Resources Blog

Breeze AI: What Works, What Needs Improvement, & What HubSpot Users Are Saying

Written by Adam Sharrow | Apr 8, 2026 3:00:01 PM

From the April 8th, 2026, edition of How Teams Work

A few weeks ago, my co-founder Andrew was at HubSpot House event during SXSW.

He expected a lot of presentations about AI, and that part didn’t disappoint.

But what stood out most were some of the discussions that came up during the Q&A.

Because the questions that people were asking weren’t about what HubSpot’s AI tools are or what they can do.

It seems like we’re all past that.

The questions sounded more like:

  • “We tried this tool already… how has it improved?”
  • “Why would I use this instead of ChatGPT or Claude?”
  • “What’s actually worth using in HubSpot vs outside tools?”

We talked about how these questions signify a shift.

These aren't beginner questions.

These are questions from people (and teams) who have been experimenting, hit some friction, and are now trying to make real decisions about where AI actually fits into their workflow.

The Shift

Most of the teams we talk to and work with have already tried HubSpot’s Breeze tools when they were first introduced.

They may have played around with Breeze Copilot to analyze CRM data or tried to generate a blog in Content Hub. The feedback we often hear is that the experience is underwhelming due to:

  • The outputs feeling generic or inaccurate.
  • The tool lacking features or customization.
  • The tools simply not (yet) working for their intended use case.

Many of them defaulted back to the tools that worked before.

ChatGPT. Claude. Gemini. External tools that felt more flexible and easier to rely on.

But now it feels like that conversation is shifting again.

What we’re hearing, both from that event at SXSW and in our own conversations, is that HubSpot users are revisiting its AI features due to seeing genuine improvements.

Now, they want to figure out what's actually worth using today.



What People Are Actually Asking

Teams are no longer starting from zero when it comes to AI. When deciding on what tools to use, they are working off of real experience.

The questions that teams are asking are no longer beginner questions.

They’re trying to figure out what’s actually worth using and what’s not.

When investigating HubSpot’s suite of AI tools, it’s common to hear questions like:

  • “Which of the Breeze tools are actually worth using?”
  • “Why should I use this instead of ChatGPT?”
  • “Are HubSpot credits worth paying for?”
  • “Are the Breeze agents ready, or is it still early?”
  • “Can I control what data sources Breeze is pulling from?”
  • “Can I exclude sensitive data properties?”

These questions signal that the market is evolving and people are becoming more AI savvy.

The Core Tension

There isn’t a single tool that does everything well. Instead, you’re choosing between tools that are optimized for very different things.

We’re also noticing this tension between using HubSpot’s tools (where your data lives) or other tools that might be more powerful and flexible.

That’s what we see teams trying to figure out:

Breeze AI:

  • Is directly embedded in your HubSpot
  • Has access to your CRM data
  • Allows you to work without leaving the platform
  • Can be tied directly into existing processes and automations

External tools like ChatGPT, Claude, and Gemini:

  • Are more flexible and feature-rich
  • Are better at open-ended tasks
  • Give you more control over prompts and outputs
  • Are easier to experiment with for different use cases

Those differences become evident pretty quickly once teams start using both.

So the tradeoff becomes:

  • Context vs flexibility
  • Integration vs capability

Breeze is strongest when context matters. When the desired output depends heavily on your HubSpot data, it has a clear advantage.

External tools are stronger when flexibility matters. When you’re looking for more creative, customized solutions, they tend to perform better.

So the real question becomes:

Which tools are the best fit for the specific use cases inside your process?

Where Things Break Down

From what I’ve seen, most of the friction comes from early experiences with specific tools in HubSpot.

A lot of teams tried early versions of the Breeze Copilot, blog generator, or the AI report builder, expecting them to save time during day-to-day work. Instead, it felt like many of the tools:

  • Did not have enough features or customization to be useful
  • Had outputs that felt too generic or inaccurate
  • Were buggy or simply didn’t work well

Even things like AI summaries on records, which should be straightforward, weren’t always consistent enough to rely on.

Connectors with third party LLMs didn’t always produce clean or usable outputs, especially when pulling from multiple sources.

Now with Breeze agents, there’s a new layer of hesitation.

  • Letting something draft is one thing.
  • Letting something take action is another.

This hesitation is especially true for companies in highly regulated industries like financial services or healthcare that work with sensitive data. Even if that data is stored in sensitive data properties and not used by AI, there’s still hesitation around how these tools operate within the system.

All of that experience and friction adds up.

This is why many teams have been hesitant to jump back into testing HubSpot’s AI tools.

Why It’s Not Working (for Some Teams)

A lot of the frustration we hear isn't actually about the tools. It's about what's feeding into them.

AI outputs will only be as good as the data and structure in your CRM. And for a lot of teams, that foundation is not as solid as they think.

There are usually common trends that indicate that you don’t have the right data, processes, and people in place to use AI tools effectively:

  • If your data is inaccurate, the output will be too
  • If your properties are built on inconsistent definitions, results won’t hold up
  • If your team isn’t following consistent processes, tools will be difficult to adopt

This is the same pattern that shows up across your system.

The tool reflects how your team actually works.

If your HubSpot data is messy, your AI outputs will be messy.

If your team doesn't have defined processes, the AI has nothing to follow.

It won’t fix broken systems. It will amplify them.

What We’re Seeing

Across the teams we work with who are adopting AI tools, I see two types of users.

1. Interested but hesitant

The first type of user usually has tried introductory tools like Breeze Copilot but they didn’t see enough value to rely on it day to day.

So they default back to ChatGPT or Claude for anything important.

Their main concerns are usually:

  • “I don’t trust it to send emails or update records”
  • “I’m not sure what data it’s using”
  • “This feels worse than what I can do manually”

There’s interest, but not enough confidence to build it into their process.

2. Active but overwhelmed

On the other side, some teams are actively using multiple different AI tools for different use cases.

They’re jumping between HubSpot and external tools like ChatGPT or Claude, depending on the task.

But there’s no clear system for when to use what.

The result:

  • inconsistent usage across the team
  • no repeatable process
  • difficulties scaling what’s working

In both cases, the issue isn’t access to AI.

It’s deciding what to use, where it fits, and how it should be used consistently.

What Actually Works

HubSpot’s AI tools have changed a lot, and if you haven’t checked them out in a while, it’s a great time to revisit.

There have been meaningful improvements over the past year, and there are areas where we’re seeing these tools become genuinely useful in day-to-day work.

The teams seeing the most success aren’t trying to use every AI tool in HubSpot.

They’re focusing on specific areas where it clearly complements the work they’re already doing.

Use case: Working inside the CRM with real context

  • Tool: Breeze Copilot

  • Outcome: Faster execution of day-to-day work directly inside HubSpot

  • Insight: Early versions of Copilot were extremely limited, but it’s expanded significantly. It now works contextually with the record or asset you’re viewing and can summarize records, create records, build workflows, generate reports, and answer questions about HubSpot features.

Use case: Scaling outreach and support with AI agents

  • Tool: Breeze Prospecting Agent and Customer Agent

  • Outcome: Assisting with lead outreach and customer engagement

  • Insight: These agents are designed to handle specific, repeatable interactions like outbound prospecting or responding to common customer inquiries. The value comes from automating simple interactions so your team can focus on work that’s important.

Use case: Triggering AI actions inside workflows

  • Tool: AI workflow actions

  • Outcome: More dynamic automation driven by data and context

  • Insight: AI workflow actions allow you to trigger things like data enrichment, run agents, or pass data to third-party LLMs as part of a workflow. These are most effective when built into structured processes, where AI is enhancing specific steps rather than replacing the entire workflow.

Use case: Using HubSpot and external AI tools together

  • Tool: HubSpot connectors for ChatGPT, Claude, and Gemini

  • Outcome: External AI tools can analyze and act on real CRM data

  • Insight: These connectors allow third-party LLMs to work directly with your HubSpot data, instead of relying on generic inputs. That shows up in things like deeper analysis, more relevant outputs, and the ability to use these tools inside HubSpot workflows. Teams can bring the flexibility of external AI into their actual processes, rather than keeping it separate from the system.

Where This Is Going

Here’s a good rule of thumb when testing AI tools:

If you try to do something with an AI tool and it doesn't work, add a calendar reminder for six months from now to try again.

These tools are advancing at such a rapid pace that it’s important to circle back to test and monitor what may not have previously worked.

At this point, it’s expected that HubSpot will continue to expand and enhance its AI capabilities across the platform.

We’re already seeing that play out with more agents, deeper workflow integration, stronger data enrichment, and more ways to connect external tools.

That’s not slowing down.

What’s changing is how teams are approaching it.

Instead of trying to adopt everything at once, teams are testing new tools as they’re released and figuring out where they actually fit into their day-to-day work.

The teams getting the most value will be the ones that stay focused on where it actually improves how they work, not just what’s new.

We'll keep sharing what we're seeing actually work, and what doesn't, as this evolves.

Want to get the "How Teams Work" newsletter sent to your inbox? Subscribe Now.