- Role
- Sole builder — concept, product design, prototype
- Domain
- Research-driven, long-form interview show (generic)
- Surface
- 8 connected screens · idea to published
- Data
- Fictional sample data
- Stack
- Next.js 16 / React 19 / TypeScript / Tailwind CSS v4 / typed mock data
- Status
- Concept prototype · unsolicited · not affiliated with any real show or company
The Operational Problem
Long-form, research-driven shows accumulate operational drag in every corner. The guest pipeline lives in inboxes, research lives in scattered documents, factual claims go unverified until the edit lock, clips stall waiting for sign-off, and sponsor deliverables get chased by hand. The scarce resource is the host's attention — and most of it gets spent on things that never needed it. The prototype reframes the whole operation around protecting that one resource.
The Product Principle
One rule drives every screen: the host always knows what needs their judgment, and everything else keeps moving without them. The decision inbox is the feature — and it works by keeping everything that does not need the host out of it. Operators triage and run the day; the host is presented only with the calls that are genuinely host-level. Two design commitments hold the line: AI drafts and humans approve (never the reverse), and evidence has tiers, so a claim is only ever as strong as its weakest source.
System Design
Eight connected screens carry an episode from idea to published, sharing one mock data source so every view stays internally consistent. Each surface is a focused queue with an explicit gate before work reaches the host.
- Command Center — a daily operator brief, a production queue from idea to published, and a Host Decision Inbox scoped to host-level calls only.
- Episode Workspace — brief, research vault, question bank, a claims-and-evidence tracker, transcript, clip map, and publishing checklist, one place per episode.
- Research Vault — a reliability-tiered source library (Confirmed down to Do Not Use) with an AI-drafts / human-approves research-packet flow.
- Guest CRM — a candidate pipeline with fit scoring, warm-intro tracking, and follow-up discipline.
- Clip Studio, Sponsor Ops, and Rights & QA— transcript-driven clip candidates with an explicit "Ready for Host" gate, sponsor deliverables with proof-of-delivery, and a publish-readiness gate covering rights, claims, and sponsor compliance.
Prototype Walkthrough
Selected views from the clickable Next.js prototype. Every screen carries a persistent banner — 'Concept prototype · fictional sample data · not affiliated with any real show or company.' All guests, sources, sponsors, episodes, and metrics are fictional and illustrative; actions are visual only and create no records. Click any frame to enlarge.
What It Demonstrates
- A single product principle (protect the host's attention) carried consistently through every one of eight screens.
- A human-in-the-loop research flow with tiered evidence — AI in the drafting seat, a person in the approval seat — rather than treating AI as an oracle.
- The ability to walk into an unfamiliar operational domain, find the real bottleneck, and ship a focused, navigable system as a live prototype rather than a deck.
- Concept prototype, not production — a clickable Next.js front end with typed mock data; no backend, no auth, no external integrations.
- Built unsolicited from public context to demonstrate product thinking.
- Every guest, source, sponsor, episode, and metric is fictional sample data, marked as such in the product itself.
- Not affiliated with, commissioned, or endorsed by any real show or company.
- Actions (generate, assign, approve, mark ready) are visual only and create no records.
