GFAVIP Display Strategy

Own the content and ops. Commodity the glass. Don’t become the SaaS people cover with masking tape.

← Cyborg signage notes  ·  Cyborg wave  ·  Headless SSO  ·  GFAVIP Perks  ·  CBS displays  ·  Stage Agent

One-sentence product

GFAVIP Display is a member-aware display operating system: orgs schedule templates and playlists; devices only pull frames. Hardware is interchangeable — e-ink room signs, old tablets, Pi kiosks, TVs — so we’re never locked to a bad vendor panel or a $40/year “digital signage” tax that breaks the restaurant menu.

Working surface: gfavip.com/display (name may evolve; the architecture should not).

Learnings we are standing on

Source Learning Implication for GFAVIP
Cyborg signage / Todd Dailey thread Industry often sells mediocre SaaS; endpoints are ~$40 ESP32 e-ink + Wi-Fi; AI can write firmware Compete on product + ops, not on proprietary glass
Thread counter-argument Hard part is credentials, certs, updates across many rooms for years That is the product we sell — reliability and content ownership
ZecTrix Note 4 class E-ink + mic + speaker + battery = sticky note / personal sign Same CMS can feed “room now” and later “voice sticky” slots
CBS 2025 displays Pi + browser kiosk + central dashboard URL works for venues Web-first renderer remains MVP for complex layouts
ESP32 TFT notes / RegardingWork display Device registry + poll/API pattern already in our history Don’t restart from zero — evolve to GFAVIP identity
E-paper console · Cart A Status + dump architecture: device I/O, cloud brain, vault write Room signs poll; optional button dumps to HEOS/agents
GFAVIP badge CBS launch + evergreen identity; provision device ↔ member Display screens are org/room identity; badges are person identity — same family
Stage Agent · CBS avatar Event presence on big glass is a different form factor, same content OS Templates: stage crawl, now/next, sponsor — not only door signs
Headless SSO · Wallet Agents and machines need token identity without browser theater Devices get scoped keys; humans manage via GFAVIP login
Office structure · graph engineering Named jobs, human gates, residue in the vault Content pipeline: agents draft, humans approve high-blast screens

Positioning

We are We are not
A display CMS + device registry for GFAVIP orgs and events A locked hardware franchise
Member- and event-aware (rooms, CBS, masterminds, desks) Generic “any logo on any TV” commodity SaaS only
Open to BYO hardware (Pi, tablet, e-ink, Note-class) Dependent on one panel OEM
Obsessed with OTA, stale-content fallbacks, recovery docs A demo that dies when Wi-Fi or the vendor dies

Enemy: expensive signage contracts that leave operators with broken backends and tape on prices. Friend: builders who will flash an ESP32 if we give them a clean JSON contract.

Architecture (strategy view)

gfavip.com/display  (or display.gfavip.com)
│
├── Control plane (humans + agents)
│   ├── org / rooms / events
│   ├── templates & playlists
│   ├── schedules & overrides
│   ├── approval gates (stage / public)
│   └── device registry (key, last_seen, firmware, location)
│
├── Content API
│   ├── GET  /d/:token          → HTML/kiosk frame (Pi, tablet, TV browser)
│   ├── GET  /api/frame/:id     → JSON for e-ink / low-power boards
│   └── POST /api/voice-in      → optional sticky capture (later)
│
└── Endpoints (commodity)
    ├── Web kiosk (Chromium / old iPad home screen)
    ├── ESP32 e-ink room sign (~$40 class)
    ├── Note-class voice sticky
    └── Stage / big glass (Stage Agent templates)

Same rule as cyborg principles: one job per device; brain and memory stay central.

Content model (what goes on glass)

Template When Blast radius
Now / next (room or stage) Meetings, summit rooms, masterminds Medium — wrong time looks amateur
Org status (Big 3, HEOS-style) Team desk / HQ wall Low–medium — internal
Perks / sponsor CBS, events, lobbies High — needs human gate
Menu / simple list Simple venues (restaurant thesis) Medium — must be easy to edit offline-ish
Sticky / note Personal or kitchen board Low — voice/agent fill later
Fallback Stale device / offline Critical — “See host desk” never blank-brick

Phased roadmap

Phase Ship Success looks like
0 · Spec This page + API sketch + 2 templates (now/next, fallback) Written enough that an agent or junior can implement
1 · Web kiosk MVP URL per screen; login as GFAVIP org; edit now/next in browser Runs on old tablet + one Pi at HQ or event
2 · Device registry API keys, last_seen, content_version, name/room fields You can list dead devices without walking the building
3 · E-ink endpoint JSON frame for Cart A / Note-class / TRMNL-class One conference-room door sign works for a full day on Wi-Fi
4 · Event pack CBS / summit templates; sponsor slots; human approval for public glass Replace ad-hoc slides and RegardingWork-only path for GFAVIP moments
5 · Ops product OTA story, recovery USB, alert if content stale > N hours Someone who isn’t Mike can re-flash and rejoin Wi-Fi from a one-pager
6 · Capture (optional) Voice sticky / dump button → agent → approved sticky frame Same as cyborg dump, landing on glass instead of only markdown

Hardware strategy (buy later, design now)

  1. Phase 1: free/cheap glass you already own (tablet, laptop, TV + Pi)
  2. Phase 3: one e-ink ESP32 class from Cart A or Note 4-class — prove JSON path
  3. Fleet: only after registry + recovery docs exist
  4. Never: ship 50 devices before one URL kiosk is boring-reliable

Catalog context: gadgets list · e-ink path: cyborg-epaper

Identity & blast radius

  • Human admin — GFAVIP login; edits playlists and rooms
  • Device — long random key; can only pull its frame (and later post voice dumps if allowed)
  • Agent — can draft content via PowerLobster / headless SSO; cannot publish high-blast templates without a human gate
  • Public stage — double-gate: wrong slide is a brand incident

Same spirit as badge provision + revoke and human gates at expensive steps.

Success metrics

Metric Why
Screens online / last_seen < 1h Ops health
Content age (stale frames) Avoid blank or wrong day
Time to edit “now/next” (non-engineer) Product is for hosts, not only Mike
Recovery time for a bricked device 5-year cost of ownership
Event screens live without heroics CBS / mastermind proof

Anti-goals

  • Rebuilding a full video ad network in year one
  • Requiring custom hardware purchase to use the product
  • Letting agents auto-post sponsor claims to lobby TVs
  • Opaque firmware with no recovery path
  • Pricing that only works if customers forget how to cancel (the tape problem)

Near-term actions (this month)

  1. Decide host: gfavip.com/display vs display.gfavip.com
  2. Spec Phase 1 templates: now/next + fallback HTML
  3. Stub device token auth and one kiosk URL
  4. Run on one physical screen for a week of real meetings
  5. Write a one-page recovery runbook (Wi-Fi, reflash, factory content)
  6. Only then order a second form factor (e-ink)

Related pages

Cyborg signage

Hardware + industry thesis that feeds this strategy.

Read →

CBS 2025 displays

Pi kiosk tutorial — venue deployment pattern.

Read →

GFAVIP badge

Person-scale display / identity cousin.

Read →

GFAVIP Perks

Event content that could land on lobby glass.

Read →

Stage Agent

Big-glass event presence.

Read →

Headless SSO

Machine identity for devices and agents.

Read →

Living notes

This is the strategy spine for GFAVIP Display. Implementation tickets should link back here. Field hardware notes stay on /cyborg-signage so product strategy doesn’t turn into a gadget catalog.

Comments

Approved comments appear below. Log in once with GFAVIP — it applies across the whole site. GFAVIP login

View comments archive