For a global mobility and transport technology company, Bloxx is a critical backend platform supporting parking experiences across mobile applications, web applications and on-street devices. Its functionality spans Orchestration, Parking Right, Rates Calculation, Inventory, Enforcement, Authentication, Eligibility and ParkingRight Reporting.
Bloxx connects customer-facing applications with street-based parking devices, making its reliability, security and performance important to the wider parking ecosystem. Persistent took ownership of an estate spanning 71 repositories, multiple APIs, applications and devices.
The transition demanded deep domain learning alongside continuing delivery. The team needed to reconcile outdated or redundant reference material with live system behaviour, modernise services running on different .NET versions and address several hundred security vulnerabilities. At the same time, it had to progress a backlog covering application maintenance, infrastructure, deployments, quality assurance and customer-driven change requests.
The immediate task was clear: establish accountable ownership before increasing delivery pace.
Taking Full-Cycle Ownership of Bloxx
Persistent combined structured knowledge transition, full-cycle engineering ownership and Claude Code-enabled delivery to strengthen the platform’s engineering foundation.
The team studied application behaviour, validated reference material and developed the domain understanding needed to support Bloxx across multiple cities and customer environments. It then assumed end-to-end responsibility for application support, defect resolution, health monitoring, change requests, feature development, security remediation, infrastructure and deployment support, quality assurance and penetration-test-related fixes.
Human engineers retained responsibility for architecture, design, testing, code review and release decisions throughout. Persistent also migrated Bloxx from legacy .NET versions to .NET 10, creating a more consistent technology foundation and reducing the risks associated with ageing frameworks. Vulnerabilities identified through SonarQube, GitHub CodeQL and other security-scanning tools were systematically remediated.
GitHub’s 2024 survey of 2,000 enterprise software-team respondents found that more than 97% had used AI coding tools at work at some point. Respondents linked those tools with development efficiency, code quality and easier onboarding. For Bloxx, AI was embedded within established controls rather than treated as a standalone experiment.
That foundation made it possible to introduce Claude Code without diluting engineering accountability.
Embedding Claude Code in a Human-Led Workflow
The client provided developers with access to Claude as part of its engineering enablement programme. Persistent University delivered a six-hour programme covering structured prompting, rules, skills and techniques for improving the quality and consistency of AI-assisted engineering work.
Claude Code supported .NET modernisation, security remediation, bug fixing, feature development, infrastructure changes, deployment activities, quality assurance and penetration-test-related changes. Engineers used it to analyse code, accelerate repeatable work and prepare changes for review.
The workflow remained human-led. Engineers first assessed requirements, current application behaviour, business rules and technical dependencies. Claude Code then supported analysis, implementation, migration and remediation. Every change continued through code review, functional testing, security scanning, quality assurance, pull-request controls and release approval.
McKinsey research found that generative AI can reduce the time needed for documentation and new-code tasks, while offering more limited gains for highly complex work. Its findings reinforce the importance of developer expertise, iteration and structured controls.
Persistent also delivered business and platform enhancements, including pollution-day parking across multiple consecutive days, automated recurring reporting and support for the revival of the Rates Calculation engine for downstream applications. The impact then became visible in engineering measures.
Business Impact
From Transition Risk to Higher Engineering Throughput
- Pull requests increased 92.5%, from 173 to 333 across the comparison periods.
- Non-merge work commits increased 113.1%, from 327 to 697.
- Median pull-request cycle time fell 41.5%, from 1.35 hours to 0.79 hours.
- Active repository coverage increased from 38 of 71 repositories to 67 of 71.
- Claude-assisted delivery saved approximately 254 engineering hours.
- Total lines changed increased from 42,133 to 176,616.
- Average change size increased from approximately 244 to 531 lines per pull request and from 129 to 253 lines per work commit.
- Support throughput rose from approximately 50 to more than 70 tickets per month with one fewer resource.
- Outstanding vulnerabilities identified through application security-scanning tools were resolved.
- A recurring business report moved from manual preparation and email coordination to automatic delivery to the relevant mailbox on the first day of each month.
These gains were achieved while standard review, testing and approval controls remained in place. Higher throughput came from clearer ownership, modernisation and disciplined AI-assisted engineering.
Extending the Model Beyond Bloxx
Following the success of the Bloxx transition, the client began exploring Persistent ownership of additional applications with similar support and modernisation needs. It is also evaluating expansion from Level 3 support into Level 1 and Level 2 support.
Persistent continues to identify opportunities around CI/CD migration to GitHub Actions, monitoring-tool rationalisation, further platform modernisation, security remediation and standardised engineering patterns. The engagement shows how an accountable AI-assisted approach can extend across multi-repository environments involving framework upgrades, defect resolution, infrastructure work, deployment support, quality assurance and application maintenance.
Persistent combined end-to-end ownership, specialised domain learning, modernisation expertise, systematic security remediation and measured Claude Code adoption. The result was stronger engineering coverage, faster pull-request cycles and a more durable foundation for long-term Bloxx ownership.
Assess Your Engineering Estate. Build an Accountable AI-Assisted Delivery Model. Talk with Persistent.
