Partner-supported cases
Observed in context and examined through the artefacts teams used to make work happen.
Bachelor Thesis · Portfolio Case Study · 2026
Human-Agent Service Blueprint
From problem qualification to a justified Human-AI work system.
An evidence-based service-design method that helps teams move from a solution-led AI use case to a reasoned decision about task allocation, readiness, architecture, control, integration, and next steps. The result is an interactive web application, an offline workshop kit, and an exportable Decision Snapshot.
01 · The challenge
Start with the work, not the agent.
Many AI projects begin with an assumed solution: “We need an agent.” The research showed that the real bottlenecks often appear earlier—in problem understanding, roles, data access, handoffs, governance, and operational readiness. The Blueprint reverses that logic: qualify first, then shape the simplest system that can reliably do the job.
Not: “How do we build an agent?”
But: “Which form of automation is justified for this work system?”
02 · Research → Design
Patterns across real project work.
Six partner-supported project cases formed the research field. Participant observation and artefact analysis were brought together through cross-case thematic analysis and translated into a Design Science artefact.
Research moves
Participant observation, artefact analysis, and cross-case thematic analysis.
Design Science artefact
A practical method for turning empirical patterns into reasoned design decisions.
- Roles and handoffs
- Data, process and integration
- Trust, visibility and control
- Low-code as bridge and stress test
- Team and partner configurations
03 · The Blueprint
A guided process with room to revise.
- 01Orient
- 02Qualify
- 03Design
- 04Synthesis
- 05Decide
- 06Reflect
Design principles
- Problem before technology
- Qualify before building
- Allocate tasks by complementarity
- Use the minimally sufficient architecture
- Make control visible
- Design for verification and recovery
- Shape integration together
- Clarify lifecycle responsibility
04 · Digital implementation
A case evolves as decisions become clearer.
A linear macro process keeps the work legible while allowing teams to revisit and correct earlier decisions.
The Living Case Record carries the case state forward; adaptive next steps avoid a rigid completeness checklist.
The Service Canvas synthesises the work system, and a Decision Snapshot exports a portable milestone view.
Glossary guidance and persistent local cases support continued work; an Offline Workshop Kit enables the same method without the web application.
05 · Optional assistance
AI as a sparring partner, not an autopilot.
When asked, the optional AI Copilot reviews the current case, surfaces contradictions or evidence gaps, and proposes concrete additions. Each suggestion is reasoned, shown as a change preview, and applied only after explicit confirmation.
AI recommends evidence explains humans deliberate humans decide.
06 · Outcome
A Decision Snapshot made for the next conversation.
The export brings the current rationale and operating conditions into a portable format, so a decision can travel beyond the workshop or prototype.
- Goal and system boundary
- Solution category and maturity
- Current decision and conditions
- Human oversight and verification
- Blockers, recovery, and fallback
- Operational responsibility
- Next test, action, and owner