What This Update Actually Is
HubSpot has launched a generally available, HubSpot-hosted Model Context Protocol (MCP) server. In plain English: it's a secure bridge that lets MCP-compatible AI tools talk directly to your HubSpot CRM.
This isn't a new chatbot or a Breeze feature. It's a protocol layer. Any AI tool that supports the MCP standard can now connect to HubSpot and interact with your data using natural language, while respecting every permission rule you already have in place.
The GA release ships both read and write access. Here's what's now in scope:
- CRM objects: contacts, companies, deals, tickets, carts, products, orders, line items, invoices, quotes, subscriptions, and now segments (lists)
- Engagements: calls, emails, meetings, notes, and tasks
- Write actions: create and update CRM records, log activities directly
- Organizational context: users, teams, reporting structures, owners, roles, and seats
- Marketing and content: campaigns, campaign metrics, landing pages, website pages, and blog posts
One important caveat: engagement objects are blocked when your HubSpot account has sensitive data mode turned on. If that setting is active in your portal, the conversation history and call data won't be accessible through the MCP server.
Why HubSpot Shipped This
The external problem is a real one. Humans who work inside AI tools every day keep having to stop, open HubSpot, look something up, copy it out, and paste it back into wherever they were working. That context-switching kills momentum and introduces errors.
The internal frustration is even more familiar. Your CRM holds the best intelligence your company has about your buyers. But most AI tools can only see what you paste into them. That gap means your AI is working with a fraction of the picture.
HubSpot's answer is to bring the data to the AI instead of the other way around. The MCP server closes that gap with a permission-aware connection that doesn't require exporting, syncing, or rebuilding any existing setup.
Notion, Writer, Perplexity, and Glean have already connected their products to HubSpot using this server. That's not a beta list. That's the GA ecosystem building in real time.
How to Use It Step by Step
If you're already connected from the beta, nothing changes. All new scope groups and tools are automatically available. No reconfiguration required.
If you're new to the HubSpot MCP server, here's how to get started:
- Create an MCP auth app. In your HubSpot account, go to Development > MCP Auth Apps and create a new app. HubSpot generates OAuth credentials automatically.
- Create a managed OAuth connection. Connect any MCP client that supports OAuth with PKCE to https://mcp.hubspot.com using the client ID and secret from your new app.
- Start prompting. Your AI tool can now query, create, and update HubSpot data using natural language.
A few example prompts to test right away:
- "Show me all deals closing this quarter over $50K, grouped by stage"
- "Create a follow-up task for the Acme Corp deal and assign it to me for Wednesday"
- "Summarize the engagement history with FitHub over the last 30 days"
- "Update contact Sarah Chen's lifecycle stage to Customer"
Note that OAuth 2.1 with PKCE is required for all connections. Not every MCP client supports this yet, so confirm your client's compatibility before you start.
What It Touches in Your HubSpot Strategy
This isn't a single-feature update. It's an infrastructure change that ripples across every hub you use.
On the CRM side, your contact, company, deal, and ticket records are now queryable and writable through any connected AI tool. That means a rep can update a lifecycle stage, log a call note, or create a follow-up task from inside their AI workspace without ever opening HubSpot.
Key Takeaway
Write access is the real unlock here. Read-only AI connections are useful for lookups. Read-write connections let insights become action without switching tools. That's the difference between AI as a dashboard and AI as a teammate.
On the sales side, engagement history (calls, emails, meetings, notes, tasks) is now in scope. That means your AI tool can synthesize a three-month relationship history and recommend a next step based on what was actually said, not just what fields were filled in.
If you're already using HubSpot's AI-powered sales tools, this compounds quickly. The Prospecting Agent's buying signal detection surfaces in-market accounts. The MCP server lets your external AI tools act on those signals immediately, in whatever workspace your reps already live in.
On the marketing and content side, campaign data, landing pages, website pages, and blog posts are all now accessible. A content strategist can ask their AI tool to pull campaign metrics and draft a performance summary without logging into HubSpot Reports.
Key Takeaway
Permissions travel with the connection. A user who can't see certain records in HubSpot can't see them through the MCP server either. Your data governance rules don't have a gap here. That's intentional and important.
For RevOps teams, this changes how you think about AI-assisted operations. We've been covering the pattern of HubSpot pushing AI deeper into the stack across multiple releases. This one connects the broader trend of HubSpot's AI-connected CRM to an open protocol that works outside HubSpot's walls.
Who Should Care Most
This update has a high ceiling and a low floor. Here's who gets the most value right now:
- Sales reps and managers who already use AI writing or research tools daily. The MCP server lets those tools pull live deal context instead of relying on what the rep remembers to paste in.
- RevOps teams who build custom AI workflows. If your team is building agents or automations on top of AI platforms, you now have a standards-based, permission-aware HubSpot integration without custom API work.
- Developers and technical admins at companies already using MCP-compatible tools like Notion AI, Glean, Writer, or Perplexity. Connecting takes three steps and doesn't require a developer to build an integration from scratch.
- Marketing ops humans managing campaign performance across multiple tools. Pulling campaign metrics through natural language saves the tab-switching that eats reporting time.
If your company is on the free plan and curious about AI-connected CRMs, this is a real entry point. There's no tier restriction on the MCP server itself.
George's Take
“The MCP server isn't an AI feature. It's a decision about where AI lives in your business. If your team's AI tools can't see your CRM, they're guessing. Now they don't have to.”
We've looked at a lot of portals where AI tools sit completely disconnected from the CRM. Reps paste in deal notes by hand. Marketers summarize campaign data manually. The intelligence lives in HubSpot, but the work happens somewhere else. That's the gap the MCP server closes. And it's not just a productivity win. It's a strategy shift. When AI tools can read engagement history, not just structured fields, they stop being lookup tools and start being reasoning tools. I've been saying for a while that the B2B buyer journey now requires your systems to work together in real time. This is HubSpot's clearest move yet toward making that actually possible for humans at every plan level.
If you're ready to connect your AI tools to HubSpot the right way, or if you want a second set of eyes on how this fits your current portal setup, book a strategy call with the Sidekick team. We'll help you map the connection, check your permissions structure, and make sure your AI tools are working with the full picture.
Frequently Asked Questions
What is the HubSpot MCP server?
The HubSpot MCP server is a secure, HubSpot-hosted implementation of the Model Context Protocol. It lets any MCP-compatible AI tool connect to your HubSpot CRM and interact with your data using natural language. It supports both read and write actions and respects your existing HubSpot user permissions.
Which HubSpot plans include the MCP server?
The MCP server is available to all HubSpot accounts on any plan, including free. You need access to HubSpot's Development tools and an MCP-compatible client that supports OAuth 2.1 with PKCE to get started.
What can the HubSpot MCP server read and write?
It can read and write CRM objects (contacts, companies, deals, tickets, and more), engagements (calls, emails, meetings, notes, tasks), segments, campaign data, landing pages, website pages, blog posts, and organizational data like users and teams. All actions respect your existing HubSpot permissions.
Is the HubSpot MCP server secure?
Yes. The MCP server uses OAuth 2.1 with PKCE for all connections. Every action respects your HubSpot user permissions, so users can only access or edit records they already have permission to see. If your account has sensitive data mode enabled, engagement objects are blocked from MCP access.
How do I connect an AI tool to HubSpot using the MCP server?
There are three steps. First, create an MCP auth app in your HubSpot account under Development > MCP Auth Apps. Second, connect your MCP-compatible client to https://mcp.hubspot.com using the generated OAuth credentials. Third, start prompting your AI tool in plain English to query or update your CRM.
Which AI tools work with the HubSpot MCP server?
Any AI tool that supports the Model Context Protocol with OAuth 2.1 and PKCE can connect. Notion, Writer, Perplexity, and Glean have already integrated with HubSpot using the MCP server. Custom agents and internal AI tools built on MCP-compatible frameworks can also connect.
