How do MSPs use Claude and ChatGPT with their PSA?

MSPs install a free MCP server connector for their PSA that runs locally on their own machine, then ask plain-English questions in Claude, ChatGPT, Codex, or Copilot. The AI runs the underlying command for them - triage the queue, build a client card, find SLA breaches, run cross-client analytics. You paste one sentence to install, enter your API credentials once, and after that you just ask questions. No code, no dashboards, no exports.

What it looks like day to day

You ask in plain English; your AI agent runs the command under the hood.

• “Triage my queue and tell me what’s breaching SLA today.”
• “Give me the whole picture for Acme Corp on one screen.”
• “Which clients had a backup that hasn’t succeeded in 7 days?”
• “How many SLA breaches per client last quarter?”

You never type the command. The connector keeps a local copy of your data, so cross-client questions come back instantly instead of as hundreds of slow API calls.

Which AI works with this

• Claude Desktop and Claude Code connect to the local connector directly - the most common MSP-owner starting points.
• ChatGPT (Plus and up) connects to remote MCP servers over HTTPS; you expose the local binary via a bridge.
• Codex CLI and GitHub Copilot (Agent mode) run the local connector too.

Full per-agent setup is on the which agent? → page.

Read this next

• Is this safe with client data? the Trust Center → · are MCP servers safe for client data? →
• New to the term? what is an MCP server? → · what is a Claude Code Skill? →
• See it on a real tool: ConnectWise PSA → · Autotask → · HaloPSA → · NinjaOne → · Datto RMM →