Halcrow provided a cross-functional team embedded directly in Bovara's product development:

Team structure:

• 1 hardware sourcing specialist (supply chain, OCPP standards, controller compatibility)

• 2 backend engineers (charging management system, payment integration, session handling)

• 2 mobile developers (iOS + Android native apps)

• 1 DevOps engineer (cloud infrastructure, monitoring, real-time data streaming)

Critical embedding elements:

• Direct access to hardware suppliers: We leveraged existing relationships with charging controller manufacturers. No "request for quote" theatre—we knew which suppliers could deliver OCPP-compliant units at scale.

• Shared technical infrastructure: Bovara gave us direct AWS access, payment gateway credentials, app store accounts. No "submit a ticket to deploy." We moved at the speed decisions could be made, not the speed bureaucracy allowed.

Law 3 (Direct Infrastructure Access): We weren't integrating via formal APIs at arm's length. We worked inside Bovara's technical infrastructure. When something needed testing with real hardware, we had charging controllers on-site for immediate validation.

Build Structure: Layered Development with Proven Foundations

Phase 1: Hardware Foundation (Month 1 - June 2023)

Objective: Lock in hardware supply, validate OCPP compatibility.

Problem: Charging stations are commodity hardware in markets like the US and EU. In Australia in 2023, local supply was immature. Most manufacturers were overseas. Lead times were long. Quality was inconsistent.

Solution: Leveraged existing supplier relationships to identify manufacturers with:

• OCPP 1.6 compliance (industry standard protocol for charging station communication)

• Proven track record deploying in Australia/New Zealand

• Reasonable lead times (8-12 weeks, not 6+ months)

• Responsive support (when hardware fails in the field, you need fast replacements)

Result: Secured supplier agreement by end of June. First batch of 20 charging controllers ordered. Hardware de-risked before writing a single line of code.

Law 9 (Incremental Exposure Reduction): We didn't order 200 controllers upfront. First batch: 20 units. Validate in pilot deployment. Then scale. If hardware failed, we're out $15K, not $150K.

Phase 2: Software Foundation (Months 2-3 - July-August 2023)

Objective: Working backend charging management system, mobile app prototype.

Critical architectural decision: Don't build charging management from scratch.

Charging management systems handle:

• Station connectivity (track which stations are online/offline)

• Session management (start/stop charging, track energy delivered)

• Payment processing (pre-authorization, final billing)

• Load balancing (distribute power across multiple chargers at same location)

• Fault handling (what happens when station loses power mid-session?)

This is solved infrastructure. Open-source projects like STEVE (OCPP server) and Open Charge Map already handle this. Rebuilding from scratch would take 6-9 months and introduce risk.

Our approach:

• Deployed open-source OCPP server (handles station communication, session management)

• Built custom API layer on top (consumer-facing endpoints for mobile app)

• Focused custom development on: user authentication, payment integration (Stripe), mobile app UI/UX

Time saved: At least 2 months. We didn't spend July-August building charging session logic. We spent it building the consumer experience layer while the open-source foundation handled the hard infrastructure problems.

Law 10 (Avoid Multi-Front Wars): Don't fight the "build a charging management system" war and the "build a great mobile app" war simultaneously. Use proven infrastructure for the first, focus energy on the second.

Phase 3: Mobile App Development (Months 2-4 - July-September 2023)

Objective: iOS and Android apps that drivers actually use.

Core features:

• Station map (find nearby chargers)

• Session start (tap to begin charging)

• Real-time progress (see charging status, energy delivered, estimated completion time)

• Payment (pre-authorize card, charge when session completes)

• History (past charging sessions, invoices)

The notification dependency problem emerged in August:

Early beta testers disabled push notifications (common privacy behavior). When they did, charging progress updates stopped displaying. The app relied on push notifications to wake up and fetch new data. No notifications = no updates = terrible UX.

Root cause: Mobile operating systems (iOS/Android) aggressively suspend background processes to save battery. Apps can't continuously poll for updates in the background. Most developers solve this with push notifications—server sends notification, OS wakes app, app fetches new data.

But if users disable notifications (which many do), this pattern breaks completely.

Fix: Decoupled charging updates from push notifications.

Implemented gRPC streaming—a real-time communication protocol where the server can push data to the client without relying on OS notification infrastructure. When the app is open, it maintains an active connection to the server. Server pushes charging progress updates directly. No notification dependency.

Result: Charging progress updates work reliably whether notifications are enabled or not. User can open app anytime during charging session and see real-time status.

Phase 4: Pilot Deployment (Month 5 - October 2023)

Objective: Deploy 10 charging stations in Sydney metro, test with real drivers.

Pilot location strategy:

• 3 stations at shopping centers (high-dwell-time locations—people shop for 1-2 hours, perfect charging window)

• 4 stations at workplace parking (commuters charge during work day)

• 3 stations at apartment complexes (overnight charging for residents)

What we learned from pilot:

• Session start latency: Some stations took 8-12 seconds to respond to "start charging" command. Felt slow. Root cause: OCPP message queue backlog when multiple cars tried to start simultaneously. Fixed by implementing connection pooling and load balancing.

• Payment pre-authorization: Users expected to know max charge upfront. We pre-authorized $50 on their card, charged actual amount when session ended. Clear communication in app: "We'll hold $50, you'll only be charged for what you use."

• Charging interruptions: If station lost power or network connection mid-session, session ended abruptly with no user notification. Implemented "graceful termination" logic—if session ends unexpectedly, app notifies user, still charges for energy delivered up to that point.

Phase 5: Public Launch (November 2023)

Objective: Expand from 10 pilot stations to 30+ stations across Sydney, Melbourne, Brisbane.

By November, the platform was stable:

• Hardware supply chain validated (controllers worked reliably, supplier could scale)

• Backend infrastructure proven (open-source OCPP server handled hundreds of sessions without issues)

• Mobile app refined (real-time updates worked, payment flow clear, session management intuitive)