Case study · HubSpot
Reporting used to live only inside HubSpot's UI. I launched the first public version of the reporting API and moved reports and dashboards onto HubSpot's object data model, so reporting can be queried across the product and by AI agents.
A first for HubSpot
v1
The first public version of HubSpot's reporting API
Objects
reports and dashboards moved onto the data model
MCP
reporting opened to AI agent workflows
COMPANY
HubSpot
MY ROLE
Senior Product Manager
TYPE
Platform and API
STATUS
Phases 1 and 2 live
THE CONTEXT
Why did reporting need a public API?
Teams across HubSpot were working toward a headless HubSpot, one that could run efficiently without the UI. For reporting, that had led to three different ways of supporting reports, and all of them leaned on private internal APIs that were never meant for outside use.
What was I responsible for?
Move fast to launch HubSpot's first public reporting API: as close to 1:1 with our internal flows as possible, and built from the start for non-human users, including setup.
01 · HOW IT CAME TOGETHER
This wasn't a feature request. It was a platform gap we knew we had to fill.
Phase 1
WHY NOW?
→ Teams across HubSpot were working toward a headless HubSpot, run without the UI.
→ For reporting, that had produced three different ways of supporting reports.
→ All three leaned on private internal APIs that were never meant for outside use.
Phase 2
AS CLOSE TO 1:1 AS POSSIBLE
→ Our internal API was robust: creation, governance, management and sharing.
→ But it was messy, built for internal use where “it works” was good enough.
→ We mirrored the internal flows as closely as possible in a clean public API.
→ It was designed from the start for non-human users, including setup.
→ We tested with internal users and a handful of early developers, then moved quickly into public beta to make HubSpot's Fall 2026 Spotlight at UNBOUND.
Where AI helped: creating test apps, writing and reading docs, and running test scenarios.
Phase 3
ONE PHASE AT A TIME
→ The goal: most of competitor parity in the first launch, then build from there.
→ Phase 1, live: report and dashboard management.
→ Phase 2, live: secondary features like scheduling and tags.
→ Phase 3: true report creation, planned by March 2027. The blocker is internal: our teams use competing agent-ready SQL dialects to create reports.
→ Built to be used by agents: HubSpot's own AI agent tools now call the public API instead of forced workarounds and messy implementations.
Where AI helped: deciding what went in each phase. We reviewed 25+ direct and indirect competitors, and what their users say about them.
02 · THE KEY DECISION
IT WASN'T
“Keep patching three separate workarounds built on private APIs.”
IT WAS
“One public API that matches our internal flows, built for agents as well as people.”
Speed mattered. Our internal API already handled creation, governance, management and sharing, so we mirrored those flows and cleaned them up for outside use instead of starting over.
03 · THE TRADE-OFF
✓ In v1
Phase 1: report and dashboard management through the API.
→ Left for later
Phase 2, now live: secondary features like scheduling and tags. Phase 3: true report creation through the API, planned by March 2027 once we resolve competing agent-ready SQL dialects internally.
My role: I owned the API launch and the move of reports and dashboards onto the object data model, and my team did most of the heavy lifting. I partnered with the Developer Experience team on comms and feedback, and the API team on compliance and launch procedures.
04 · THE OUTCOME
v1
HubSpot's first public reporting API
MCP
reporting now reachable by MCP connectors and AI agents
2 of 3
API phases live, with report creation coming by March 2027
05 · WHAT TO TAKE FROM IT
1
An API is only as good as the model under it. Fixing the foundation first made every later integration easier.
2
Build for agents from day one. HubSpot's own AI tools now call the same public API developers use, instead of working around private ones.
3
The fastest path to a public API was the internal one people already relied on. Clean it up instead of starting from scratch.
Sources: My work at HubSpot, and the Reporting API guide (HubSpot Developer Docs).
Your turn.
I mentor career switchers from CS, support, marketing and analytics. Start with a free 30-min clarity call. We'll map your story and your next step. No sales pitch.