Managing what your HubSpot AI knows

Close the gap between what your AI can reach and what it should.

Three questions about your AI's knowledge layer

An AI is only as safe as the knowledge it can reach, and most teams have not decided what that should be.

HubSpot's knowledge vaults bound what Breeze knows, but a connected model like Claude can draw from elsewhere.

Without a decision on where the intelligence layer lives, the AI's knowledge is whatever it happens to find.

Why does what the AI knows decide how safe it is?

What an agent can access is the real boundary on what it can say or do.
Knowledge vaults let you curate that, attaching files, CRM content and segments to a specific agent.
Set the knowledge deliberately and the AI's behaviour follows.

Why does what the AI knows decide how safe it is?

Data_Library_Illustrations_Color_LightBG

Why is Breeze's answer not Claude's answer?

Vaults govern Breeze, HubSpot's own AI, by attaching curated context to each agent.
A model reached through the connector draws on what the connector can see, not on a vault.
So one business can have two different knowledge boundaries, and needs to choose each deliberately.

Why is Breeze's answer not Claude's answer?

Why decide where the intelligence layer lives?

The choice is to route the model through curated HubSpot context, or give it broader CRM access.
That decision sets the AI's knowledge, its risk surface and what it could leak.
Made deliberately, the intelligence layer becomes a control rather than an accident.

Why decide where the intelligence layer lives?

Customer_Platform_Illustrations_Color_LightBG

Name a media community platform

Outstanding HubSpot Partners with amazing work ethics

"They analysed our business model and community structure carefully and came up with an efficient, sustainable plan."

Julia Smith

Head of Digital and Community

Ready to discuss your HubSpot project?

Let's take our relationship up a level.

Simply fill in the form below...

(I'll get back to you ASAP)

Prefer another way?

FAQs

How long does setting up the knowledge layer take?

Usually 2 to 3 weeks. The first stage decides what each agent should and should not know. The second builds the vaults, or scopes the connector access, and documents the boundary.

Can HubSpot control what its AI knows?

Yes. Knowledge vaults in Breeze Studio attach curated files, CRM content and segments to a specific agent, so it draws only on what you choose. PYB sets these up so the boundary is deliberate.

What about a connected model like Claude?

A model reached through the connector draws on what the connector can access, not on a Breeze vault. So the intelligence layer for Claude is a separate decision, and PYB helps you choose between curated context and broader access.

What HubSpot products does this need?

Knowledge vaults are available across the hubs from Starter upward, with the richer AI features higher up. The connector-based path depends on how the integration is set up.

Does HubSpot meet our AI governance requirements?

HubSpot holds SOC 2 Type II and ISO 27001. PYB adds its own ISO 27001, ISO 9001 and ISO 42001 certifications, so the people governing your AI are themselves audited against the AI and information-security standards your buyers and regulators ask about.

Talk to PYB about where your HubSpot AI's knowledge should live.

A 15-minute call to walk through what each agent should and should not know, and whether to route Claude through curated context or broader access, and what closing the gaps looks like. No prep, no pitch deck.

Book a 15-minute call