For years, HubSpot has focused on making our platform the best place for marketing, sales, and service teams to get their work done. With AI, we’ve shifted toward having the platform do more of that work for them – using agents that qualify leads, resolve tickets, rescue deals, and drive results across the business. That’s why we describe HubSpot as an agentic customer platform. But agents don’t click around dashboards or browse interfaces; they call APIs, parse structured outputs, and take action. Software designed for humans now has to evolve so it’s truly usable by agents as well.
Access by itself isn’t sufficient, though. Agents also need depth. An agent looking only at raw records can’t tell what’s typical for a given business, or what’s proven effective for hundreds of thousands of similar companies. As we recently wrote, the real AI race isn’t about models or data alone; it’s about context. That belief guides everything we build. It’s why we were early to release an MCP server, and why we’ve continually broadened what agents can read, write, and execute since then.
And that was just the start. Our broader vision is this: Agents can run on HubSpot. And agents can run HubSpot. Running on HubSpot means any agent – whether built by us or by others – can plug into HubSpot’s data, context, and capabilities as core building blocks. Running HubSpot means agents can operate the platform end-to-end through our APIs, MCP server, CLI, and whatever new access methods emerge next.
What we’re enabling so agents can run…