AI Assisted Customer Journey Builder
Turning a broad business goal into an editable draft journey for stakeholder review, with assumptions made explicit.
The problem
Stakeholders need a shared understanding of a proposed workflow before detailed development begins. Abstract requests are difficult to discuss and often hide key assumptions, which leads to rework later.
Target user
Product manager, operations lead, or business analyst. Key task: turn an ambiguous request into a tangible journey the team can review and align on.
Product goal
Make the AI output editable, visually structured, and explicit about assumptions so the team can move from an abstract request to a reviewable product artifact quickly.
User flow
- 1Enter a business objective, target user, and constraints (or start from an example).
- 2Review the AI-generated journey proposal and its stages.
- 3Edit stages, reorder them with move controls, and set their status.
- 4Open the assumptions panel and mark assumptions as confirmed or unresolved.
- 5Review the open questions.
- 6Export a concise product-concept summary for a review meeting.
Design decisions
- The draft is explicitly a draft — a visible "Editable draft" badge sets expectations and invites correction.
- Move controls over drag-and-drop — up/down controls are more accessible, keyboard-friendly, and reliable on mobile than drag interactions.
- Assumptions are a first-class artifact — confirming or leaving assumptions unresolved turns hidden risk into meeting agenda items.
- Export closes the loop — the review summary converts an interactive canvas into something a stakeholder can read and share.
AI interaction design
Input is JourneyGenerationService.generateJourney(request). Output is a JourneyProposal with summary, stages, assumptions, and open questions. The human review loop is: edit stages → reorder → confirm assumptions → export. The AI output is never treated as final.
Technical approach
Next.js App Router with a React client component for the builder. TypeScript types for JourneyRequest, JourneyStage, and JourneyProposal. Atomic component structure with reusable atoms and feature-scoped stage/assumption components. The mock generator selects a template based on the goal keywords. The export view renders a table and copy-to-clipboard summary.
Tradeoffs
- Generation is template-based rather than model-driven, which keeps the demo deterministic and fast.
- No drag-and-drop, by design, to prioritise accessibility and mobile usability.
- Journeys are not persisted; the artifact exists for the session and can be copied out.
Success measures
- Time to produce an initial journey
- Number of assumptions resolved in a review
- Stakeholder agreement
- Rework avoided before implementation
Next steps
Add persistence and versioning, stakeholder comments, and an optional real model adapter; then validate whether generated journeys reduce time-to-alignment in real reviews.