# Clone — Platform Vision (Byron, 2026-08-01)

**Exploration phase — mockups + Fable design pass + UX critique BEFORE committing to build.**

## What Clone is
A **presence platform**. Anyone lands, auto-registers with Google, and brings a *presence* to life — an avatar/persona/agent representation of a real or historical person. Functionally close to "persona" (Byron is NOT renaming to persona, but concedes they're near-synonymous; the clone-vs-persona line is itself an open question to articulate).

## Cloning modes
1. **From media** — clone someone from an audio tape or video (flagship emotional case: **clone my grandfather so his grandkids can ask him to tell a story**; "ask Grandpa / ask Dad / ask Mom a story").
2. **Yourself** — reads/looks at everything about you and can act like you / for you.
3. **A public persona** — a person's public-facing self (company spokesperson, brand voice).
4. **Historical / anyone** — e.g. **create George Washington**. Byron flags this as the higher-RAI-implication path (consent, likeness, truth, disclosure).

## The bigger frame — physical-world representation + agent-ability
This isn't just Q&A. It's an **AI persona / avatar / agent with a physical-world representation**, hyper-integrated into **sensor data**: phone cameras, mics, speakers, even **robotic or other interfaces** (temperature, external capabilities). Lineage: this is where **AgentSim** was going, but the point here is a *thing's ability to act and interact* — not simulate. The difference from a talking archive: it can **act**, not only recount.

## Use cases to pontificate (Fable to expand)
- Grandpa/Mom/Dad telling stories to kids (memorial / living memory).
- An avatar that **represents a company**.
- A **press advocate / press-release spokesperson** who can speak in press releases.
- A **teacher** avatar.
- Historical figures (George Washington) as education/experience.

## The big question (must be answered + wrapped into the experience)
**Why and where do you need human interfaces** — and why does this **preserve the human feel**? Thesis: Clone is a **bridge** for people who aren't ready to use/leverage AI natively — a warm, human surface into the whole space. Update the proposals around this.

## Landing experience (critical)
- **No cookie / unknown visitor → it jumps right in.** A **canned, decent-quality avatar (voice + video)** is present and greets you immediately — not a static marketing page.
- **After Google ID → personalized:** "Hi, how's it going — I know you're [name]…" (name from Google profile). Immediately friendly.
- **Immediate create-offer:** "Create something from what you have — a document, a video, an audio, a photo," OR **record yourself right there** (camera/mic on phone or computer) and create your own on the spot.
- Must feel frictionless, warm, alive, and let you **see what it's doing** (the transparency/x-ray idea from the first mockup).

## Auth + monetization
- **Anybody can auto-register** via Google ID.
- **Freemium:** ~**$1 of AI credit** on a new account by default; unlock more when you upgrade.
- **Track** like every other fleet system (analytics). **Publish** it and get it onto **info.arnao.ai ASAP**.

## Fidelity = HOW + WHAT (what makes it convincing) — Byron, 2026-08-01
A clone must analyze and capture not just **WHAT** a person says (facts, stories, opinions) but **HOW** they say it — **voice, cadence, mannerisms, speech patterns, verbal tics, catchphrases, humor, rhythm, the pauses, how they open and close a thought.** The *manner* is what makes it convincing; content without manner is an impostor. So the corpus + model must explicitly model both dimensions, and the **"watch it learn" pane must visibly capture manner** ("the way he trails off", "always starts a story with 'So —'", "his laugh"), not only facts. Ties to the voice clone (Chatterbox / ElevenLabs) + a **speech-pattern / style-analysis layer** over the transcript. This is also the honest, evidence-bound version: the clone speaks in their real patterns *because* those were observed, and it discloses where it's extrapolating manner it never saw.

## Mirror mode
**Real conversations that remember you** (persistent memory across sessions). (Byron: "mirror mode," not "miramos.")

## Quality / speed / cost slider (must be prominent in the UI)
- A **prominent slider** for **real-time tuning between quality and speed**, likely **intersecting cost**.
- Range: **insanely-fast realtime** ⟷ **more latency for higher-quality avatar platforms**.
- Tiered avatar backends: the **cheap/cheesy platform that didn't work well before** → a **paid server we run** → a **premium platform** (HeyGen / D-ID / or a recommended best-in-class — Fable/Claude to recommend). "For more money, we do the better one" must be a selectable option.

## Form factors + UX bar
- **Portrait AND landscape**; **trifold vs folded-trifold**; **mobile** with heavy **sensor integration**.
- The differentiator is a **killer, frictionless interface**. Byron wants it **eye-popping and very compelling**.

## Process
1. **Fable pass** (designer + strategist) across all of this — focus on UI design, answer the big questions, pontificate use cases, define 2-3 mockup directions.
2. **UX-critique agent** — lots of concrete feedback to make it eye-popping across portrait/landscape/trifold/folded/mobile + sensors.
3. **A few mockups** built from that.
4. Update the **PRD/proposal**; **publish** + **info.arnao.ai** fast.

Prior art: first landing+wizard mockup at clone-begin.arnao.ai; working chat MVP at clone.arnao.ai; concept studio at clone-studio.arnao.ai. Related PRDs in ~/clone-app/PRD*.md.

## KEY CONNECTION (Byron, 2026-08-01) — "The Sitting" = the create-side of the Family History project
Fable's bold idea "The Sitting" (clone people **while alive** via a guided interview) is NOT new scope — Byron already has a **family history project** with an "upload a video OR get interviewed" mechanism. The unification: **Family History = the capture pipeline** (guided interview / video upload → source material + on-camera consent record + heirloom); **Clone = the destination** where that person becomes a living, answer-back presence. One pipeline: interview → source → clone. This makes the memorial flagship a *living ritual* ("sit for your clone / interview your parents this Thanksgiving") instead of a posthumous reconstruction, and solves consent structurally (captured on camera). Wire Clone's create-flow to accept a Family-History interview as a first-class source. (Family history project details being located; connect concretely — see the search brief.)