Phase 1 — Alpha Playbook

Goal: prove the core loop retains a small group of real users. Not growth. Not press. Retention and qualitative truth.

Who to invite (the atomic network)

Pull from existing Cinevva users, in priority order:

1. Top 3D-asset generators. They already have inventory to place, so they hit the creator aha fastest. Query the generations table for users with the most completed 3D generations.
2. Active community members. People who engage (radio, Discord, replies). They give feedback and tolerate rough edges.
3. Users with a social following. They produce shareable screenshots later. Do not over-index here for alpha; they matter more in beta.

Target 50, invite ~80 to clear no-shows. Exclude @cinevva.com from "real user" metrics.

Invite mechanic

• Personal email from Mariana, not a mass blast. Tone: "We built something new and want you to be one of the first fifty people inside it."
• Unique invite link granting access. The world is not publicly reachable in Phase 1.
• In-world feedback button (screenshot + free text to a D1 table).
• Weekly 20-minute calls with the 5-10 most active testers.

The first session must deliver the loop

Founders' Quest framing: the alpha is a frontier being settled, and the invitee is a founding citizen. The opening is not "here's an empty plot." It is a short guided objective chain that doubles as the tutorial: arrive, claim land, place your first object (creator aha), reach the first landmark, leave your mark, earn a founder badge. This gives a solo player something to do when no one else is online, routes the few online players into collision points, and gives a reason to return.

Hard requirement before any invite goes out: the world must be seeded (AI + curated) so the first invitee lands in a populated place, not a tech demo. Empty world = dead alpha.

Pre-launch seeding checklist

• AI-generated + curated content placed across the starting area (target: feels inhabited, not a void).
• 5-10 internal showcase plots built by the team to demonstrate the ceiling.
• The mission/loop is playable end-to-end solo.
• First-object-in-under-60-seconds path verified on a fresh account.
• Analytics wired: session length, D1/D7 return, first-object time, objects placed, plots claimed, reactions, feedback events.
• Crash/error reporting live. 48-hour load test with simulated 50 players passed.

Weekly cadence

• Mon: pull last week's retention + activation numbers, review feedback table.
• Wed: ship the top 1-2 friction fixes identified by testers.
• Fri: calls with most-active testers; capture the best plots as screenshots/clips for the eventual beta trailer.

Phase 1 success criteria (gate to Phase 2)

These must hold before opening a public beta:

• Island/world loads playable in under 5 seconds on broadband desktop.
• 50 concurrent at 60fps on mid-range desktop; zero server crashes over a 48-hour test.
• First object placed in under 60 seconds from landing, for a majority of new testers.
• Retention signal: a recognizable cohort returns in week 2 without being prompted. If nobody comes back on their own, the loop is not fun yet. Do not paper over this with reminder emails.
• 5-10 plots good enough to screenshot for marketing.

If the retention signal fails, we do NOT proceed to beta. We fix the loop or rethink the thesis. See kill-criteria.