Solutions for 6 stages of the revenue journey
Our domains
How we work
New to HubSpot?
We set up Breeze AI agents and chatbots in your HubSpot portal so they draw on your data, follow your rules, and hand off to a person whenever they hit a wall. From picking the right use case through building the knowledge base to measuring results. We’re rolling out the same agents on our own portal right now.
HubSpot Partner since 2018
Breeze comes built into HubSpot, and agents activate in a few clicks. That’s exactly why so many companies end up in the same spot: the AI features are on, but nobody can say what the company actually gets out of them. The difference between an agent that helps and one that hurts isn’t the model. It’s the setup.
Breeze is already in your license, and you can activate an agent in an afternoon. Without a chosen use case, source material, and an owner, though, it stays a trial run: the team tries the agent out, nobody measures the results, and within a month everyone’s forgotten about it.
A badly configured chatbot is worse than none at all. It answers off the mark, spins the customer in circles, and won’t let them through to a person. Every conversation like that costs you a piece of trust in your brand.
The worry that an agent will make something up from your data is a fair one. An agent turned loose on outdated documentation answers confidently, and wrong. Without prepared sources and guardrails, it can’t be trusted, and the team figures that out fast.
The agent answers, but nobody reads the conversation transcripts, tracks how many queries it resolved, or tunes it based on that. AI isn’t a project that finishes once: without ongoing care, answer quality drops with every change to your offering.
An AI agent is only as good as the source material, rules, and oversight it’s given. Turning the feature on is the easy part of deployment. The outcome is decided by everything around it: picking the right use case, the knowledge base, the guardrails, and the measurement.
A company turns on a chat agent with default settings and points it at the entire website as its source. The first few days look good: the agent answers smoothly and the team celebrates.
Then a customer asks about return policy terms. The agent answers based on a page that’s three years out of date and nobody ever deleted. The customer holds the company to that answer, support spends hours explaining, and the sales director wants the agent switched off.
The company turns the agent off and closes the book with ‘AI isn’t for us yet.’ But it wasn’t the AI that failed. It was the deployment.
None of this has to happen. Deploying an AI agent has known steps and known risks, and both can be managed.
We treat deployment like onboarding a new team member. First we pick a use case where the agent has a real chance to help, prepare its source material, set the rules, and test it before it ever talks to a customer. We’re going through the same process ourselves right now: rolling out the Breeze Customer Agent on our own customer support and the Prospecting Agent in our own sales team. What we recommend to clients, we try on ourselves first. Deploying AI agents is one discipline within our implementation practice; for exactly what each Breeze agent can do, see our HubSpot Breeze AI product page.
An AI agent isn’t a feature you flip on. It’s a new colleague who needs training.
We review the volume and structure of your queries, the state of your source material, and your license. We’ll recommend where an agent makes sense, and tell you plainly where it doesn’t yet. Not every process is ready for AI.
The agent answers only from what you approve. We build the knowledge base, clean out outdated material, and connect sources so answers rest on current data.
We set the agent’s role, tone, and boundaries: what it can handle on its own and when it hands the conversation to a person. Before launch, we test it against real queries from your own operation.
From day one, we track how many conversations the agent resolved, where it handed off, and where it fell short, and tune it accordingly. Deployment isn’t the finish line: the agent needs upkeep, and so does the knowledge base it runs on.
The typical first step into AI in HubSpot: an agent that answers repetitive customer questions on the website, in chat, and by email, around the clock. Your support team gets to focus on the cases that actually need a person.
We connect the knowledge base and approved sources, set the tone of the responses, and define the handoff rules to your support team. The agent answers only from material you’ve approved.
A chatbot that helps visitors find an answer, qualifies their interest, and hands the conversation to sales or support along with the full context. No helpless ‘I don’t understand that’ loop.
Without source material, the agent has nothing to answer from. We help build or clean up the knowledge base: article structure, content updates, and rules for who maintains it going forward.
An agent that prepares the ground for your sales reps: company research, context pulled from the CRM, and draft first-touch outreach. The rep decides and sells, and leaves the routine work to the agent.
We set up the agent on top of your CRM: who to track, which signals to evaluate, and in what tone to draft outreach. The rep approves every draft; nothing goes out on its own.
The quality of the drafts depends on the quality of the data. We review the data fields the agent draws on and recommend what to fill in so the personalization is grounded in fact.
Agents that save time inside the company: content written in your tone, plain-language answers pulled from reports, and an assistant for everyday CRM work. We deploy them wherever routine work is eating your team’s time.
We set the Content Agent up with your brand voice and rules for working with source material, so the output sounds like you. The team gets a workflow where AI drafts and a person approves.
The Data Agent answers questions against your CRM data: whoever asks gets a number and the context around it. We set which data the agent can access and how to verify the results.
Whichever agent you’re deploying, the deployment skeleton is the same. Three things decide whether an agent helps or hurts.
An audit of your existing documentation, a cleanup of outdated content, and a structure for the sources the agent is allowed to draw on. This includes deciding what the agent shouldn’t know.
We define the agent’s role and tone, the topics it avoids, and the moment it hands the conversation to a person. We test the rules against scenarios from your own operation before the agent ever meets its first customer.
We set up metrics (resolved conversations, handoffs, satisfaction), train your team to manage the agent, and agree a cadence for regular tuning. Ongoing agent operation can be handed to our HubSpot managed services.
The typical first step into AI in HubSpot: an agent that answers repetitive customer questions on the website, in chat, and by email, around the clock. Your support team gets to focus on the cases that actually need a person.
We connect the knowledge base and approved sources, set the tone of the responses, and define the handoff rules to your support team. The agent answers only from material you’ve approved.
A chatbot that helps visitors find an answer, qualifies their interest, and hands the conversation to sales or support along with the full context. No helpless ‘I don’t understand that’ loop.
Without source material, the agent has nothing to answer from. We help build or clean up the knowledge base: article structure, content updates, and rules for who maintains it going forward.
An agent that prepares the ground for your sales reps: company research, context pulled from the CRM, and draft first-touch outreach. The rep decides and sells, and leaves the routine work to the agent.
We set up the agent on top of your CRM: who to track, which signals to evaluate, and in what tone to draft outreach. The rep approves every draft; nothing goes out on its own.
The quality of the drafts depends on the quality of the data. We review the data fields the agent draws on and recommend what to fill in so the personalization is grounded in fact.
Agents that save time inside the company: content written in your tone, plain-language answers pulled from reports, and an assistant for everyday CRM work. We deploy them wherever routine work is eating your team’s time.
We set the Content Agent up with your brand voice and rules for working with source material, so the output sounds like you. The team gets a workflow where AI drafts and a person approves.
The Data Agent answers questions against your CRM data: whoever asks gets a number and the context around it. We set which data the agent can access and how to verify the results.
Whichever agent you’re deploying, the deployment skeleton is the same. Three things decide whether an agent helps or hurts.
An audit of your existing documentation, a cleanup of outdated content, and a structure for the sources the agent is allowed to draw on. This includes deciding what the agent shouldn’t know.
We define the agent’s role and tone, the topics it avoids, and the moment it hands the conversation to a person. We test the rules against scenarios from your own operation before the agent ever meets its first customer.
We set up metrics (resolved conversations, handoffs, satisfaction), train your team to manage the agent, and agree a cadence for regular tuning. Ongoing agent operation can be handed to our HubSpot managed services.
your support or sales team handles repetitive questions at real volume
you have a knowledge base or documentation the agent can answer from, or you’re ready to build one
you’re on HubSpot Professional or Enterprise (Breeze agents require these license tiers)
you have a person who will own the agent: reading transcripts and approving changes
you want your team spending time on cases that need a person’s judgment, not on routine work
your query volume is low and every question is different, a person handles those faster and better
your source material doesn’t exist or is out of date, the first step is a knowledge base, not an agent
you’re expecting AI to fix a broken process, automated chaos is still chaos
the agent would need data from systems outside HubSpot, that calls for development and integrations first, then the agent
How many queries does your team handle a month, and what share of them repeat?
Is there documentation the agent could answer from, and is it current?
Who will check and tune the agent’s answers after launch?
Breeze agents require HubSpot Professional or Enterprise. Their work draws on credits; we’ll walk through how credits work and what your license covers together during use-case selection.
For a detailed rundown of what Breeze can do, see our HubSpot Breeze AI product page.
Deploying a single agent usually takes one to two weeks; multiple agents with workflow connections take two to three. What matters most is the state of your source material: the better your knowledge base, the faster the deployment.
Breeze supports Czech, Slovak, and English, among other languages, which matters if you’re serving a CEE audience. We verify the quality of answers in your specific languages during testing against real queries from your own operation, before the agent ever talks to a customer.
The agent answers only from the sources you approve during deployment: the knowledge base, selected website pages, documents. That’s exactly why preparing the source material is the core of the whole deployment. On top of that, we set guardrails: topics the agent avoids, and situations where it hands the conversation to a person.
It hands the conversation to a person, along with the full context. We set the handoff rules together: by topic, by how confident the answer is, and by what the customer wants. The goal is that a customer can always reach a person.
Yes, we’re rolling them out right now. We’re deploying the Breeze Customer Agent on our own customer support and the Prospecting Agent on our own sales team. We carry the lessons from our own operation, including what didn’t work, into every client deployment.
Scope depends on the number of agents, the state of your source material, and any workflow connections you need. We confirm scope and price together during scoping.
A good first step can be a free HubSpot audit, which checks how ready your portal is.
You can, Breeze is built as a no-code tool. Our work isn’t in the clicking, it’s in everything around it: use-case selection, source-material prep, guardrails, testing, and measurement. If you want to take it on yourselves, we’re happy to walk you through what to watch for on a consultation.
Start with a free audit: we’ll look at how you’re using your portal today, the state of your source material, and where an AI agent makes the most sense to start. Or skip ahead and walk us through a specific use case.