Smartsheet announced on June 11, 2026 that its enterprise customers can now plug into a new Model Context Protocol (MCP) server, giving AI assistants like Microsoft Copilot, ChatGPT, and Google Gemini Enterprise direct access to work data stored in the platform. The integration, unveiled from the company’s Bellevue headquarters, means project managers and team members can interrogate their sheets, reports, and dashboards using natural language inside the AI tools they already use—no custom coding required.

Smartsheet's MCP Server: A Direct Line to Your Project Data

The core of the announcement is a server that implements the Model Context Protocol, an open standard originally developed by Anthropic that has rapidly gained industry support. Think of MCP as a universal USB-C port for AI: it defines how AI models discover and interact with external data sources and tools. With Smartsheet's MCP server, an enterprise AI assistant can securely fetch live information from a company’s work management environment—budget figures, task progress, resource allocation—and use it to answer questions or trigger actions.

Until now, integrating Smartsheet with an AI assistant required building a custom plugin or relying on third-party middleware. The MCP server is a ready-to-use endpoint that Smartsheet hosts and maintains. According to the company, customers simply activate it from their admin console, copy the provided URL, and paste it into the supported AI platform. Once connected, the assistant understands the schema of the user’s Smartsheet workspaces—sheets, columns, reports, dashboards—and can perform read-only queries. Write operations, such as updating a cell or adding a row, are planned for a later release.

The supported assistants cover the enterprise AI trifecta:
- Microsoft Copilot (via Microsoft 365 Copilot extensibility)
- ChatGPT Enterprise (from OpenAI)
- Google Cloud Gemini Enterprise (via Vertex AI agents or Gemini app)

Smartsheet emphasized that all data flows are governed by existing enterprise permissions, meaning a user can only query what they are allowed to see in Smartsheet. The MCP connection uses OAuth 2.0 authentication and TLS encryption, and no Smartsheet data is stored by the AI model or used for training.

What This Means for Different Roles

For the Project Manager or Knowledge Worker

If your organization uses Smartsheet to track projects, and you spend your day inside Microsoft Teams or Outlook, the Copilot integration is a game-changer. Instead of opening Smartsheet and hunting for a row, you can ask Copilot directly: “What’s the status of the Q3 marketing campaign deliverables due this Friday?” Copilot will reach into Smartsheet via MCP, find the relevant rows, and present a summary—complete with assignee names and due dates. Similarly, a finance analyst using ChatGPT Enterprise can ask, “Show me the total budget spent on the Phoenix office renovation across all sheets,” without manually exporting data.

Because the MCP server understands the structure of your Smartsheet account, questions don’t require you to remember precise column names. You can use fuzzy language, and the AI will map it to actual fields. This drastically reduces the friction of extracting insights from structured work data. A field technician could ask Gemini on their phone, “What urgent maintenance tasks are still open for Site B?” and get a real-time answer pulled from a shared Smartsheet.

For IT Administrators and Security Teams

The setup process is straightforward but requires attention to access controls. Smartsheet enterprise admins must first enable the MCP server from the Admin Center under “Integrations.” There, they generate a unique endpoint URL and an associated API key. The URL is then registered with the AI platform:

  • In Microsoft 365 Copilot, admins use the Copilot Studio to create a connector that points to the MCP server URL. Microsoft’s documentation provides a wizard for configuring permissions and authentication.
  • For ChatGPT Enterprise, an admin adds the URL in the “Workspace Settings” under “Data Sources,” entering the OAuth credentials provided by Smartsheet.
  • Google Cloud Gemini Enterprise integrates via Vertex AI Agent Builder, where the MCP server is added as a tool.

Crucially, the Smartsheet MCP server enforces object-level permissions. If a restricted sheet is not shared with a particular user, their AI assistant will not return data from that sheet, even if they ask. This is a significant security advantage over earlier, less granular integrations.

Smartsheet also provides audit logs for every MCP query—who asked what, when, and which assistant was used. This helps compliance teams keep track of AI interactions.

For Developers and Platform Builders

While the out-of-the-box integration targets business users, developers can extend the MCP server. Because MCP is an open protocol, a developer could write a custom client that interacts with the server using the standard JSON-RPC messages over stdio or HTTP. This means a company could build its own internal chatbot that sits on top of Smartsheet data, using the same MCP endpoint. Smartsheet has published the full API specification for the MCP server, encouraging a rich ecosystem of tools.

Moreover, the MCP server currently exposes a set of “tools” in MCP parlance: list_sheets, get_sheet_data, search_sheets, and summarize_report. Developers can call these tools programmatically without dealing with the SQL-like Smartsheet API of the past. This abstraction saves time and reduces complexity when building automation.

How We Got Here: The MCP Revolution and Smartsheet’s AI Journey

The Model Context Protocol emerged in late 2024 as Anthropic’s answer to the fragmentation of AI-to-tool integrations. It was quickly backed by major players: OpenAI announced support in early 2025, followed by Microsoft and Google later that year. The idea was simple: rather than every developer building one-off connections, a standard protocol would let any AI model talk to any data source if both sides spoke MCP. This mirrors the role REST played for web APIs.

Smartsheet saw early potential. The company had already dabbled in AI with its own dashboard assistant and formula builder, but these were siloed within the Smartsheet interface. The MCP server is a strategic pivot: instead of pulling users into Smartsheet’s own assistant, it meets them where they are—inside the enterprise copilots that are becoming ubiquitous.

In a press release, Smartsheet’s VP of Product Engineering noted, “Our customers live in tools like Copilot, ChatGPT, and Gemini. Giving them a secure, zero-friction way to access their work data from those interfaces is the natural next step for enterprise work management.”

The timing aligns with a broader trend: by mid-2026, almost every major SaaS platform had released an MCP server for enterprise use. However, Smartsheet’s focus on structured work data—project plans, budgets, resource sheets—makes this integration particularly high-value, because that data is often scattered across dozens of sheets and is cumbersome to query using traditional search.

What to Do Now: A Quick-Start Guide

If you’re a Smartsheet enterprise customer, here’s how to get started today:

  1. Verify your Smartsheet plan: The MCP server is available to Enterprise and Premier plans. If you’re on a lower tier, you’ll need to upgrade. Confirm with your Smartsheet account manager.
  2. Enable the MCP server: Log into the Smartsheet Admin Center, navigate to Settings > Integrations > MCP Server, and toggle it on. The system will generate a unique URL (e.g., https://mcp.smartsheet.com/enterprise/your-instance). Save this URL, along with the client ID and secret shown.
  3. Choose your AI assistant and configure:
    - Microsoft Copilot: In the Microsoft 365 admin center, open Copilot Studio. Create a new conversational agent, and add a “Data Source” from the menu. Choose “Custom MCP Server” and paste the Smartsheet URL. Follow the prompts to authenticate using OAuth. Test with a sample query like “Show me open tasks in the Alpha Project sheet.”
    - ChatGPT Enterprise: As an admin, go to chat.openai.com/manage, select your workspace, and click on “Connections.” Choose “MCP Server” from the available connectors, enter the URL and credentials, and save. Users in the workspace can immediately start asking questions prefixed with “using Smartsheet…”.
    - Google Gemini Enterprise: In the Google Cloud console, navigate to Vertex AI Agent Builder. Create a new agent, and under “Tools,” add a new tool of type “MCP Server.” Input the URL and authentication details. Publish the agent and assign it to users. They can then invoke it within Gemini by name, e.g., “@SmartsheetAgent show me the latest report.”
  4. Train your team: Communicate the new capability and provide a few example prompts to get them started. Emphasize that they must still follow data governance rules—what they ask in Copilot is logged and auditable.
  5. Monitor usage: Use Smartsheet’s audit logs (in the Admin Center under “Security” or via the Smartsheet API) to track which queries are being made. Set up alerts for unusual activity, such as a spike in queries from a single user that might indicate data scraping.
  6. Consider security policies: If your organization restricts the use of certain AI assistants, you may want to allow connection only to approved platforms. Smartsheet lets you whitelist which AI platforms can connect to your MCP server, blocking any others.

Outlook: The Era of Interoperable Work Data

Smartsheet’s move is a bellwether. As MCP adoption accelerates, the boundary between different enterprise applications will blur. In the near future, you might ask your AI assistant to “prepare a status report by pulling data from Smartsheet, Jira, and Salesforce” and get a consolidated answer without ever opening a browser tab. For Smartsheet, the MCP server positions the company not just as a destination for work management, but as a data layer that feeds the entire enterprise AI ecosystem.

We can expect Smartsheet to expand the server’s capabilities: write support, as mentioned, plus deeper integration with Smartsheet features like Bridge workflows and Resource Management. There’s also talk of making the MCP server available to smaller business plans, though no timeline has been announced.

For now, enterprise users should test the integration on a pilot project and gather feedback. The ability to talk to your project data naturally, inside the AI tools you already use, is not a gimmick—it’s a productivity lever that could change how teams interact with structured information.