When we launched the Gradient MCP Server, it connected your Gradient data to Claude Desktop, Claude Code, and Microsoft Copilot. Today we are adding ChatGPT to that list.
The Gradient MCP Server is the bridge between your Gradient account and the AI agents you are already using. It uses the Model Context Protocol — an open standard for connecting AI agents to external data sources — to give your AI of choice direct, secure, read-only access to your Reconcile and Microsoft Module data.
With ChatGPT now supported alongside Claude Desktop, Claude Code, and Microsoft Copilot, you can ask questions about your billing, reconciliation, and vendor data in plain language, inside whichever AI you already work in, without exporting a spreadsheet or copying data manually.
The MCP Server is read-only in Version 1, which means your AI agent can pull and reason over your Gradient data but cannot make changes to Gradient or your PSA. That scope is intentional and will expand in future releases.
With the connection active, you can ask your AI things like:
"Walk me through this month's reconciliation and flag anything that looks unusual."
"Which clients have the thinnest margins this month?"
"This vendor just raised prices 5 percent. What does that do to my overall margins?"
"Show me assigned versus unassigned Microsoft license counts by client."
"Help me prepare talking points for an upcoming QBR with this client."
The agent requests the relevant data through the MCP Server, Gradient returns it scoped to your organization, and you get an answer in plain language rather than a dashboard you have to interpret yourself.
The Gradient MCP Server currently supports four agents:
Support for additional platforms is on the roadmap. This article will be updated as new agents are added.
Each MCP key is tied to your organization. An agent connected with your key can only ever see your own Gradient data. It cannot reach any other organization's information. Gradient runs the data logic on its own servers and returns a clean, validated result, which means your data is not passed directly to the AI model in an uncontrolled way.
Treat your MCP key like a password. If you believe a key has been exposed, delete it and create a new one. Access is revoked immediately.
You can create an MCP key from Settings > Data Settings > API Keys inside your Gradient account. Give the key a recognizable name, copy the snippet from the Key created window, and follow the connection instructions for your preferred agent.
Full setup documentation is available at support.meetgradient.com/mcp-server.
Version 1 focuses on reading and reasoning over your data. Future releases will add the ability to update mappings and make configuration changes directly from your agent, as well as support for additional AI platforms.
If you are already using the MCP Server with Claude or Microsoft Copilot, adding ChatGPT support takes minutes. If you have not set it up yet, now is a good time to start.