Research · Design · Knowledge · Build
From real work to the systems that make it possible.
I connect research, product strategy, service design, knowledge architecture and the building of AI systems. The starting point is the problem to solve with the team.
This practice can involve working as a Forward Deployed Designer, shaping knowledge architecture or prototyping. The role follows the context and the responsibilities the project needs.
How I move from a problem to a useful system
This framework sets out my approach and the questions to address in an engagement. The work section separately describes contributions and outcomes from past projects.
- 01
Observe real work
Use interviews, observation, informal practices and exception cases to understand what people actually do and why.
Make visible: actors, tasks, information, tools, workarounds and constraints.
- 02
Establish a baseline
Agree with teams on what needs to improve and how to recognise improvement. Look for available data before setting a target.
Measure as relevant: end-to-end time, waiting, rework, errors, exceptions and approval latency.
- 03
Structure domain knowledge
Define domain concepts, entities, relationships and rules. Connect information to its sources and distinguish facts, assumptions and decisions.
Design: domain model, ontology, knowledge graph and update rules.
- 04
Question the workflow
If we built this way of working today, which steps would still be necessary? Which decisions, relationships or obligations must be preserved?
Produce: a simpler scenario, its assumptions and the changes in responsibility.
- 05
Allocate roles and decisions
Clarify what belongs to a person, an agent, an automation or software, and design the handoffs between them.
Specify who executes, knows, decides, approves, can override and remains accountable.
- 06
Design the context for AI
Specify the knowledge, instructions, sources, permissions and task information needed at each step. This is where Context Engineering fits into my approach.
Define: relevant context, information selection, freshness, access, limits and traceability.
- 07
Prototype and evaluate
Test assumptions on representative situations with users and engineering. Include difficult cases from the prototype stage.
Define: expected outputs, acceptance criteria, test cases, sources, human review, escalation and failure modes.
- 08
Build, integrate and refine
Build the parts of the system within the scope agreed with the team, prepare their integration and observe usage to improve them. Make production and operational responsibilities explicit.
Track: quality, exceptions, adoption and effects on work against the baseline.
Responsibilities defined for each project
I work with business, product, design and engineering teams. My involvement can span problem framing through building; together, we clarify contributions, interfaces and accountability for decisions.
- Business intent and real work
- I lead research, problem framing and the design of how work should happen.
- Knowledge and context
- I connect business entities, relationships and rules to the information people and AI systems need.
- Prototype and evaluation
- Together, we test assumptions against technical constraints and user feedback.
- Architecture and production
- We agree on what each person builds and on security, integration, reliability and maintenance requirements.
- Adoption and impact
- Together, we follow usage, friction and observed outcomes.