You’re paying for AI your team isn’t using.
Most portals we see use a fraction of their monthly AI credits, and credits don’t roll over. The ones that do use them often spend them on low-value jobs.
Credits expire unused.
Your plan includes AI credits every month. Unused ones are gone at the end of it.
Credits go to the wrong jobs.
A few busy workflows drain the pool while the work that matters waits.
Your agent gives stale answers.
The Customer Agent went live, then nobody kept its knowledge current.
The team uses ChatGPT instead.
People paste customer data into outside tools, outside the CRM and outside your rules.
Are you using the AI you pay for?
Pick your tier and how much you used last month. See what your unused credits could be doing, and where they’re best spent.
Breeze Assistant and standard enrichment use no credits.
Pick your tier and last month’s usage. You'll get:
- What you’re likely leaving unused
- What those credits could be doing
- Where they’re best spent
Credit amounts and rates are approximate, based on HubSpot's published credit rates as of September 2026. HubSpot sets them and can change them.
The delivery model matters more than the tool.
Every one of these lives inside your HubSpot. What changes is how deep the AI reaches into your data, and who you pay for it. We start with what HubSpot already does and go further only where it stops.
HubSpot native · you pay HubSpot
Breeze Assistant
AI help inside HubSpot for everyone: drafting, summaries and research on the record.
- Fits when
- your team is under-using HubSpot and wants quick wins.
- You pay:
- Nothing extra. It’s included with your seats.
Breeze agents
The Customer, Prospecting, Knowledge Base and Data agents, set up on your brand, rules and knowledge, and grounded in your ICP and deal history.
- Fits when
- you’re on Service or Sales Hub Pro with standard questions and modest volume.
- You pay:
- HubSpot, in credits, as the agent resolves conversations or works leads.
AI in your workflows
Breeze actions inside the workflows you already trust: classify, summarize and fill properties automatically.
- Fits when
- the step is simple and the volume is modest.
- You pay:
- HubSpot credits for each action.
Pivot hosted · one flat monthly fee
Workflows with Claude actions
The same workflows, but the AI step runs on Claude through Pivot, with outside data, longer rules and better reasoning.
- Fits when
- volume makes credits expensive, or the logic outgrows Breeze.
- You pay:
- Pivot’s flat monthly fee. No credits.
Custom Claude agents
An agent built for your job, grounded in your own data and rules, with Claude doing the reasoning. It can run on your site or inside HubSpot.
- Fits when
- you don’t have the seats Breeze needs, volume is high, or the knowledge lives outside HubSpot.
- You pay:
- Pivot’s flat monthly fee. No seats, no credits.
Both · HubSpot up front, Pivot behind it
Hybrid: HubSpot front end, Claude back end
Your HubSpot chat and inbox stay the interface, with rep handoff and logging, while Claude does the reasoning behind it.
- Fits when
- reps live in the HubSpot inbox but Breeze’s answers aren’t good enough.
- You pay:
- Your existing seats, plus Pivot’s monthly fee.
Embedded record cards and pages
Pivot’s intelligence shown right on HubSpot records and pages: intent, AEO results, agent activity. No new tab, no new login.
- Fits when
- your reps ask “so what do I do with this?”.
- You pay:
- A small monthly add-on.
Most agents answer FAQs. Ours apply your rules.
A standard Customer Agent looks up an answer in a knowledge base. Our custom Customer Agents send the hard questions to Claude, which works through your policies and the customer's details step by step before it answers, and hands off to a person when it should.
Ask the agent a question. You'll get:
- What it knew about the customer
- Which of your rules it applied
- What it did next
Eight questions. One honest answer.
The same test we run on scoping calls. If every answer is no, build it in HubSpot, and we'll help you do it. Any yes is a job Breeze can't do well on its own.
1. Does it need information from outside HubSpot?Your ERP, a website, a document, a public database
2. Are the rules longer than a page?Breeze workflow prompts are limited to about 3,000 characters
3. Does it need a structured answer, not one value?A score plus reasons, or several fields at once
4. Does it take more than one reasoning step?Check this, then decide that, then write something else
5. Are you missing the tier or seats it needs?Some agents need Professional or specific seats
6. Does the knowledge live outside HubSpot?Drives, wikis or people’s heads, not your HubSpot knowledge base
7. Will customers read what it writes?Customer-facing text needs to be right, every time
8. Could another team use up the shared credits?Workflows stop when the pool runs out
Answer the first question to start. You'll get:
- Build it in HubSpot, or not
- Which questions decided it
- The right next step
The differentiator isn't the AI. Everyone has AI. It's the write-back.
Your team already knows how to judge these records. We teach an agent your rules, run it on your data, and put the answer on the record, so your workflows, your lists and your reps can use it without anyone opening a spreadsheet.
- TriggerA record changes, a form comes in, or a schedule runs.
- GatherThe agent pulls what it needs from HubSpot and your other systems, not just what's on the record.
- JudgeIt applies your rules, at full length, and reasons through each step.
- WriteThe answer goes into a HubSpot property, attributed and reversible.
- ActYour workflows, lists and reps act on it. Results can show in a card on the record.
Breeze shows you what's already in HubSpot. Our agents and record cards bring in what HubSpot doesn't have, in the screens your team already uses.
Agents we’ve built, and what they changed.
Monthly client reports, drafted by an agent
12–25 hours reclaimed a month
Each client report used to take 1–2 hours of pulling and formatting data across 15+ accounts. An agent in our own HubSpot, trained on our templates, data sources and voice, now pulls the dashboards, drafts the summary and flags trends, so the team analyzes instead of compiling.
Two agents, two people, one month
12 hours saved in the first 30 days
As a pilot, we turned on two Breeze agents for two people on our team. In the first 30 days, HubSpot measured 12 hours saved, 6 hours per person, using tools already included in the subscription.
Customer agents with Claude behind them
Advanced logic, in HubSpot’s chat
For questions a knowledge base can’t answer, custom Customer Agents send the conversation to Claude through Anthropic’s API, apply the client’s own rules, and hand off to a person with a summary when a human should decide.
A deal deduplicator that cleans up the pipeline
Duplicate deals found and resolved
Duplicate deals inflate the pipeline and split activity across records. A custom agent finds deals that are really the same opportunity, decides which one should survive, and flags the rest for merging, so forecasts count each deal once.
400+ tickets a month, routed by an agent
72% → 94% routing accuracy
Senior staff spent 15–20 hours a week categorizing tickets. A custom coded action calls Claude to classify each ticket by category, urgency and sentiment, with a human review queue for edge cases. Time to first response fell 60%, and 15+ hours a week came back.
173 workflows, cleaned up before AI went live
Agents worked on day one
Conflicting workflows and inconsistent data were producing unreliable AI answers. Workflows were consolidated by 29%, lifecycle stages standardized and the data restructured, so the Customer Agent and Breeze Assistant worked correctly from the first day.
Fear shouldn't be the guardrail. Governance should.
AI without context is just expensive autocomplete. The answer isn't limiting what it can see. It's expanding access safely, with the same rules IT has used for decades.
Read-only by default
Agents can read what they need. Write access needs your approval, property by property.
Every write is attributed and reversible
You can see which agent changed what, and roll it back.
The code decides what can’t be undone
The model reasons; our code, not the AI, controls anything irreversible.
Keys and conversations kept safe
API keys live in encrypted secrets. Internal HubSpot conversations stay out of AI with “never log” settings. Paid business AI accounts only.
You’re never a hostage
Every agent is documented, one per client, and the rules are yours to keep.
Three levels. Same team either way.
Simple activations don't need a six-week project. Custom builds deserve more than a checkbox setup. Every price is fixed before we start.
Agent Activation & Training
- Configuration and brand voice
- Team training
- Breeze credit forecasting
- Delivered in 1–2 weeks
Best for: Prospecting, Social Post, Knowledge Base and Deal Loss agents
Agent Implementation & Optimization
- Knowledge base audit
- CRM and ticketing integration
- Performance dashboard
- Delivered in 2–4 weeks
Best for: Customer Agent, Data Agent and ABM Landing Page Agent
Custom AI Solutions
- Claude or OpenAI API integration
- Cloud hosting and monitoring
- 30 days of post-launch support
- Delivered in 3–6 weeks
Best for: live CRM queries, custom coded actions and predictive models
Agents Pivot hosts (Claude actions, custom and hybrid agents, record cards) add a flat monthly fee for hosting, monitoring and fixes, with no seats and no credit meter. Several agents or AI across teams? Start with a Project Blueprint, $2,500 and credited. Onboarding HubSpot? AI agents added to your onboarding are bundled at a lower price.
Not sure where AI fits? Start free.
Both end with specific recommendations you keep, whether or not you work with us.
AI Readiness Assessment
We look at your HubSpot and your goals, and tell you exactly what AI makes sense. If all you need is a $2,500 activation, we’ll say so.
What we cover- Your HubSpot setup and subscription tier
- AI agent opportunities matched to your goals
- Data quality: what needs cleaning up before AI can help
- Breeze credit forecasting, so there are no surprise costs
- 2–3 quick wins you can turn on right away
- Clear next steps and a realistic timeline
You get: a written AI readiness summary with specific recommendations.
Support Email Audit
Is your support inbox costing you customers? Send us one week of support email, with customer details removed. We’ll show you what automated triage, SLAs and routing would have done with it, and how much agent time it would have saved.
That’s the work automated triage takes over. The audit shows how much of it applies to your inbox.
From credit check to first agent in weeks.
- Free credit checkUse the tool above, or connect HubSpot read-only for a free instant audit that includes your AI usage.
- Put your credits to workWe set up Breeze on the jobs worth the most, and train your team to use it.
- First custom agentWhere Breeze stops, we build one agent, test it on twenty of your real records, and let you judge the results.
- ScaleAdd agents as each one proves itself.
What you'll want to know.
Why not just use ChatGPT ourselves?
Free ChatGPT answers from the open internet, including your competitors' claims and details it makes up. Ours answers from your data, in your HubSpot, in your voice, and when it doesn't know, it says so rather than inventing an answer.
Will you try to replace Breeze?
No. Breeze is the right start. We set up what it does well first, and build custom only where it can't do the job well on its own. The free test above is how we decide.
Is our data used to train AI?
No. We use paid, business AI accounts whose terms don't train on your data, keep API keys in encrypted secrets, and keep internal HubSpot conversations out of AI with HubSpot's "never log" settings.
Can the AI change our data?
Only where you've approved it. Agents are read-only by default. Anything that writes is attributed to the agent and reversible, and the code, not the model, decides anything that can't be undone.
What if we stop working with you?
You keep everything. Agents are documented and the rules are yours. You're never a hostage.
What HubSpot tier do we need?
It depends on the agent. The Customer Agent needs Service Hub Professional or Enterprise, the Prospecting Agent needs Sales Hub Professional or Enterprise, and Breeze Assistant is included at every tier. Pivot-hosted agents need no extra seats. We map it to what you already own.
How long does it take?
1–2 weeks for an activation, 2–4 weeks for a full implementation, and 3–6 weeks for a custom build. Most teams see value in the first week.
What if an agent gets something wrong?
Fixing what broke is included in the monthly fee and never metered. We'd rather you tell us than put up with it.
Talk it through in 15 minutes.
Bring what you want AI to do. We’ll tell you whether Breeze can do it, what it costs in credits, and where a custom agent would do better.