Skip to content
  • There are no suggestions because the search field is empty.

HubSpot AI and Breeze: understanding AI tools and capabilities

Understand how HubSpot Breeze Assistant, AI agents, Data Agent and data enrichment work, including permissions, Credits and governance considerations.

HubSpot includes a growing range of artificial intelligence tools designed to help users create content, understand CRM information, research data and automate suitable activities.

These capabilities are delivered through Breeze, HubSpot’s AI technology, and are increasingly embedded throughout the platform rather than provided as one standalone tool.

Available functionality depends on your HubSpot products, subscription, seat, permissions, account-level AI settings and use of HubSpot Credits.

What is HubSpot Breeze?

Breeze is the name HubSpot uses for its AI technology and related capabilities.

Depending on your HubSpot account and configuration, these can include:

  • Breeze Assistant
  • HubSpot’s specialist AI agents
  • Agent Hub and the agent builder
  • Data Agent
  • AI-powered content generation
  • AI actions within workflows
  • AI context and knowledge vaults
  • Data enrichment

It is useful to think of Breeze as an umbrella for HubSpot’s AI capabilities rather than one individual tool.

Breeze Assistant

Breeze Assistant is HubSpot’s conversational AI assistant for sales, marketing and service teams.

Users can interact with Breeze Assistant using natural-language prompts. Depending on the tools, data and permissions available to the user, it can help with activities such as:

  • Generating and refining content
  • Summarising CRM information
  • Preparing for meetings
  • Reviewing performance information
  • Exploring reports and data
  • Working with files and supported connected applications
  • Completing supported actions in HubSpot

Breeze Assistant can use information available in your HubSpot account to provide more relevant responses. User permissions still determine which data and actions an individual can access.

Reviewing Breeze Assistant outputs

AI-generated responses should be reviewed before they are relied upon or shared externally.

The quality of a response can depend on:

  • The completeness and accuracy of your CRM data
  • The information Breeze Assistant is permitted to access
  • The context provided with the request
  • The clarity and specificity of the prompt
  • Your HubSpot configuration
  • The limitations of the underlying AI models

Breeze Assistant should therefore be treated as a productivity and decision-support tool rather than an automatic source of truth.

AI agents and Agent Hub

HubSpot also provides AI agents designed to support more specific goals or processes.

Unlike a general conversational assistant, an agent is normally configured around a defined purpose, set of instructions, sources of information and permitted actions.

Depending on the HubSpot products and features available to your organisation, agents can support areas such as:

  • Customer service
  • Sales prospecting
  • Data research and analysis
  • Content creation
  • Record review and processing
  • Other defined business processes

Agent Hub provides a central area for accessing and managing supported agents, context and related AI functionality.

HubSpot’s custom agent builder can be used to define an agent’s:

  • Instructions
  • Actions
  • Knowledge
  • Inputs
  • User and team access

The custom agent builder is currently a beta feature. Its availability depends on eligible Professional or Enterprise subscriptions, relevant permissions and HubSpot Credits.

Agent permissions and access

An agent should be given access only to the information and actions required for its intended purpose.

Administrators should consider:

  • Who can create or edit the agent
  • Who can run it
  • Which CRM records it can read or update
  • Which HubSpot tools it can use
  • Which files, integrations or knowledge sources it can access
  • Which actions it can perform
  • How its outputs will be checked
  • How its HubSpot Credit use will be controlled

Permission to run an agent does not override a user’s underlying permissions for other HubSpot features. If an agent relies on a tool the user cannot access, the user may be unable to complete the relevant action.

Custom agents should be tested before publication to confirm that they use the intended inputs and knowledge, follow their instructions and produce an appropriate output.

Data Agent

HubSpot’s Data Agent applies AI to CRM and operational data.

It can be used to answer questions, research information, transform data and support data-focused automation.

Its capabilities include:

  • Smart properties: CRM properties populated with AI-generated answers
  • Smart actions: Data Agent actions used within workflows
  • Smart columns: AI-generated columns used to research and analyse information in Data Studio

Depending on the configured source, Data Agent can use information including:

  • Existing CRM property data
  • Company websites
  • Web research
  • Call transcripts

Data Agent requires appropriate AI settings and user access. It also uses HubSpot Credits when generating responses for records.

Because its output can be written into CRM properties or used by automation, prompts, data sources and results should be tested carefully before being applied at scale.

Data enrichment

HubSpot also provides data enrichment for contact and company records.

Enrichment can add or update business information using:

  • HubSpot’s commercial dataset
  • Third-party data providers
  • Publicly available information

Depending on the record and the information available, enrichment can help populate details such as company name, job title, industry or company size.

This information can then support:

  • Segmentation
  • Reporting
  • Prospecting
  • Qualification
  • Routing
  • Automation
  • General CRM data quality

Data enrichment is optional and is turned off by default. Super Admins can control whether enrichment is used manually, automatically or continuously, and which users can enrich records.

Turning enrichment off prevents future enrichment but does not remove information that has already been added to CRM records.

Data enrichment and Data Agent are different

Although both features can add information to CRM records, standard data enrichment and Data Agent serve different purposes.

Data enrichment matches supported contact and company records with verified business information from HubSpot’s commercial dataset, third-party providers and public sources.

Data Agent uses AI prompts and selected sources to research, analyse or transform information for a defined purpose.

Organisations should understand whether a property contains:

  • Information entered by a user
  • Existing CRM data
  • Information added through standard enrichment
  • Information generated or researched by Data Agent

This distinction is important when information will be used for reporting, qualification, routing, automation or business decisions.

AI context and knowledge vaults

HubSpot’s AI tools can use two types of supporting information: context and knowledge vaults.

Context contains foundational information about your organisation that can be used across multiple AI tools. This can include:

  • Business details
  • Products and services
  • Brand information
  • Ideal customer profiles
  • Personas
  • Team and process information
  • Custom organisational context

Knowledge vaults contain information added for specific projects or agents.

Maintaining accurate context can help AI-generated outputs reflect your organisation more closely. Knowledge vaults can provide more focused information for a particular agent or use case.

Access and visibility should be considered carefully when adding files, customer information, business processes or other internal material.

AI settings, permissions and data access

HubSpot provides account-level AI settings and user permissions for controlling access to AI functionality and supported data sources.

Depending on the feature, administrators may be able to control access to:

  • Generative AI tools
  • Breeze Assistant
  • CRM data
  • Customer conversation data
  • Files
  • Data Agent
  • Agent creation and management
  • Individual agents and their actions

Enabling AI at account level does not mean every user needs access to every feature.

Apply the principle of least privilege by granting users and agents only the access required for their responsibilities. Permissions should be reviewed whenever roles, processes or AI use cases change.

Privacy, security and organisational governance

HubSpot provides technical settings for managing AI access and data use, but your organisation remains responsible for deciding how AI should be used.

Your organisation should consider:

  • Data protection requirements
  • Information security policies
  • Internal AI policies
  • Brand and content standards
  • Approval processes
  • Regulatory and contractual obligations
  • Restrictions on confidential or sensitive information
  • Human review requirements
  • Record retention and audit requirements
  • Acceptable use of external web research

Sensitive information should not be entered into AI prompts or made available to an agent unless your organisation has confirmed that doing so is appropriate.

A feature being technically available within HubSpot does not automatically make it suitable for every user, process or category of data.

Accuracy and human review

AI-generated information can be useful, but it should not be assumed to be complete or correct.

Human review is particularly important where AI is used to:

  • Produce client-facing content
  • Interpret CRM information
  • Research organisations or contacts
  • Populate CRM properties
  • Qualify or prioritise records
  • Trigger automation
  • Recommend or perform operational actions
  • Support decisions with financial, legal or customer consequences

Reviewers should be able to identify the source of important information and understand whether it came from CRM data, enrichment, external research or AI generation.

The level of review should reflect the potential impact of an inaccurate output or action.

Availability and HubSpot Credits

HubSpot’s AI functionality continues to develop, and each tool can have different requirements.

Availability may depend on:

  • Your HubSpot products
  • Your subscription tier
  • Your assigned seat
  • Your user permissions
  • Account-level AI settings
  • Participation in a beta
  • HubSpot Credits

Features including Data Agent and custom agents can consume HubSpot Credits. Administrators should understand expected usage, review available estimates and configure limits where supported before introducing credit-consuming features at scale.

Always check the current HubSpot documentation and the availability shown within your own account before designing a process around a particular AI capability.

What can Forbidden help with?

Forbidden can help you:

  • Review your current HubSpot AI configuration
  • Understand relevant Breeze features and use cases
  • Review AI permissions and access
  • Identify appropriate opportunities for AI within your CRM processes
  • Improve the CRM data and processes on which AI relies
  • Design and test suitable HubSpot automation and agent use cases
  • Review HubSpot Credit considerations
  • Help users understand how to use HubSpot AI effectively

Forbidden can support the technical configuration and practical adoption of HubSpot’s AI tools. Your organisation remains responsible for decisions relating to data protection, security, regulatory compliance, acceptable use and internal AI governance.

Further information

HubSpot regularly changes and expands its AI functionality. Refer to its current documentation when assessing or implementing a particular feature.