What This Update Actually Is
The HubSpot connector for ChatGPT already existed. It let you pull up individual HubSpot records inside a ChatGPT conversation. That was useful, but limited.
This update is a meaningful expansion. HubSpot added seven new capability areas to the connector, aligning it with what's available on their MCP server. Here's exactly what's new:
- Campaign management and attribution: create and update campaigns, manage assets, and pull deal and revenue attribution reports.
- Landing page creation and content analytics: build or update landing pages and access performance data including views, submissions, and conversion rates for pages and blog posts.
- Leads record access: create and update lead records using engagement data, track lead pipelines, and analyze lead-to-deal conversion rates.
- Custom and activatable object access: view, create, and update custom objects plus HubSpot's activatable objects like appointments, projects, listings, services, and courses.
- Flexible CRM queries: search and analyze records across CRM objects, properties, and associations using plain-language requests.
- Schema discovery: let ChatGPT explore your specific HubSpot account's objects, properties, associations, and field definitions before it acts.
- Connector feedback: submit product feedback to HubSpot directly from inside a ChatGPT conversation without switching tabs.
One important guardrail: the connector mirrors your existing HubSpot permissions exactly. ChatGPT can only touch what your HubSpot account already allows you to touch. Super admins control which permissions are enabled and can revoke access at any time if they're opted into HubSpot's governance beta.
Why HubSpot Shipped This
The gap this update closes is a familiar one. Humans who work in HubSpot every day already know the data is there. The frustration is the distance between having a question and getting an answer.
Building a report takes time. Finding the right property name takes time. Switching between HubSpot, a spreadsheet, and a ChatGPT tab to connect the dots takes time. That context-switching adds up, and it slows down decisions that should be fast.
HubSpot's answer is to make ChatGPT an actual participant in your workflow, not just a drafting assistant that sits beside it. With schema discovery, ChatGPT can learn your specific data model first, then answer questions that reference your real property names and object configurations. That's a meaningful shift from generic AI help to account-aware AI action.
How to Use It Step by Step
Before anyone on your team can use the expanded connector, a super admin (or a user with HubSpot Marketplace permissions) needs to set it up. Here's the sequence:
- Have a super admin install the HubSpot connector from the ChatGPT plugin or connector marketplace.
- Review and approve the permissions the connector requests. Approve only what your governance policy allows.
- Grant access to the specific team members who need it. Don't do a blanket rollout until you've tested permission boundaries.
- If you're a ChatGPT workspace admin, check whether you need to enable the updated connector version at the workspace level.
- Each user connects with their own HubSpot credentials. Walk them through this step so the permission mirror is set correctly for each individual.
- If your team was already using the connector, check whether you need to upgrade or reconnect it. The new write capabilities (campaigns, landing pages, leads) won't appear in older connector versions.
- Start with schema discovery. Ask ChatGPT to describe your HubSpot account's objects and properties. This confirms the connector is reading your specific configuration correctly before you take any write actions.
- Test a read-only query first. Ask for a campaign performance summary or lead-to-deal conversion rate. Confirm the data matches what you see inside HubSpot before trusting write actions.
What It Touches in Your HubSpot Strategy
This update isn't isolated to one hub. It ripples across Marketing Hub, the Smart CRM, and any custom data model your team has built. Here's where to think carefully.
Marketing Hub: Campaigns and Content
Campaign management and attribution access means marketers can now ask ChatGPT to pull deal and revenue attribution data inside the same conversation where they're making campaign decisions. If your team has ever struggled to connect campaign spend to closed revenue, this is worth exploring. We've written about how to build revenue attribution your sales team actually believes, and having that data accessible through a natural-language query changes how fast those conversations can happen.
Landing page creation through ChatGPT is a write action, which means it carries real risk if permissions aren't set correctly. Test this in a sandbox or with a single trusted user before opening it to your broader marketing team.
Key Takeaway
Write actions like creating campaigns or landing pages require eligible subscription tiers and correctly scoped permissions. Audit your user permissions before enabling these features team-wide.
Smart CRM: Custom Objects and Leads
Custom and activatable object access is the most technically significant part of this update for ops teams. If your portal uses custom objects to model something unique to your business (service agreements, equipment records, membership tiers), ChatGPT can now query and write to those objects. That's powerful, but it requires your custom object schema to be clean and well-documented first.
Sloppy property names and inconsistent object configurations will surface quickly when AI starts working with your data model. If your portal has accumulated years of custom properties without naming conventions, now is the time to clean that up.
The natural-language CRM query capability pairs well with other AI tools HubSpot is shipping. If you're also using Breeze Assistant for task work, you're building a workflow where AI assists at multiple touchpoints rather than just one.
Key Takeaway
Schema discovery is the connector's most underrated feature. Have ChatGPT describe your account's data model before taking any write actions. It surfaces misconfigured properties and gaps you might not have known were there.
Governance and Permissions
HubSpot's governance beta gives super admins the ability to revoke connector access at any time. If your organization has data handling policies around third-party AI tools, the fact that this connector runs through OpenAI means you'll want legal and IT aligned before a broad rollout. HubSpot's own data handling documentation covers how data flows when the connector is active.
Who Should Care Most
Not every HubSpot customer needs to act on this today. Here's who should move first.
- Marketing ops leaders who run campaign reporting manually. The campaign attribution access alone reduces hours of report-building each month.
- Revenue ops teams managing custom objects. If your data model is more complex than standard contacts, companies, and deals, schema discovery and custom object access give AI a map to work from.
- HubSpot super admins at growing companies who already use ChatGPT regularly. This is the natural next step once your team is comfortable with AI in daily work.
- Business owners who want faster answers without hiring an analyst. Natural-language CRM queries lower the technical bar for pulling meaningful data out of HubSpot.
Humans who should wait: teams with messy CRM data, unclear permission structures, or no governance policy for third-party AI tools. Get the foundation right first or the connector will surface problems faster than you can fix them.
George's Take
“The connector isn't magic. It's a mirror. It reflects the quality of your HubSpot setup back at you through AI. Clean data, smart permissions, and a clear schema make it brilliant. Messy portals make it dangerous.”
I've seen this pattern across hundreds of portals. When you give AI access to your CRM, the first thing it does is expose every assumption you made when you built it. The connector's schema discovery feature is genuinely useful, but it's also humbling. It shows you exactly how your data model looks to a system that doesn't fill in the gaps charitably. That's not a criticism of HubSpot's update. It's a reminder that the tool is only as capable as the foundation underneath it. We've talked about where the line is between a tool and something that starts deciding for you, and this connector sits right on that line. Used carefully, with proper permissions and a clean data model, it's one of the most practical AI integrations HubSpot has shipped. Used carelessly, it's a fast path to writing bad data into your CRM at conversational speed. Start slow. Verify outputs. Keep humans in the loop on write actions until you trust what you're seeing.
If you want help evaluating whether your HubSpot portal is ready to support this kind of AI integration, or if you want a second set of eyes on your permissions and data model before you roll it out, let's talk. Book a strategy call with the Sidekick team and we'll walk through your specific setup so you can move forward with confidence, not guesswork.
Frequently Asked Questions
What's new in the HubSpot connector for ChatGPT?
HubSpot expanded the ChatGPT connector to support campaign management, landing page creation, leads records, custom and activatable object access, natural-language CRM queries, and schema discovery. These additions align the connector with HubSpot's MCP server capabilities. Existing users may need to upgrade or reconnect the connector to access the new features.
Who can use the HubSpot ChatGPT connector?
The connector is available to all HubSpot customers across all plans who also have an OpenAI account. However, specific features like campaign management, content analytics, and custom object access may require a higher-tier HubSpot subscription and additional permissions. A super admin must install and configure the connector before other users can access it.
Is the HubSpot ChatGPT connector safe to use?
The connector mirrors your existing HubSpot permissions exactly, so ChatGPT can only access what your account already allows. Super admins can revoke access at any time through HubSpot's governance beta. Since ChatGPT is an OpenAI product, review HubSpot's data handling documentation and align with your IT and legal teams before a broad rollout.
Can ChatGPT create or update HubSpot records?
Yes. The expanded connector supports write actions including creating and updating campaigns, landing pages, leads records, custom objects, and activatable objects. Write access depends on your HubSpot subscription tier and user permissions. Test these capabilities with a single trusted user before enabling them across your team.
What is schema discovery in the HubSpot ChatGPT connector?
Schema discovery lets ChatGPT explore your specific HubSpot account's objects, properties, associations, and field definitions before taking any action. This means ChatGPT works with your actual data model rather than a generic HubSpot structure. It's the recommended first step when setting up the connector so you can confirm it's reading your configuration correctly.
Do I need to reconnect the HubSpot ChatGPT connector after this update?
Possibly. HubSpot states that customers already using the connector may need to upgrade or reconnect it to access the new campaign, content, custom object, and write capabilities. A ChatGPT workspace admin may also need to enable the updated connector version at the workspace level. Check both sides of the integration before assuming the new features are live.
