Can it work?
Architecture tests the data, integrations, operating constraints and implementation path.
How I work
The same four phases apply whether the system is an ERP, an operational platform or an AI workflow. Timing depends on the work.
Four phases
Phase 1
I start by naming the operational problem, who it affects and why it matters now. I talk to the people doing the work and look at the tools and data they use today. Operational reality becomes clear, testable requirements. We decide what to build first and what can wait.
Phase 2
I design the data, integrations and workflows before code begins. Screens are shaped around the tasks people repeat every day.
Phase 3
The application, integrations and business logic are built and tested. Where it helps, AI is added with a human review step.
Phase 4
The system is released with care for existing data and continuity. The team receives a system they can run, with documentation and handover.
Behind the method
Architecture tests the data, integrations, operating constraints and implementation path.
Product and UI/UX decisions follow the real tasks, reviews and handovers of daily operations.
Prepare for discovery
Bring one operational problem, your current steps and tools, the people involved and the decisions still open. Mark unknowns: you can describe the problem before you have a complete specification.
Fill in what you know. Write “unknown” beside anything that needs checking.
Illustrative scenario, not a client case study or a tested software design.
A request arrives by email. An administrator copies it into a spreadsheet. The manager cannot approve it because the supplier quote is missing. The administrator follows up, then returns the request for review.
These questions do not establish that an integration is available or that new software is the right choice.
Use a written description or a redacted sample. Leave passwords and private customer records out of an initial message. Implementation scope, cost, milestones and support need an engagement-specific agreement.
General discovery principles: GOV.UK discovery guide and user research in discovery. The worksheet and fictional example are my preparation guidance.
Discuss one workflow and your open decisions