BezelIQ Studio intent
There is no unified interface for producing and reviewing SDD artifacts before build begins. Agents currently produce prose instead of machine-readable contracts, briefs are lossy Priya interpretations rather than mechanically extracted from upstream documents, and there is no place where Justin can take a project from idea to build-ready with confidence that dev agents have everything they need. The gap between intent and execution is invisible until Quinn builds the wrong thing. The cockpit ex
There is no unified interface for producing and reviewing SDD artifacts before build begins. Agents currently produce prose instead of machine-readable contracts, briefs are lossy Priya interpretations rather than mechanically extracted from upstream documents, and there is no place where Justin can take a project from idea to build-ready with confidence that dev agents have everything they need. The gap between intent and execution is invisible until Quinn builds the wrong thing. The cockpit exists as a monitoring tool but there is no design and planning workspace.
| metric | target |
|---|---|
| Intake to Build-ready in one session | Justin completes intake, triggers artifact generation, reviews all four artifacts, and approves exec gate without leaving bezeliq.ai |
| Zero artifact reads outside Studio | Justin never opens a raw JSON file to understand a project |
| New studio add cost | Adding Sales/Marketing/Finance studio requires only a new route and module — no changes to platform shell, nav, or auth |
| Auth toggle | Enabling Supabase OAuth gate requires one config change — no code change |
There is no unified interface for producing and reviewing SDD artifacts before build begins. Agents currently produce prose instead of machine-readable contracts, briefs are lossy Priya interpretations rather than mechanically extracted from upstream documents, and there is no place where Justin can take a project from idea to build-ready with confidence that dev agents have everything they need. The gap between intent and execution is invisible until Quinn builds the wrong thing. The cockpit ex
| metric | target |
|---|---|
| Intake to Build-ready in one session | Justin completes intake, triggers artifact generation, reviews all four artifacts, and approves exec gate without leaving bezeliq.ai |
| Zero artifact reads outside Studio | Justin never opens a raw JSON file to understand a project |
| New studio add cost | Adding Sales/Marketing/Finance studio requires only a new route and module — no changes to platform shell, nav, or auth |
| Auth toggle | Enabling Supabase OAuth gate requires one config change — no code change |
There is no unified interface for producing and reviewing SDD artifacts before build begins. Agents currently produce prose instead of machine-readable contracts, briefs are lossy Priya interpretations rather than mechanically extracted from upstream documents, and there is no place where Justin can take a project from idea to build-ready with confidence that dev agents have everything they need. The gap between intent and execution is invisible until Quinn builds the wrong thing. The cockpit exists as a monitoring tool but there is no design and planning workspace.
Justin can take any project from intake through artifact approval — producing complete machine-readable SDD artifacts (INTENT.json, BLUEPRINT.json, SPEC.json, POLICY.json) via a structured Studio interface — and advance to Build knowing dev agents have zero guesswork.
- item: metric: string, target: string
- item: metric: string, target: string
- item: metric: string, target: string
- item: metric: string, target: string
Machine-readable source fields
Justin can take any project from intake through artifact approval — producing complete machine-readable SDD artifacts (INTENT.json, BLUEPRINT.json, SPEC.json, POLICY.json) via a structured Studio interface — and advance to Build knowing dev agents have zero guesswork.
| type | success | description |
|---|---|---|
| Justin (exec) | Completes intake, triggers artifact generation, reviews everything in one place, advances to Build with confidence — without reading raw JSON or leaving bezeliq.ai. | Platform owner and sole human executive. Runs intake, reviews agent-produced artifacts as rendered human views, approves or rejects at exec gates, authors artifacts directly when needed. |
| Agents (Priya, Vikram, Zara) | Artifacts land in Studio automatically after task completion. No manual upload step. | Artifact producers. They receive tasks via the RedKey platform, produce SPEC.json, BLUEPRINT.json, and UX schemas, and surface outputs to Studio for Justin's approval. |
There is no unified interface for producing and reviewing SDD artifacts before build begins. Agents currently produce prose instead of machine-readable contracts, briefs are lossy Priya interpretations rather than mechanically extracted from upstream documents, and there is no place where Justin can take a project from idea to build-ready with confidence that dev agents have everything they need. The gap between intent and execution is invisible until Quinn builds the wrong thing. The cockpit exists as a monitoring tool but there is no design and planning workspace.
1
bezeliq-studio
2026-04-28
- direct
- confident
- precise
- calm
- Technical jargon in UI labels (HCS, task.complete, roles.exec)
- Verbose helper text
- Playful or casual tone — this is a professional tool
- Use studio metaphors — project, artifact, review, approve
- Label actions by what they do, not by what they trigger ('Approve Design' not 'Post task.complete')
- Short labels — never explain a button in the button text
BezelIQ Studio
An interpretation workspace. Justin comes here to think, review, and decide — not to execute. Calm, spacious, and confident. More whitespace than the cockpit, which is a dense operations monitor. This feels like a design studio, not an ops center. The machine complexity — JSON, HCS, Supabase — is completely invisible.
- Dense information panels that feel like a monitoring dashboard
- Exposing raw JSON or HCS message payloads to Justin
- Multi-step forms that feel bureaucratic or slow
- Any modal or flow that blocks access to the rest of the studio
- Progressive disclosure — show the right level of detail at each step, not everything at once
- Artifacts are always rendered as readable human views, never raw JSON
- Every gate has a clear approve or reject action — no ambiguous states
- Studio picker is always accessible — switching studios is one click
- Forge design system — consistent with BezelIQ platform visual language
- Platform shell — nav, studio switcher, shared layout, routing
- Dev Studio — intake flow, artifact production triggers, artifact rendered views, exec gate approval
- Ops Studio embedded inside the shared Studio shell via /ops and /cockpit-frame
- Marketing Studio embedded inside the shared Studio shell via /marketing and /marketing-frame
- Product/tool launch surfaces such as the AD SEO product route
- Auth architecture wired but ungated — Supabase OAuth ready to enable
- bezeliq.ai primary domain — Cloudflare Pages deployment
- Dreamborn Forge design system — more whitespace variant than cockpit
- Sales Studio and Finance Studio — future modules
- Native Vue rewrite of Ops Studio cockpit internals — follow-on
- Native Vue rewrite of Marketing Studio internals — follow-on
- Auth enforcement — Supabase OAuth gate is designed but not active
- Agent-authored artifacts for other studios (Sales briefs, Finance models)
- Mobile-responsive layout — desktop only in v1
- Machine artifacts are never shown as raw JSON to Justin — every artifact has a rendered Studio view
- All four artifacts (INTENT.json, BLUEPRINT.json, SPEC.json, POLICY.json) must be present and approved before a project advances to Build
- Auth architecture is present from day one (Supabase OAuth) even if not enforced in v1 — flipping auth on requires config change only, no architecture change
- Platform shell must accommodate Dev Studio, Sales Studio, Marketing Studio, Finance Studio, and Cockpit as routes — adding a new studio requires no change to the shell architecture
- Dreamborn Forge design system — no custom CSS that conflicts with forge tokens
| metric | target |
|---|---|
| Intake to Build-ready in one session | Justin completes intake, triggers artifact generation, reviews all four artifacts, and approves exec gate without leaving bezeliq.ai |
| Zero artifact reads outside Studio | Justin never opens a raw JSON file to understand a project |
| New studio add cost | Adding Sales/Marketing/Finance studio requires only a new route and module — no changes to platform shell, nav, or auth |
| Auth toggle | Enabling Supabase OAuth gate requires one config change — no code change |
- Justin has to open a JSON file or read raw markdown to understand what an artifact contains
- A project shows as build-ready with one or more artifacts missing or unapproved
- Adding a second studio (Sales, Marketing) requires refactoring the platform shell or navigation
- The interface resembles the cockpit — dense rows, small text, status pills everywhere — rather than a spacious design workspace
- An exec gate action (approve, reject) is not clearly visible as the primary action on a review screen
- Auth is bolted on after launch, requiring a structural change to routing or state management