MCP · Claude and AI agents

Let Claude work in your OrgLines workspace

OrgLines runs an MCP server, so Claude and other MCP-compatible AI agents can work with your CRM, mail, tasks, calendar, ERP, finance, logistics and HR as a set of tools. The agent works with your real data, inside your permissions, and every important change comes back to you for confirmation.

Who it is for

  • Teams already using Claude or other AI agents daily
  • Owners who want to run the business from one chat
  • Managers who combine OrgLines data with other tools in one AI conversation

The Model Context Protocol (MCP) is an open standard that lets an AI assistant use external tools and data sources in a controlled way. Instead of copying numbers from OrgLines into a chat, you connect the chat to OrgLines. Claude can then look up stale leads, list the week's tasks, check stock levels or prepare a freight order itself — using the same records your team sees.

Control stays with OrgLines, not with the agent. Access is granted through a consent screen, is limited to one workspace and can be revoked at any time in Settings. The server decides which actions need confirmation: when a write matters, the agent receives a proposal, not a result, and the change only happens after you approve it.

How to connect Claude

  1. 01

    Open the MCP section

    In your OrgLines settings, find the MCP section with the server address.

  2. 02

    Add the server to your AI client

    Add it as a remote MCP server in Claude or another client that supports remote MCP servers.

  3. 03

    Sign in and approve access

    OrgLines shows which client is requesting access and what access it asks for. The client then acts on your behalf in this workspace only.

  4. 04

    Ask in plain language

    "Show leads with no answer for three days", "plan my week", "what did we ship yesterday?"

  5. 05

    Confirm what matters

    Proposals for changes, deletions, external messages and financial actions wait for your approval.

  6. 06

    Revoke when you want

    Remove the client's access in Settings at any time.

What you can do

01

Broad set of tools

Personal workspace, leads, mail, companies and tasks, finance, logistics, HR, retail, website status, knowledge base and memory — all available to the agent.

02

Action classes

Every tool is classed as read, create, change, destructive, external or financial. OrgLines itself decides which of them need your confirmation; changes such as deleting data, sending outside the workspace or anything financial come back as a proposal first.

03

Confirmations on the server side

OrgLines itself decides which writes need your confirmation — the agent cannot skip it.

04

Your plan, your role, one workspace

The agent sees only the modules and data your plan and role allow, and only in the workspace you approved.

05

Transparent memory

What the agent remembers is stored as records you can review, correct and delete.

06

Linked records

The agent can link records across modules and find related ones, so a note, a task and a lead stay connected.

07

Digests and reminders

Ask for the morning brief or weekly review, or schedule a reminder that arrives in Telegram.

08

Revocable at any time

Access granted on the consent screen can be withdrawn in Settings whenever you want.

What an agent can do, by module

  • Personal: notes, tasks, calendar events and an ICS feed for your phone, reminders, habits, goals, time tracking, receipts and spending, subscriptions, warranties, documents with expiry dates, meeting notes
  • Leads: create and update leads, bulk updates, notes, timeline, stale leads and a leads summary
  • Mail: threads and messages, drafts, sending and scheduled sending, snooze, labels, blocking senders, clean-up
  • Companies and work: organizations, contacts, projects, tasks with checklists and comments, reports, employees, invoices, stock levels
  • Finance: overview, cash flow, break-even, forecast variance, scenarios, payroll summary, purchase requests, upcoming obligations
  • Logistics: freight orders, trips with crew, status and weights, carriers, drivers, fleet, dispatch summary
  • HR: people, applications and stages, vacancies, forms, internships, and reports such as funnel, sources, turnover and retention
  • Knowledge base, memory, your own functions and outbound webhooks

The safety model

Giving an AI agent access to a business system is only reasonable if the system, not the agent, sets the limits. In OrgLines the agent works with the rights of the user who connected it, in one workspace, and the server returns a proposal for every write that matters. You see the proposed change and confirm it, or nothing happens.

FAQ

Questions about MCP · Claude and AI agents

Didn't find an answer? Write to us and we will get back to you.

Ask a question
What is MCP?

The Model Context Protocol is an open standard that lets AI assistants such as Claude use external tools and data sources in a controlled way.

Which agents are supported?

Claude and any other client that supports remote MCP servers.

How do I connect it?

Open the MCP section in your OrgLines settings, add the server address to your AI client, sign in and approve access on the consent screen.

Can the agent delete data or send emails without asking?

No. The server decides which actions need confirmation, and changes such as deleting, sending outside the workspace or anything financial come back as a proposal you confirm first.

Does the agent see everything in my company?

No. It works in the workspace you approved and sees only what your plan and your role allow.

How do I stop an agent's access?

Revoke it in Settings. The client loses access to the workspace.

Is it the same as the built-in AI assistant?

It uses the same tools and the same confirmation rules. The built-in assistant runs inside OrgLines; MCP lets you use your own AI client instead.

Bring your business into one place

Create a free workspace in a minute and invite the AI assistant to your team.

Start free