Two years after Generative AI took off, one thing is clear: AI will transform the way businesses operate. But for most small and medium businesses, the AI landscape feels overwhelming. They don’t need the hype; they need practical help. The reality is that 50% of small businesses fail within five years. Not because starting a business is hard, but because scaling one is brutal. When you’re caught between limited resources and ambitious growth goals, you need tools that deliver actual results, not more complexity. That’s what we’re building with Breeze — AI designed specifically for scaling companies, built directly into HubSpot’s customer platform. In this post, we’ll take you behind the scenes of how we’re building Breeze, the most common use cases, and how it’s already helping real businesses drive results. The Breeze Tech Stack: Building for the AI-First Era Software is evolving. What started as desktop programs in the 1990s and moved to cloud-based SaaS in the 2000s is now entering a new era: software that doesn’t just serve, but works toward outcomes. This is the shift from “software as a service” to “results as a service.” We’re not building AI because it’s the latest tech trend — we’re using AI to deliver what customers are actually after: more leads, faster sales cycles, and better customer satisfaction. With Breeze, scaling companies can quickly and easily get value from AI connected to their customer data. We’re distinguishing Breeze in three ways to enable HubSpot customers to get the most out of AI: Our platform unifies structured data (like CRM records) and unstructured data (such as emails, calls, and transcripts) with external data like buyer intent and company news to give Breeze the full picture. Our “all-on-one” approach provides context across the customer journey, unlike point solution AI agents that fall short without a full understanding of the GTM. Our agent ecosystem connects humans with AI agents to extend their teams — some built by HubSpot, others by partners and citizen developers — all seamlessly connected to customer data to solve any need. The Four Layers of Breeze We’re building Breeze in four layers with data and context as the foundation. It’s something HubSpot is uniquely positioned to provide as the source of truth for nearly 250,000 companies with millions of users. From there, we help customers make sense of that data, customize and extend Breeze, and use it to power their GTM — all without having to become AI experts themselves. 1. Context Layer: Unified Data for Powerful AI AI is only as good as the data that powers it. Imagine asking your AI assistant to help with a customer issue, but it can‘t see their recent emails, doesn’t know about their last support call, and has no idea about the new product update that might solve their problem. That‘s the reality of most AI tools today — they’re smart, but they’re working blindfolded. Providing complete customer context is essential to AI delivering on its full promise, and this is what makes HubSpot stand out from other tools in the market. Only HubSpot has the depth and breadth of data across the customer journey, and can unify it to train our AI systems. That’s what makes Breeze’s Context Layer so unique: Breeze Knowledge Models: Not one, but multiple specialized models that pull all of your data into one clear picture of your customer, their interests, objections, and more. We combine the latest foundation models — from OpenAI, Stability AI, Google, Anthropic, and others — with our own custom models and deep understanding of the GTM for scaling businesses, so you get the best of AI without worrying about which model is best this month. Structured Data: This has always been our strength — customer records, company information, contact data, and all the rich CRM information that HubSpot has organized for over 15 years. Unstructured Data: This represents up to 80% of a business’s data that, until now, has been trapped, unable to be used. Think emails, call transcripts, meeting recordings, conversations. Our acquisition of Frame.ai in 2024 allows us to use this data for the first time, a capability lacking in most solutions on the market today. External Data: Think of this as all the data that exists out in the world, beyond your company. We’re able to perceive this external data across text, audio, and video, including over 200 million buyer and company profiles, information from earnings releases, funding rounds, and other relevant news posted across the internet. When you’re building a business, you don’t have time to hunt through multiple systems or keep up with AI developments. We absorb that complexity and bring together your internal and external data so Breeze has the complete picture. 2. Intelligence Layer: The Brain of Breeze Our Intelligence Layer is the cognitive engine that processes your data into insights — it’s the “brain” of Breeze. While nascent today, we’re thinking about this in three ways to make Breeze even more powerful in the future: Reasoning: Breeze won’t just look up answers — it will think things through, connecting different pieces of information to work out solutions, like your best employees tackling tough problems. Memory: Unlike basic chatbots that forget conversations the moment they’re over, Breeze remembers. This is the next evolution in AI personalization and quality. Breeze will learn from previous conversations and remember key user facts and preferences that can be used across all of their interactions. No more repeating yourself. Prediction: Instead of just reacting to problems, Breeze spots patterns and anticipates what might happen next through predictive analytics. For example, instead of finding out a client may not renew 30 days before expiration, Breeze flags concerning signals (decreasing usage, logins, fewer support tickets) months before the contract ends. You shouldn’t need a massive data science team to turn raw business data into insights you can act on. With Breeze, every business will be able to reason through problems, remember every interaction, and predict what comes next. 3. Integration