Graphene Studio · Closed Beta

You're a Creator.
Here's how this works.

The 90-second tour. Build multi-chapter audio fiction shows end-to-end — from premise to multi-voice ElevenLabs render — with the same pipeline that produces the public Graphene network.

Closed beta · No billing · Use the platform reasonably

The pipeline

  1. Step 1

    Pick (or build) a universe

    Universes are shared-lore worlds — characters, settings, motifs, voice rules — that seasons inherit. You can write inside an existing one (like Turing Logs) or build your own private world.

    Browse universes →
  2. Step 2

    Create a season

    Pick a universe, write a one-line premise, choose chapter count. The platform drafts a chapter-by-chapter plan in ~30s. You can edit before kicking off prose generation.

    Create your first season →
  3. Step 3

    Generate prose

    Each chapter is a single GPT call (~5–15s). Auto-refine runs three craft critics in parallel (tension, dialogue, repetition), folds their findings, generates take 2, promotes the better one. ~90s end-to-end per chapter.

  4. Step 4

    Review takes

    Every prose change leaves a take row. Open the Takes & Critique modal on any chapter to diff alternates, run additional critics, override applied findings, or roll back to a prior take. The bible auto-updates with deltas the model proposes — review them in the Bible modal.

  5. Step 5

    Render audio

    ElevenLabs multi-voice TTS, ~10–30s per chapter. Per-segment cache reuse means re-rendering one chapter doesn't re-TTS the others. Each chapter gets its own MP3 plus magazine-style pullquotes auto-picked from the prose.

  6. Step 6

    Publish (when you're ready)

    Flip the 'Publish to /graphene' toggle on the season row. Until then, only you (and super-admins) can see the season — it stays off the public slate, off RSS feeds, and out of search. Toggle it back off anytime.

    Open the seasons admin →

Visibility model

Your seasons are private by default. Other creators can't see them. The public /graphene slate doesn't show them. RSS feeds don't carry them.

When you're ready to share one, flip the Publish to /graphene toggle on the season's row. Toggle it back off anytime — listeners on the public slate stop seeing it; subscribers in podcast clients keep what they already downloaded.

Universes follow the same model. Yours are private; the platform ones (Turing Logs and any others built by the network owner) are shared lore that every creator can build inside.

What's not enabled (yet)

  • ·Direct billing. The closed beta has no charges or usage caps. The paid Studio tier ships once we have demand signal.
  • ·Public discovery for creator shows. Even when you publish to /graphene, your season appears on the slate — there's no per-creator profile page or search.
  • ·Cross-creator collaboration. Universes are owned by one creator; multi-creator universes aren't a thing yet.
  • ·Custom voice cloning. You pick from the existing ElevenLabs voice bank. Custom voices are a Studio-tier feature.

Stuck on something?

Every part of this is one DM away from getting unstuck. Sandon runs the platform — message him with anything from “I broke a thing” to “I want a feature” to “what would you do here?”

Welcome to Graphene Studio | HiveJournal