HubSpot Launches Deep Research Connector with ChatGPT

HubSpot announced it has become the first CRM to launch a Deep Research Connector with ChatGPT—and the first third-party MCP connector available in the ChatGPT Registry built with a remote Model Context Protocol (MCP) server. This connector extends ChatGPT’s capabilities while eliminating the need for users to install or configure anything locally.

According to HubSpot, the new remote MCP server changes everything. It represents a major leap forward from local servers by eliminating the need for users to install or run anything on their machines. Instead, they can access capabilities through their browser using trusted and reliable authentication flows such as OAuth . It democratizes access to AI for more than 250,000 businesses, while maintaining enterprise-grade security and respecting user-level permissions.

The decision to build in-house became clear once the company considered what hosting its own infrastructure could enable. By building on HubSpot’s platform, the company could leverage enterprise-grade CI/CD, auto-scaling, observability, and even region-aware API call routing built on Cloudflare Workers. It also gave the company direct control over how MCP interacts with existing scopes model for OAuth, allowing the company to ensure user-level permissions were respected throughout any interactions with its MCP tools.

HubSpot’s MCP server is built with the Java MCP SDK. The company chose Java for long-term maintainability and development within its standard technical stack. HubSpot’s backend libraries and infrastructure tooling are already centered around Java microservices.

For our initial release, HubSpot introduced a few adaptations to better align the SDK with its Dropwizard backend. These changes demonstrate the SDK’s flexibility, even in an enterprise setting.

Longer term, HubSpot intends to take its learnings on its MCP server and ChatGPT connector and enhance the entire developer platform to not only respect user permissioning but also mature the OAuth stack to the latest standards in OAuth 2.1 and beyond.

“Through thoughtful engineering decisions made at the rapid pace of AI development, we successfully implemented a complex protocol as a secure, customer-ready feature in a matter of weeks. This showcases how our team tackles challenging problems at the cutting edge of AI technology while delivering real customer value,” the company posted.

For more information about this news, visit www.hubspot.com.