Roast index / HubSpot Breeze AI
Breeze AI — 1,000 credits per blog post, and none of them roll over
An AI that meters you by the question, priced in a currency only HubSpot prints, with an expiration date.
Breeze is HubSpot’s AI story, and credit economics is a genre of pricing that AI briefly made fashionable. HubSpot took it to its logical conclusion: an internal currency, with an expiration date.
The rate sheet is the comedy, unintentionally excellent:
| Agent action | Credits | At $0.01/credit |
|---|---|---|
| Generate one blog post (Content agent) | 1,000 | ~$10 |
| Resolve one support conversation | 50 | ~$0.50 |
| Qualify one lead fully | 1,000 | ~$10 |
| One AI workflow action | 10 | ~$0.10 |
Every AI action prices like a mini-SaaS in itself. Included credits don’t stack across products — buy three hub subscriptions and your credit pool is whatever your most expensive subscription decreed. Unused credits don’t roll over. The credits are, functionally, airline miles for a flight you paid full fare on.
The AI costs you $10/post at retail… while the same token at any frontier model provider is a faction of a cent. The delta is the credit monopoly margin.
The job, done lean
The genuinely useful AI in a GTM stack is embarrassingly simple:
- Sentiment scoring feedback and tickets (-1..+1, on your model provider)
- Clustering raw text into feature candidates
- Drafting summaries a human reviews
SuperSuite’s shared AI client does exactly this — an OpenAI-compatible client you configure per deployment. Your key, your provider, your rate limits, your spreadsheet. When the next model arrives better and cheaper, you switch a variable. Nobody’s “agent” re-priced itself at 1,000 credits.
Kindred roast, same meter, different door: Breeze agents vs. MCP — why your answers should live where the questions are asked.
The use case fit
You need it to: score feedback, cluster tickets into themes, draft a summary — with a model you chose, at a rate a CFO can model.
One act, no interval
The 31st of the month. Credits reset at midnight.
The credits meter, a play in one act
— Somewhere, an included credit pool shared across three products politely refuses to stack.
The lean alternative
honest tools, fairly priced
Open AI client
SuperSuite apps use an OpenAI-compatible client you can point at any provider — OpenAI, Anthropic-compatible gateways, or a self-hosted model. Sentiment in SuperFeedback, synthesis in ZenHero, and your API key, your bill, your terms.
Reading this at renewal season?
Run the porch checklist on your own account — six questions, one coffee.
What your team actually opened last week · what each used surface costs standalone · the three jagged thresholds most likely to hit you · the math at your three-year volume · your real exit cost · and an honest word on when the all-in-one earns its keep.
Take the checklist with you →Straight answers
How much does HubSpot Breeze AI cost?
AI actions are metered in HubSpot Credits. Included credits vary by (many) tiers — but don't roll over month to month. Beyond that, credits cost $0.01 each, e.g. 1,000 credits to generate one blog post, 50 per resolved AI support conversation, 100 per prospecting recommendation.
What does 1,000 credits mean in dollars?
At the pay-as-you-go rate of $0.010 per credit, one AI-generated blog post is about $10 — before the seat and subscription costs. A busy content agent could meter its way past a freelancer's monthly engagement.
Is there a way to use AI in the CRM without a credit meter?
Yes — bring your own model. SuperSuite apps call an OpenAI-compatible endpoint that you configure: pick providers, cap spend at the provider level, and keep the rate sheet in a spreadsheet instead of a catalog.
Answer blocks above are emitted as FAQPage structured data, because that is how information should travel now.