As a member of HubSpot's AI Partner Advisory Council, I've had the privilege of getting an early look at Breeze Studio—their next-generation AI agents platform. After spending time in the closed beta, I can confidently say this represents something genuinely transformative: a logical application of AI that feels firmly embedded into the product rather than bolted on as an afterthought.
But here's the tricky reality: whilst the technology is impressive, the success of AI agents ultimately comes down to user adoption. And that's where things get interesting—and challenging.
The Adoption Hurdle: Why Quality Matters More Than Features
The most critical insight from my beta experience isn't about the technology itself—it's about human behaviour. Users will need time to hand over control to these agents, and as we've seen with ChatGPT, Claude, and other AI experiences, it only takes a couple of poor-quality interactions for people to abandon the technology entirely.
This isn't unique to HubSpot. It's the fundamental challenge facing every AI agent platform today. The difference is in how HubSpot is approaching the problem.
Two Flavours of AI: Assistants vs Agents
HubSpot has created two distinct types of AI tools, each serving different needs:
Breeze Assistants function much like Custom GPTs or Copilot Agents—you provide custom instructions, give them access to tools, and interact through a familiar chat interface. The crucial difference is that these assistants have access to both files and CRM objects, including contacts, companies, deals, tickets, knowledge articles, blog posts, landing pages, and calls.
This organisational context is huge. Instead of generic AI responses, you get outputs that understand your business, your customers, and your processes.
Breeze Agents operate differently. These are pre-configured, autonomous task performers. You assign them a job—researching prospects, handling customer handoffs between teams, writing blog content—and they work in the background. No back-and-forth chat required. They complete their tasks, often taking several minutes or more, and notify you when finished.
The notification model represents a fundamental shift from AI as a conversational tool to AI as a background worker.
Solving the Three Big Questions
HubSpot have identified three critical questions that customers ask about AI agents:
-
What do I use AI for? With infinite possible use cases and flashy demos everywhere, knowing where to start becomes paralysing.
-
How do I make it good? There's a vast gap between basic AI outputs and genuinely useful ones. Most organisations struggle to bridge this gap.
-
How does AI actually change my business? Too many AI implementations become what HubSpot calls "copy-paste Olympics"—moving content around manually without genuine automation or business impact.
HubSpot's approach addresses these questions by focusing on two key vectors: connection to your business and ease of use. Rather than trying to be the most powerful agent platform, they're positioning at the intersection of what's deeply integrated with your business context and what's simple enough for anyone to deploy.
The Framework That Changes Everything
What struck me most during the beta was HubSpot's agent framework. They've created a system where building effective agents has gone from taking days to taking hours—sometimes under 60 minutes.
Every agent is built from three components:
- Instructions: The core prompt that defines behaviour and outputs
- Tools: Capabilities that let agents find data, write data, or take actions
- Knowledge: Proprietary business context from files, CRM records, or uploaded documents
This framework scales across multiple agents and enables both HubSpot's product teams and customers to build powerful integrations rapidly.
Agents can be automated based on business events. Instead of remembering to run research on a new prospect, the agent triggers automatically when someone submits a form. Instead of manually creating handoff reports when deals close, the agent generates them based on property changes. That's POWERFUL!
This moves AI from human-initiated tasks to event-driven automation—a fundamental shift in how work gets done.
The Marketplace Opportunity
Currently, HubSpot's agent marketplace features a handful of tools, but I predict this will expand rapidly. The foundation is already there with integrations like Zoom, Shopify, and Jira. What's coming next will likely be driven by MCP (Model Context Protocol) integrations, creating a rich ecosystem of third-party capabilities.
The speed at which effective agents can now be built means we'll likely see an explosion of both HubSpot-created and partner-developed tools. The framework makes it possible to create functional agents that solve real business problems, not just impressive demos.
Why This Feels Different
After years of watching AI implementations in business software, what makes Breeze Studio noteworthy isn't just the technology—it's the approach. HubSpot isn't trying to recreate the wheel with a completely new interface or workflow. They're embedding AI deeply into existing business processes and CRM workflows.
The agents know your data. They understand your business context. They can take action within your existing systems. Most importantly, they're designed to work autonomously based on the business events that actually matter to your organisation.
During the preview, one HubSpot team member noted something that resonated: in the future, users might interact with 10% of the software they use today but accomplish 10 times more work. The complexity gets abstracted away by intelligent agents whilst the actual business outcomes multiply.
The Road Ahead
The beta experience has been largely positive, but the real test comes with broader adoption. HubSpot is taking a measured approach, starting with customisable pre-built agents before opening up fully custom agent creation.
For partners and solution providers, this represents a significant opportunity. The ability to create agents that combine proprietary knowledge with HubSpot's CRM context could fundamentally change how services are delivered and scaled.
But success isn't guaranteed. The quality of early experiences will determine whether businesses embrace this technology or abandon it after a few disappointing interactions.
The Bottom Line
Breeze Studio represents a logical evolution of AI in business software—not flashy demos, but practical automation that's embedded into real workflows. The technology works, the framework scales, and the business context provides genuine differentiation.
For businesses evaluating AI agents, the question isn't whether this technology will change how work gets done—it's whether platforms like Breeze Studio can deliver that change reliably enough to earn user trust and adoption.
Based on my beta experience, I'm optimistic. The pieces are in place for something genuinely transformative, but as with any AI implementation, the quality of execution will determine success or failure.
Comments