connecting to the provider.

Local AI providers do not connect: If you are using a local provider such as LM Studio or Ollama, confirm that the local AI service is running, reachable at the configured address, and not blocked by firewall rules.

Responses are slow or incomplete: This may occur when the AI provider is under high load or when the prompt is too large. Try simplifying the request or breaking it into smaller steps. For local providers, verify CPU, GPU, and memory availability.

Unexpected results: AI responses depend on the selected model and the prompt. If the output does not meet expectations, try rephrasing the prompt or providing more context.