In a widely discussed LinkedIn post, David Elkington, founder and CEO of Atonom, while blasting HubSpot’s autonomous decision to sell their customers’ enrichment data (which stands reversed, following widespread criticism), called customer data, not Breeze, as HubSpot’s ‘most valuable asset.’ 

In context, Elkington was pointing to what he believed motivated the decision. For marketers, however, the remark carries a broader lesson.

His assertion reveals the position Breeze AI holds in the HubSpot community.

Breeze is a significant addition to HubSpot, and it has been a massive success for HubSpot. Fueled heavily by the launch and maturity of Breeze, HubSpot’s total revenue for 2025 grew 19% YoY to $3.13 billion, with its total customer base surging past 288,000 globally. 

Yet Elkington’s observation is a useful reminder that AI capabilities should be viewed in perspective. 

In today’s post, we’ll provide the perspective while covering Breeze’s powerful capabilities. Let’s begin.

What AI features does HubSpot have in Breeze?

HubSpot’s AI suite is supported by 3 pillars, each with distinct functionality and pricing:

• Breeze Assistant is conversational AI, a chat window in your account that answers questions and completes small tasks on request. For example, it can summarize deals, create lists of unsubscribed contacts, research companies using CRM and external data, or generate reports from prompts. Copilot is available on all plans and tiers.

• Agents perform tasks with minimal supervision. HubSpot offers agents for social media, sales prospecting, customer service, and content creation, with additional options such as call recaps, deal-loss analysis, and account handoffs available through the agent marketplace. Access to specific agents depends on your subscription; most require the Pro or Enterprise tier.

• Breeze Intelligence focuses on data management by automatically enriching contact and company records, reducing form fields for returning visitors, and providing a buyer-intent tool to identify companies browsing your site. This tier uses a separate credit system, purchased in packs, which should be reviewed before enabling automatic enrichment for your entire database.

HubSpot AI features are also available outside these categories, including record summaries, report generation, workflow suggestions, and text proofreading. 

What is Breeze Assistant in HubSpot? 

HubSpot Breeze Assistant is the platform’s centralized conversational AI interface. 

It is integrated within your portal to help marketing, sales, and customer service teams efficiently find data, create content, and complete manual tasks without navigating complex menus. With secure access to your internal CRM data, HubSpot Academy resources, and website analytics, Breeze Assistant provides context-specific support tailored to your business. 

The assistant serves as a platform-wide generalist, managing a range of daily tasks:

• Summarize records and activities: Breeze Assistant can instantly summarize contacts, company histories, or recent deal activities, allowing you to quickly prepare for meetings without reviewing extensive records.
• Draft and refine content: It can write sales outreach emails, draft follow-ups, optimize social media copy, and generate content directly within your daily workflows.
• Navigate and learn HubSpot: If you need assistance building complex workflows or completing specific functions, the assistant provides step-by-step guidance.
• Run custom assistants: If your admins create AI assistants in Breeze Studio, you can access and use these custom bots directly within the main Breeze Assistant interface.

What are HubSpot Breeze Agents?

Breeze Agents are autonomous, AI-powered digital teammates in HubSpot. These agents work in the background to execute multi-step workflows, such as creating content, qualifying sales leads, or resolving customer support tickets, using your existing CRM data and brand context. 

The table below lists what agents are good for Breeze across marketing and sales. 

HubSpot Breeze AI Agent	Primary function	Use cases
Prospecting Agent	Automates early-stage sales outreach and qualification.	• Monitors target accounts for buying intent and signals• Drafts and sends highly personalized, timely email outreach to leads.
Customer Agent	Provides 24/7 autonomous customer support across chat channels.	• Instantly resolves common customer queries using your existing knowledge base.• Handles transactional requests (e.g., tracking an order, resetting a password).
Content Agent / Content Remix	Automatically generates and repurposes marketing assets.	• Builds ready-to-publish blog drafts from a single keyword or prompt.• Splits a single asset (like a landing page) into social posts, emails, and case studies.
Data Agent	Cleans, enriches, and structures data directly within workflows.	• Standardizes formatting inconsistencies (e.g., phone numbers or text fields).• Extracts insights from text-heavy form responses or recent call transcripts.
Company Research Agent (Beta)	Gathers deep background context on target accounts.	Aggregates recent company news, technology stacks, and market trends before a sales call.
Deal Loss Agent (Beta)	Analyzes friction in the sales pipeline.	Evaluates closed-lost deals to surface micro-trends and recommend win-rate improvements.
Customer Health Agent (Beta)	Monitors customer retention signals.	Assesses account satisfaction metrics and flags accounts at risk of churning, suggesting next steps.
Sales to Marketing Feedback Agent (Beta)	Bridges communication between revenue teams.	Synthesizes everyday sales team data and objection insights into actionable feedback for marketing campaigns.

Buyer intent is one of the most practical tools in the lineup. It identifies which companies have visited your site, how many pages they viewed, and, through the signals tab, highlights events such as funding rounds, leadership hires, or product launches. This information is sourced from a network of sites sharing intent data via HubSpot’s tracking code. However, the quality of that data isn’t uniform. It tends to hold up better for US-based companies than elsewhere. Some signal types (technology usage, research intent) are more reliable than others (general company news, which can miss things entirely) 

Use this tool to identify strong leads. Filter results by your target market and intent criteria, save relevant views, and set up a recurring digest to avoid manual checks. 

HubSpot Breeze allows you to use external large language models. You can connect to models such as OpenAI, Anthropic’s Claude, and Google’s Gemini through marketplace connectors. However, you must select a specific model version during setup. There is no option to automatically use the latest version. If a provider retires a model, any workflow using it will stop functioning until updated. Additionally, newer or larger models may quickly reach the external provider’s usage limits, separate from HubSpot’s charges.

Data handling and governance

Anyone routing customer or prospect data through these tools should be aware of where the data is sent. HubSpot states that personal information is masked before leaving its systems, and data processed through agents and assistants is subject to a zero-data-retention agreement with the model provider. This means the data is not stored or used to train external models. 

Beyond vendor policies, governance is primarily an internal responsibility.

Teams that allow unrestricted experimentation with connected AI tools often restrict access later, once they understand the extent of data visibility. The same is true within HubSpot. Connecting an external AI assistant, such as ChatGPT, Claude, or Gemini, only provides access to data the connecting user can already view in HubSpot. However, if a super-admin establishes the connection and shares it broadly, team members may gain access to records they should not see. It is important to determine who sets up these connections and what permissions are granted.

To maximize the benefits of HubSpot AI features, organizations must be prepared. According to the Dynamic Capabilities theory, success depends not on owning resources, but on an organization’s ability to sense and seize opportunities, as well as reconfigure resources in response to change. Equally important is the human element. Managers play a critical role in determining how data is governed, how AI outputs are trusted, how customer journeys are redesigned, and how employees use AI. 

How much does HubSpot Breeze AI cost?

The short answer: HubSpot Breeze AI includes free basic features (like Copilot), while its advanced AI agents use an outcome-based credit system ($10 per 1,000 credits). This translates to $0.50 per resolved customer conversation, $1.00 per recommended sales lead, and $0.10 per data agent answer after utilizing your plan’s built-in monthly credit allowance. 

For a more detailed breakdown of pricing and credits, refer to the table below. 

Feature / product	Cost model	Details & credit consumption
Breeze Copilot	Included at no extra cost	Available across all subscription plans.
Custom Assistants	Currently free	Strategically unmetered by HubSpot to encourage user adoption (subject to change).
Breeze Agents (Flagship) (Customer, Prospecting, Data)	Consumption-based (HubSpot Credits)	Draws from your plan’s monthly credit pool or paid top-ups:• Customer Agent: 50 credits ($0.50) per resolved conversation.• Prospecting Agent: 100 credits ($1.00) per recommended lead; 10 credits ($0.10) per company research.• Data Agent: 10 credits ($0.10) per response.
Breeze Agents (New/Beta)	Free during Beta period	HubSpot provides a 30-day notice and in-product banners before converting a free agent to a credit-charging model.
Breeze Intelligence (Enrichment & Form Shortening)	Tiered Capacity Packs	Sold in monthly, non-rolling subscription packs:• 100 credits: ~$30–$45/mo• 200 credits: ~$60/mo• 1,000 credits: ~$150/mo• 10,000 credits: ~$700/mo(1 credit enriches a single contact/company record; combined forms may use >1 credit).
Basic Data Enrichment	Included at no extra cost	Baked into every HubSpot portal for basic lookups; bulk/automated tasks still require credit packs.

The real risk isn’t the pricing structure so much as the absence of any warning before you spend more than intended. Turning on automatic enrichment, tracking buyer intent broadly, or running an agent across a large workflow segment can burn through a month’s credits fast. Checking the usage dashboard regularly, and testing any new automation on a small, manually controlled group before opening it up to the entire database, is the most reliable way to avoid a bill shock. 

Getting started with HubSpot Breeze AI 

Select a recurring, time-consuming task and implement a targeted solution for it. Confirm whether it uses credits before expanding further. Once you see time savings, proceed to the next task. 

As we wrap up, please keep the following points in mind:

• ​Credit-based features can accumulate costs in no time. Enrichment, agent runs, buyer intent, and smart properties all use the same credit pool. Automate thoughtfully.
• Never trust any piece of AI content without getting it scrubbed by a human first. 
• Test new processes with a small group before deploying them to customers.
• Giving feedback matters. A thumbs-down on a wrong summary or answer feeds back into how the system improves, so it’s worth the extra click. 
• Model access may be unstable. HubSpot has previously adjusted or temporarily paused access to certain models due to factors beyond its control.