AI and agents
Ask, and Layers answers from your tasks
The Layers assistant finds the tasks in your workspace on its own and answers with links to them. Claude, Cursor and other agents connect over MCP and work with tasks, pages and forms.
Layers for teams: the first 5 seats are free, then $4 per seat a month
Buy seats · $4/seatThis text was prepared with the help of AI.
No card required · free for up to 5 people

Documents in ChatGPT and Claude
Connect Layers Documents in one minute
Use one remote MCP address. OAuth opens in Layers, and the agent gets only the permissions of the account that signs in.
- 1
Add the connector
Open connectors in ChatGPT or Claude and paste the remote MCP address. Choose OAuth when the client asks how to sign in.Remote MCP addresshttps://app.layers.md/mcp?profile=documents - 2
Sign in to Layers
Approve the connection in Layers. Passwords and access tokens stay out of the conversation. - 3
Ask for the finished document
Create from text or a template, fill known fields, open the result for editing, or send the exact document for signature.
Try it in your own words
Three document jobs, one conversation
The agent returns both the working document and a Layers action link for the next person.
“Make a lease agreement”
You: Make a lease agreement for this apartment.
Agent: I’ll ask for the missing parties, dates, address and rent, fill the template, then return the edit link and the link you can send.
“Compute final pay”
You: Compute final pay from these figures and make the document.
Agent: I’ll use only the figures you provide, create the filled document in your Drive and return it ready to review.
“Send for signature”
You: Send this document to the named recipient for signature.
Agent: After you confirm the exact person, I’ll send once and track Opened, Agreed and Signed for you.
The problem
How AI works for most teams
AI on the side
- ✗The chatbot doesn’t know your tasks or docs
- ✗Answers get copied into the tracker by hand
- ✗An agent gets full access — or none at all
- ✗No one knows who spends how much on AI
01 · Ask
An assistant that knows your tasks
An AI chat built into Layers. It searches your workspace tasks on its own and answers with links to them.
Answers
What’s on me today — in one question
The assistant finds your tasks for today and shows the priority and status of each. Every task name links to the task.

Suggestions
It suggests where to start
A new chat opens with four suggestions: What is on me today, Overdue tasks, Sprint summary and Who is doing what.
02 · Connect
The agent works in your workspace
The Layers MCP server gives an agent tasks, pages, forms and projects. What the agent does shows up right away — as if you did it yourself.
Tasks
From plan to tasks, no manual entry
The agent breaks a plan into tasks and sets assignees and due dates. The tasks land in the list with their own keys, just like ones created by hand.
Pages
Notes from the agent go straight to Pages
The agent creates a page in the right folder and adds sections with headings and lists. “Sprint 14 retro notes” sits under “Meeting notes”.
Forms
A form from the agent, questions included
The agent builds a form: a start screen and four questions as steps.

Comments
The agent comments on a task — as you
The agent’s comment appears in the task’s Activity like your own: the author is the token owner.

03 · Grant access
An agent gets exactly the access you give it
The agent signs in with a token. You set its access and expiry, and revoke it in one click.
Access
Access and expiry, set when you create the token
With Read and write access, the token creates and changes tasks, pages and comments — everything you can do yourself. Expiration: 7, 30 or 90 days, 1 year, or never.

Expiry and revoking
See who has what access
Every token has its own access, expiry date and status. To revoke one, use the button in its row.
04 · Extend
Your own MCP servers, in the assistant
Connect a wiki, a CRM or your own service over MCP, and the Layers assistant will work with it.
A connector is a name and an address
A name, the server address — and a token if the server requires authorization. Settings from another client can be pasted into the JSON tab.
05 · Control
The owner decides what’s allowed
AI features and everyone’s monthly limit live in the workspace settings.
Turn AI features on or off for everyone
Assistant, editor, meetings and podcasts — each feature has its own switch. A feature switched off disappears for everyone in the workspace.
A limit per person — or AI off
One monthly limit applies to everyone without a limit of their own. A person’s menu has Set a limit, Same as everyone and Turn AI off.
Cloud or your own server
Free for up to 5 people. Self-host whenever you want
In the cloud — no card, no time limit. On your own server you get the same MCP server, and your admin connects the model for the assistant.
One workspace
AI is just the start
Your data, under control
Do I need an AI key to connect Claude or Cursor?
No. The Layers MCP server works on its own: the agent brings its own model, and Layers gives it tools and data under your token.
How do I connect an agent?
Create a token in Settings → Access Tokens, then add the Layers MCP server address and that token to your MCP client.
What can an agent do?
More than 50 tools: tasks (including in bulk), pages, forms, projects, sprints, folders, tags, comments, search and trash. With a read-only token it can only look.
Can an agent see other people’s documents?
No. The agent works with the token owner’s permissions and sees exactly what they see.
Which models can the assistant use?
On your own server, the admin connects OpenAI, OpenRouter, Gonka or any OpenAI-compatible server — including a local model via Ollama.
How do I cap spending?
A monthly limit per person — shared or individual. AI can be turned off for a specific person, or a feature turned off for the whole workspace.
Your agents work right inside Layers
Start free in the cloud or deploy Layers on your own server.

