The MCP Server Build
Your staff ask Claude, ChatGPT or Cursor about your own customers, orders or files, and the assistant looks up the answer in your live data through a server on your own host.
Your team already asks Claude or ChatGPT for help, but the assistant can't see your order system or your customer database. Someone copies rows out of one window into another, and the answer is only as fresh as the paste.
We build one MCP server, the standard way AI assistants connect to other systems, that gives those assistants up to eight named tools into a system you own: your database, an API your developers built, or a folder of documents. Your staff ask in plain words, such as "what's still open for this customer?", and the assistant looks it up live. Up to two tools can add or update a record, and none can delete.
The server runs on your own host, and you decide who gets in, with access keys you create or sign-in through Microsoft Entra ID, Okta, Auth0 or Keycloak. You type in your database and API passwords yourself, so we never hold them, and the server logs each call. We test each tool on your live data and in up to two of your AI assistants before we hand it over.
What's included
Up to eight tools into one database, in-house API or folder of documents, each described so the assistant knows when to use it and what comes back
Small changes you can trace: up to two tools may add or update one record at a time, only in the fields you approve, and no tool can delete
Access you control, with keys you create and revoke on the server or sign-in through your Microsoft Entra ID, Okta, Auth0 or Keycloak, and passwords you enter yourself
The server running on your own Docker host or Linux server at your HTTPS address, with a 90-day log of who called which tool
Each tool tested on your live data, a test session in up to two of Claude, ChatGPT, Cursor and Claude Code, one tuning pass on the tool descriptions and one round of changes
The source code, a snapshot you can roll back to, and a plain-English handover with set-up steps for each assistant and a record of where your data goes
Before you order
Have these ready so we can start the moment you order. The delivery clock only runs while we have them.
SSH access to a Docker host, or to a Linux server with Python 3.11 or later, for a user you will add our deploy key to
A web address for the server on an HTTPS reverse proxy you already run, reachable from the internet if you want Claude or ChatGPT to use it
The system it will reach (database details, your API address and its documentation, or the folder path) and a one-line description of each tool you want
For sign-in, someone who can create app registrations in your Microsoft Entra ID, Okta, Auth0 or Keycloak, and a Claude or ChatGPT plan that lets you add custom connectors
One person who will approve the spec, run the test in your AI assistant and send the feedback
How it works
Book it — pay the fixed price and tell us the target environment. No scoping call first.
We deliver — an experienced specialist runs our standard workflow for this exact task. You can watch progress and message them in the portal.
Hand-off — you get the deliverables, the documentation, a rollback snapshot, and a walkthrough if you want one.
Who does it
A salaried Support Resort specialist with more than ten years in the field, working from a runbook we've been refining since 2003 — supervised by senior management, never a marketplace contractor. You'll see who is assigned and can message them about this task directly in your portal.
FAQ
Ask about this task
Our assistant knows the scope document and runbook behind this service. Ask what's covered, what isn't, or whether it's the right task for your situation.
Answers come from our published scope documents. Prefer a person? Switch to the "Email us" tab — a senior manager replies personally, usually the same weekday.
Related services
The MCP or API Integration
One of your systems, or your AI assistants, connected to a platform you already pay for, such as Xero, HubSpot, Shopify, Stripe or Microsoft 365, so nobody types the same record twice.
The AI Developer Day
One full day of a senior AI developer working your list: LLM features shipped, prompts and retrieval measured against real cases instead of guessed at, and token bills brought back down.
The Reverse Proxy Setup
A hardened front door for up to three applications you already run — HTTPS ends at the proxy, scanner probes are turned away, and traffic is balanced across your backends while the apps behind it stay exactly as they are.
The Server Admin Day
A senior system administrator works a full day on your server backlog, from the patch level nobody has checked to the backup nobody has restore-tested. Book one day or a run of dates, and each day ends with a written account of every hour.