An embedded dev pod, not a remote vendor

Halcrow provided a dedicated cross-functional team embedded directly inside Banktech's operations:

• 2 mobile developers covering iOS and Android natively

• 2 backend engineers handling the API layer and integrations

• 1 QA specialist focused on security testing and regulatory compliance

• 1 product lead translating requirements into real user stories alongside Banktech's own team

What made this work was proximity. Daily standups with Banktech stakeholders were real coordination on blockers and trade-offs, not status theatre. During the pilot, the team made weekly visits to the live venues, physically present to watch patrons use the app and catch confusion as it happened rather than waiting for a bug report. We had direct access into Banktech's payment infrastructure, moving from a sandbox environment to production access once security review cleared it, with no ticket-and-wait step in between. And Banktech's product team, the Halcrow dev pod, and venue operations all shared the same communication channels, so nothing got lost in translation between groups.

Build structure: a phased rollout, tested against real behaviour

Phase 1: MVP Build (February–June 2023)
Over a four-month sprint, we built native iOS and Android apps, a web platform for venue and admin use, the physical card-to-digital-balance linkage, identity verification for new users, payment gateway integration, and the foundational regulatory reporting. A few decisions here mattered more than they might look: we chose fully native apps over a cross-platform framework because performance genuinely mattered — a patron tapping their card expects an instant balance update, and any UI lag reads as "something's broken." We used a real-time connection for balance updates rather than periodic polling, because a patron needed to actually believe their money was moving the moment it moved. And because venue WiFi is often unreliable, the app was built offline-first, queuing transactions locally and syncing once connectivity returned, since a gaming machine transaction can't simply wait on a network round-trip. Financial security was built in from day one rather than retrofitted, with all financial data encrypted at rest and in transit, and identity verification handled through an established third-party provider rather than something built in-house from scratch.

Phase 2: Pilot Launch (July–October 2023)
Three pilot venues went live, and the most important lessons came directly from watching real patrons, not from a spec. The clearest one: patrons weren't reading instructions. They'd tap the card, see nothing happen because they hadn't opened the app first, and assume the whole thing was broken. We fixed this by changing the onboarding flow itself, forcing new users through a guided "tap your card now" step while the app is already open, so they see their balance update in real time and immediately understand how the system works. Onboarding drop-off fell from 40% to 12% as a direct result. Other fixes followed the same pattern of watching first and adjusting fast: the balance display needed to be three times larger so patrons could read it at arm's length, patrons wanted an explicit notification confirming exactly what they'd just spent, and the original three-tap cash-out flow got reduced to one tap plus a PIN, based directly on patron feedback that a multi-step process felt like an unnecessary barrier to getting their own money back.

We deliberately launched in just three venues rather than rolling out everywhere at once, and each one taught us something different — a second venue ran a different POS system that needed its own adapter, and a third had older gaming machines that needed a firmware update before the integration would work properly.

Roll Out in Small, Reversible Steps
We didn't launch in fifty venues simultaneously. Three venues first, learn, fix, then scale. If the onboarding flow had been broken at fifty venues from day one, the damage to adoption would have been broad and hard to undo. Limiting it to three venues for two weeks meant the problem stayed small and fixable before it ever had the chance to compound.

Phase 3: Scale Rollout (November 2023–August 2024)
By this stage the product had matured, and the work shifted from building new features to making the existing ones hold up at scale — supporting growth from roughly 10,000 to 50,000 patrons without any drop in performance, handling the small configuration differences between every individual venue's existing systems, automating compliance reporting across different state regulators, and managing end-of-day financial reconciliation across multiple venues, banks, and payment processors at once. We also migrated the underlying infrastructure to a more resilient setup that removed several single points of failure, and built a real-time monitoring dashboard for Banktech's own operations team, so that if transaction success rate dropped below 95% at any single venue, the right people were alerted immediately rather than finding out after the fact.