How a hidden operational bottleneck became a software-led advantage — and what leaders can reuse from the process.
01 · Business opportunity discovery
The initial question was whether CILE could internalise selected road-opening work and create durable value. The software emerged only after the operating model exposed a hidden constraint.
The business case combined field operations, staffing, compliance, workflow design and financial viability. The breakthrough was not a new machine. It was identifying the administrative complexity that would otherwise destroy scalability.
Internalise a repeatable field activity to gain control, resilience and economic value.
Each municipality and police zone used different deadlines, channels, documents and contacts.
The administrative burden threatened profitability, training time and operational reliability.
Encode the procedures once, surface the right rule at the right moment, and generate the operational outputs automatically.
02 · What was built
The product hides procedural complexity behind a single operator path. Click through the workflow to see what each screen changes operationally.
Screens are drawn from the project documentation. Public access to the operational prototype is not available.
What the operator sees
One structured form for the address, dates, work type and operational context.
Business effect
A single entry point removes training friction and standardises the minimum data required.
What the operator sees
The address is geocoded and the relevant municipality is identified automatically.
Business effect
The system removes manual research and reduces the risk of following the wrong procedure.
What the operator sees
Deadlines, submission channels, contacts and documentary requirements appear contextually.
Business effect
Procedural knowledge becomes an organisational asset instead of remaining in one person’s memory.
What the operator sees
Road signs are placed directly on a mapped or imported field view.
Business effect
Preparation moves into the same workflow, reducing tool switching and coordination gaps.
What the operator sees
The blocked road segment and proposed diversion are calculated and visualised.
Business effect
A complex conditional task becomes repeatable and easier to review before submission.
What the operator sees
The document package, email content and field handoff are assembled automatically.
Business effect
The output is submission-ready, traceable and consistent across municipalities.
03 · Human × Claude co-development
Claude did not replace domain work or decision-making. The value came from combining human ownership with a reasoning model and an agentic coding environment.
Owns the business problem and the consequences.
Acts as a reasoning and architecture partner.
Operates in the codebase and compresses iteration cycles.
04 · Measured results
Minutes required to prepare a standard permit workflow, based on the detailed before-and-after process analysis.
Projected cumulative value over ten years from lower administrative effort and avoided failure costs.
Procedural knowledge, document production and field coordination were brought into one repeatable workflow.
05 · What leaders can reuse
The most useful lesson is not how to reproduce this exact software. It is how to identify a constrained value pool and turn operational knowledge into a maintainable system.
Look for a process where demand, economics or strategic intent are attractive, but one recurring bottleneck prevents scale. That constraint is often a better software brief than a list of desired features.
Model roles, handoffs, exceptions, timing and failure modes before automating. Claude becomes dramatically more useful when the operating reality is explicit.
Store changing facts as data, deterministic requirements as rules, and ambiguous decisions as explicit human judgment. This separation makes the system easier to maintain and audit.
Give Claude the problem frame, source material, constraints, acceptance tests and permission boundaries. Ask it to plan before changing files, then verify outputs with observable evidence.
A prototype creates value only when someone owns the data, rule updates, hosting, security, testing, change control and user adoption. Production readiness is an operating-model decision.
06 · From prototype to production
AI can accelerate implementation. It does not remove the need to make explicit decisions about ownership, risk and the future operating model.
What data enters the system, who owns it, how is it versioned, and what may be shared with a model provider?
Is a static client-side application sufficient, or do multi-user workflows require an API, database and identity layer?
What are the trust boundaries, permissions, external dependencies, fallback paths and audit requirements?
Are personal, contractual or regulated data involved? Which processing, retention and accountability rules apply?
Which rules are business-critical, what evidence proves correctness, and what happens when an external service fails?
Who updates the rules, supports users, trains new operators and decides when the product should evolve?
Who am I
Business engineering first, then a second Master's in industrial engineering. Three years of professional projects in strategy, applied AI and finance in parallel. Open to opportunities in strategy, finance or corporate development.