Point an AI assistant at your business and let it work.
DMly runs a Model Context Protocol server. Connect Claude, Cursor, n8n or your own agent, and it can search your contacts, read a conversation, find a free slot, book it and raise the invoice. You do not write the HTTP layer or describe your data to the model.
One address. The client works out the rest.
Model Context Protocol is the standard way an AI client talks to a system it does not already know. You give the client one address, it asks what the server can do, and the server answers with a list of tools it can call.
That is the part that used to be your job. No glue service, no hand-written schema in a prompt, no keeping an integration in step with your data model. The server describes itself and the assistant picks the tool.
DMly’s tools are task-shaped rather than one per endpoint, and the answers come back trimmed for a model’s context instead of as full API payloads. The assistant gets what it needs to decide, not a wall of JSON.
What your assistant can actually do
Thirty-seven tools across eight areas of the business, plus four reference resources it can read.
Workspace
An orientation snapshot: connected channels, whether each can send right now, open conversations and today’s bookings. Clients are told to call this first.
Contacts and CRM
Search by name, phone, handle, tag, segment or lifecycle stage. Pull one person’s full picture, add contacts, apply tags, write notes and move them through stages.
Conversations
Triage the inbox by open, unread or platform, and read a thread as a transcript. Assign a conversation to a teammate or mark it handled.
Messaging
Reply on whichever channel the customer last used, or send an approved WhatsApp template when the 24-hour window has closed.
Appointments
Find real open slots using the same availability engine as your booking page, so notice periods, buffers, time off and your connected calendar all apply. Then book, reschedule or cancel.
Finance
List invoices and what is owed, summarise revenue for a period, draft an invoice, send it on WhatsApp and record money already received.
Broadcasts
Work out exactly who a broadcast would reach and freeze that audience before anything is sent.
Automations
List flows and quick automations with their trigger, channel and run counts, so the assistant can tell you what is already running.
Two ways in, depending on the client
Sign in with OAuth
The client sends you to DMly, you sign in and pick a workspace. Nothing to mint and no key to paste, so this is the shorter path for hosted and desktop clients that offer an “add a connector” flow.
What it may do: exactly what that person’s team role allows in the dashboard. Narrow the role or remove them and it bites on the next request.
Paste an API key
Mint a workspace key carrying an mcp:read or mcp:write scope and paste it into the client. Built for config-file clients, scripts, headless setups and your own agent.
What it may do: what the scope allows, narrowed to the team role of whoever created it. Revoke the key to cut it off.
Neither is a fallback for the other. If your client has a connector flow, use OAuth. If it wants a URL and a header, use a key.
An assistant with your data needs real limits
The workspace comes from the credential
Never from a tool argument. No tool accepts a workspace id, so a model cannot be talked into reaching another business.
Bounded by your team roles
A connection can never do more than the person behind it could do in the dashboard. Read-only stays read-only.
Every write is logged
The audit log records the client that connected and one row per change, so you can see exactly what the assistant did and when.
MCP server, or the REST API?
Both are included on every plan. They are for different jobs.
Ask your assistant what happened today.
The MCP server is included on every DMly plan. Start a free trial and connect a client in a few minutes.
