Introduction to the AI Assistant Last updated: 17. Oct 2026
Docly has a built-in AI assistant that helps you write and edit content — one mode focused on a single document, one that works across your whole workspace.
What the AI Assistant can do
The assistant reads your documents and proposes concrete edits — you stay in control and approve every change before it is saved. Typical tasks:
Rewrite, translate, shorten or expand text in a document.
Fill in fields, fix formatting, or restructure content.
Create, edit, rename, move or delete files across a project.
Answer questions about your content and run predefined reports.
Create and edit images — once you have picked an image model (images are off by default).
Read public web pages you point it to.
Workspace Agent: look up what is published, read the site console, the system log, shares and version history, and propose to publish or unpublish.
The two modes
Document AI Assistant — an inline panel inside the document editor. It edits one document at a time and is opened from the AI AGENT ribbon button while editing. Best for quick edits to the document you are looking at.
Workspace Agent — a full-page agent that works across many files and folders. It can create, edit, rename, move, copy and delete files, and shows all proposed changes for approval. Best for larger tasks that touch several files.
Which one should I use?
If you want to… | Use |
|---|---|
Edit the text or fields of the document you have open | Document AI Assistant |
Run a predefined report on a document | Document AI Assistant |
Change several files, or create/rename/move files | Workspace Agent |
Do a multi-step task ("refactor", "set up a new section") | Workspace Agent |
Tip: You can jump straight from the Document AI Assistant to the full Workspace Agent — click the 🚀 Workspace Agent button in the inline panel when a task grows beyond a single document.
Choosing an AI model
Both modes let you pick which AI model to use from the Select AI Model selector. Different models trade off speed, cost and capability — pick a lighter model for simple edits and a stronger model for complex reasoning. If unsure, the default is a good starting point.
Images use a separate image model, picked in the same panel. Images stay off until you choose one, and the choice is saved for your user. Image costs count toward the same monthly limit.
Billing & terms of use
The assistant requires developer access to the folder you work in. The first time you open it you must accept the terms. Key points:
You review every change — you are responsible for reviewing and approving all AI changes. Nothing is saved until you accept it.
Usage is billed — AI usage is billed to your subscription. The default spending limit is $100 USD per month, resetting on the 1st. Contact support to raise it.
No guarantees — AI-generated content is provided as-is, with no warranty of accuracy or completeness.
Both modes show the cost of the current session and your usage for the month, so there are no surprises.
Using your own AI tools
To work on Docly from your own AI tools instead — Claude Code, Cursor, ChatGPT and others — use the docly CLI, or connect the tool to the MCP endpoint (/mcp) on the Docly cloud portal. Both require developer access. See How an AI agent should use the docly CLI.