The Build That Won Us a 2026 HubSpot Impact Award

HubSpot has named Spin the APAC winner of the 2026 Impact Award for Technical Excellence.
We're proud of that, and we'll spare you the acceptance speech.

Award announcements are boring. The work that wins them usually isn't, so here's the story of this one: a bespoke insurance CRM bridged into HubSpot, a partner record rebuilt as a live 360-degree view, and an AI agent we built that did the engineering.

A sales team working blind

The client is an Australian insurance company with more than twenty years in the market. Its growth runs on a national partner network: independent retail partners who introduce their customers at the point of sale. That B2B2C channel delivers the overwhelming majority of new customers, at a fraction of the cost of direct acquisition. It is the backbone of the business.

The people responsible for that backbone spend most of their week on the road. And until recently, they were travelling blind. Open a partner's record and the questions that mattered had no answers. How many policies has this partner generated this quarter? Are their referrals trending up or quietly falling away? What share of their introductions actually convert? When were they last visited? Is this relationship healthy, slipping, or about to be poached?

None of it lived in HubSpot in any usable form. The policy and referral data sat in the client's bespoke, in-house CRM, the system of record for every policy. HubSpot, where the relationships were managed, had no connection to it. The two systems had never spoken.

The wall, and the unlock

There was a good reason they'd never spoken. As a regulated Australian financial services business, the client operates under strict data-handling expectations, and for years there were real limits on what policy data could defensibly sit in a third-party platform.

The unlock was infrastructure: HubSpot stood up an Australian data centre. With data residency onshore, bringing policy-derived data into HubSpot became defensible for the first time, and that opened the door to building something properly technical on top of it. So the brief became threefold. Bridge two systems that had never been connected. Design a CRM architecture rich enough to turn raw policy events into partner intelligence. And do it compliance-first, from the first line of code.

What we built

The foundation is a custom integration built directly against HubSpot's CRM API. The client's CRM pushes referrals and policies into HubSpot as they happen, and every record lands on exactly the right partner through a shared ID key. The integration behaves like production software because it is production software: it finds an existing deal before creating one so nothing duplicates, it backfills history through batch endpoints, it respects rate limits, and it names deals by policy number only, never by customer name, in line with ASIC requirements.

Raw deal data is only half the story. The value is in what happens once it arrives.

We rebuilt the partner record as a system we call 360 View. A partner's record is no longer a static profile. It is a layer of calculated properties that continuously translate deal activity into the numbers the sales team runs on: policies and referrals counted across five rolling time windows so year-on-year movement is visible at a glance, a conversion rate calculated on a rolling three-month cohort so one odd month can't distort the picture, average policy value, days since last referral, and a trend flag that marks each partner as growing, stable or declining.

Sitting on top of all of it is a single health score, zero to one hundred, surfaced as green, amber or red, built from six weighted inputs. It answers "how is this relationship doing?" in about three seconds.

Nobody types any of these numbers in. Ever. Every metric is calculated from the underlying deal data by an event-driven engine that recalculates the affected partner the moment a deal is created or changes stage, then writes the results straight back to the record. No nightly batch, no stale fields, no depending on someone remembering to update a spreadsheet. Calculated over manual, everywhere it was possible. That one design choice is what makes the system trustworthy enough to run a sales team on.

Who did the engineering

Here's the part that made this submission unusual, and we suspect the part that won it.

The HubSpot architecture, every custom property, the pipeline, the integration objects and the calculation logic, was deployed through the HubSpot API by a custom AI infrastructure agent that Spin designed, built and trained. Powered by Claude, operated from the command line, version-controlled in GitHub like any other software project. We didn't use AI to help a developer type faster. We built a piece of tooling and gave it the engineering job.

Training is the word that matters. The agent was grounded in HubSpot's full platform model, its objects, properties, associations, API behaviour and limits, and in deep context on the client: the partner model, the tier and visit logic, the compliance constraints of a regulated Australian financial services business. Platform expertise plus customer context is what let it produce correct, production-grade configuration instead of generic boilerplate.

Humans stayed exactly where humans belong. We did the design, the judgement and the compliance thinking, and we built the workflows and dashboards by hand in the HubSpot app, because those can't yet be created through the API. An honest division of labour: the agent did the repeatable engineering, people did the parts that require taste and accountability.

What it changed

The clearest signal is behaviour. Usage of the partner records has climbed sharply as the team has come to trust the numbers, and when a sales team starts opening a record before every call and every visit, that's the whole game. For the first time they can see the entire network's health from one screen: who to grow, who to scale, who to recover. Decisions that used to run on gut feel now run on live data that maintains itself.

And it's still growing. 360 View is live, in daily use, and being extended as the client's team finds more value in it, which is exactly how the best technical work earns its keep.

The award is HubSpot's verdict that this is the best technical build in the region this round. We'll take it. But the thing we're most proud of is quieter: a new way of delivering CRM engineering, where a trained agent does the heavy lifting and people do the thinking. This build was the proof it works on a real, regulated, multi-system problem.