I lead the delivery of AI-assisted integration programmes, pairing agentic tooling with human approval gates, evals and audit trails. The approach is proven in live testing on a large regulated migration and built on a decade of enterprise SAP integration. The interesting problem isn't whether AI can do delivery work. It's whether you can govern the change.
About
I design and lead how enterprise teams adopt AI-assisted delivery without losing the discipline that regulated environments demand. In practice that means governed delivery workflows, with scoped tools, human approval gates, evals and tracing, applied to real programme delivery. The foundation underneath is a decade of enterprise integration across SAP BTP, Integration Suite and S/4HANA in complex, regulated landscapes, which is exactly where the governance instincts come from.
Governed AI delivery, as it actually runs
Inventory-driven tickets
Delivery starts as a plan rather than a prompt. The interface inventory drives the epics and stories, decomposed into detailed tickets that an AI (or a developer in another timezone) can build from without a clarification call.
AI drafts, humans promote
AI drafts the standard artefacts (iFlows, documentation and test assets) from approved templates. Consultants review, complete sensitive mappings and promote. Nothing ships on a machine's say-so.
Traceability & change control
Every artefact traces back to its ticket and every change goes through controlled review, so the programme can always answer what changed, when and why.
Plan-vs-actual reporting
Live plan-vs-actual tracking keeps the team, the PM and the client looking at the same picture. Progress is reported as evidence rather than status calls.
What I bring
- A toolchain I built and now run. Each accelerator below exists because it removed a real delivery bottleneck, not because it demos well.
- Turn discovery into repeatable delivery by standardising inventories, patterns and artefact packs so teams move faster with fewer missed requirements.
- Reduce production risk by designing for observability, safe reprocessing and controlled change, so handover and steady-state support remain straightforward.
- Accelerate build and testing safely through governed, prompt-assisted workflows (scoped tools, approval gates, traceability), keeping sensitive mapping and decisions under human control.
Accelerators & skills
How I deliver
Discover
Start with discovery to define boundaries, constraints and a workable target design, then baseline the interface inventory that becomes the single source for planning and AI-assisted scaffolding, while sensitive mapping stays under human control.
Design
Architecture is informed decision-making rather than diagram drawing. I work in layered views from business capability down to runtime, and record every pattern choice (API-led, event-driven, hub-and-spoke) as a trade-off against latency, coupling, scale and cost. The test of a design is whether the decision can still be defended six months later.
Scaffold
Scaffold the build through template-driven, tool-assisted delivery. AI drafts the standard scaffolds and artefacts, then consultants validate, complete sensitive mappings and promote into test through stage gates.
Approved tools only, least-privilege access, and allowlisted connectors by environment.
Structured inventories, constraints and canonical inputs shape what gets drafted, not free-form prompting.
AI proposes; consultants review and approve based on risk, keeping sensitive mapping and requirements human-handled.
Decision context, artefact lineage and reviewable change sets across iterations, with clear provenance of inputs and approvals.
Adoption, cost and stability checkpoints plus readiness gates before scaling the approach across teams and waves.
Validate
Define success criteria up front, then validate through repeatable test assets (mocks, harnesses, Postman collections/scripts) and contract checks so edge cases surface early. Treat prompts and templates as versioned artefacts with review gates, and shape the platform approach around reliability, supportability and governance, including regulated constraints such as data residency, operational controls and auditability.
Cutover & operate
Drive go-live readiness through repeatable evidence from build and validation, supporting UAT and production readiness without relying on fragile one-off scripting. Keep maintainability and support in mind so solutions can be handed over cleanly into steady-state operations. Operate with observability patterns, safe change and reprocessing, reusing the same mock services and Postman assets to support triage and ongoing reliability.
Highlights
A production platform built and operated for a real community, with real members, real money and a production AI support agent.
Wickham Wave
Bespoke club operations app used by 50+ members to manage attendance, team assignment, scheduling, billing and role-based workflows across web, iOS and Android, including a production AI support agent answering member questions over WhatsApp.
50+ active members weekly
Billing admin reduced to near-zero. Gamification driving attendance streaks and friendly competition.
Attendance, scheduling & teams
Waitlists, auto-promotion, recurring events, calendar sync and tactical auto assign teams.
WhatsApp agent, live on the club number
Answers member questions grounded on club data, with a human-escalation path.
Stripe-powered billing
Checkout, portal and webhooks with flexible pricing tiers.
Role-based access control
Webhook verification, security headers and edge-level CSP.
Awards, streaks & badges
Milestone tracking with featured profile badges and attendance competition.
Automated health checks
Scheduled monitoring pipelines with safe retries and at-a-glance system status.
Platforms and tools used across enterprise delivery and side builds