[object Object]

A practitioner asks “what is Breeze, exactly?” and gets three different answers from three HubSpot pages. The product evolved fast in 2026 and the marketing surface has not fully settled. The short version: Breeze is HubSpot’s umbrella for AI features across copilot, agents, and embedded intelligence. The longer version determines what to enable, what to govern, and what to wait on.

The three layers

Breeze
  ├─ Copilot      - chat assistant inside HubSpot UI
  ├─ Agents       - autonomous specialists for recurring work
  └─ Intelligence - embedded AI in forms, reports, emails, content

These layers ship at different cadences and have different governance profiles. Treating them as one product leads to underutilization of the agents and overuse of the copilot for tasks better handled elsewhere.

Copilot

A chat interface accessible across HubSpot. It answers questions about your portal data, drafts emails and content, navigates the UI on request, and surfaces shortcuts. Admins enable copilot per user under Settings > Account Setup > AI. Common patterns:

"Show me deals over $50k closing this quarter assigned to me"
"Draft a 3-bullet meeting summary from the call transcript"
"Find contacts at Acme who haven't been emailed in 30 days"
"What's the typical close rate for deals from my best lead source?"

Train reps on prompt patterns. The copilot is dramatically more useful with concrete asks than with vague ones.

Agents

Specialized AI workers running recurring tasks in the background. Current lineup includes:

Prospecting Agent  - finds leads matching ICP, scores, writes outreach
Content Agent      - generates blog drafts, social posts from briefs
Customer Agent     - handles tier-1 support across 9 channels
Social Agent       - drafts and schedules social posts
Data Agent         - cleanses and enriches CRM properties
Company Research   - enriches company records on creation

Each agent has its own configuration, audit cards (the trust layer), and increasingly outcome-based pricing. Rolling GA continues through the year — verify status before depending on any single agent.

Intelligence features embedded across the hubs

Smaller AI features baked into existing tools: smart subject lines, content summarization on long pages, predictive forecast adjustments, smart form fields. Some are tagged “Powered by Breeze,” others are not labeled at all. Review the AI features list quarterly because it grows.

Audit cards: the trust layer

Every agent action surfaces in an audit card that shows what changed and why. This is the single feature that moves trust from “the AI is doing things to my data” to “I can see what the AI did.” Make audit-card review part of weekly team rituals during rollout.

Governance to settle before broad rollout

- Data residency: where does processing happen
- Retention: how long is request/response data stored
- LLM provider: which model handles which task
- Human-in-loop: which actions require human approval
- Per-team access: which agents are enabled for which teams
- Audit trail: how long are audit cards retained
- Opt-out: how a customer disables AI on their data

Document answers in a one-page AI policy and revisit when HubSpot updates its data terms.

Pricing models to watch

HubSpot has shifted Customer Agent and Prospecting Agent to outcome-based pricing — pay per resolved conversation or per qualified lead. Other agents remain on traditional usage models. Bills can swing meaningfully during early adoption — set spend alerts and review weekly for the first month.

What to do this week

List which Breeze features are enabled in your portal, classify each as copilot/agent/intelligence, document the governance answers above, and pick one agent to pilot with a clear success metric before expanding.

[object Object]
Share