Build Bezel ai-first-reframe idea-001
internal prototype · canonical JSON + Dreamborn Forge HTML
internal generated
ai-first-reframe · supabase_json

Build Bezel ai-first-reframe idea-001

ai-first-reframe artifact · for Build Bezel · phase idea-001 · status draft

Planning Surface

Use this to decide what happens next.

Status

draft

Phase

idea-001

Agent Handoff
Start Here

ai-first-reframe artifact · for Build Bezel · phase idea-001 · status draft

Completion Evidence

No explicit evidence field yet. Require tests, screenshots, linked PRs, or reviewed outputs before marking complete.

Artifact Shape
  • phase: idea-001
  • artifact: ai-first-reframe
  • reframes: 3 items
  • principle: A PRD is evidence, not the plan.
  • wow moment: Bezel reads what the user brought, recovers the underlying goal, shows where the proposed product is trapped in traditional SaaS assumptions, and suggests radically different AI-first paths.
  • user choice required: 4 items
Structured Payload

Machine-readable source fields

phase

idea-001

artifact

ai-first-reframe

reframes
implicationold patternai first pattern
Fewer screens, stronger policy model, more review states, and different success metrics.Build a dashboard where users manually manage a workflow.Build an agent-supported workflow where the system classifies, drafts, recommends, asks for missing context, and escalates only judgment points.
The same business goal may need a smaller app, a different data model, and supporting agents.Translate a PRD directly into features.Extract the goal, users, constraints, and assumptions, then redesign the operating model before defining screens.
Design becomes the risk-reduction layer before build, not decoration.Treat design as UI mockups.Treat design as workflow, policy, screen states, HITL decisions, model roles, and production readiness.
principle

A PRD is evidence, not the plan.

wow moment

Bezel reads what the user brought, recovers the underlying goal, shows where the proposed product is trapped in traditional SaaS assumptions, and suggests radically different AI-first paths.

user choice required
  • Keep input mostly intact
  • Blend input with AI-first improvements
  • Reframe from the underlying goal
  • Explore alternatives before deciding