Seeds of the Throne

type: product-direction
status: established
updated: 2026-09-07
scope: vault functionality, workshop engine, composition engine, project explorer
source: "[[01 Sessions/Daily/2026-09-07 - Conversational Authorship Product Direction]]"

Conversational Authorship Product Direction

Founder problem

The author wants to finish Seeds of the Throne but does not have time to manually write and maintain every part of it. Build the system to solve that concrete problem first: preserve the author's story, judgment, corrections, and approval while taking on as much of the time-intensive story-development and manuscript-production work as can be performed reliably.

The broader market hypothesis is that many other people can tell and guide a compelling story but lack the time, craft training, organization, or production capacity to complete a book alone.

Direction

Build the vault into a conversational authorship system for story-rich, craft-light creators. A user should be able to tell a story naturally, answer one useful question at a time, guide important decisions, and move from raw material to a completed manuscript without first mastering professional writing terminology.

You already have the story. We help you discover it, develop it, and turn it into a finished book.

The working analogy is vibe coding for authorship: ordinary-language intent and iterative judgment on the surface; story craft, state management, continuity, drafting, revision, and production discipline underneath.

Initial user

Start with imaginative non-writers creating fiction from worlds, characters, fragments, recordings, role-playing histories, notes, or unfinished manuscripts. Preserve an architecture that can later support memoir and family history with stronger factual, consent, privacy, and uncertainty controls.

Product boundary

The system must eventually support the complete path:

natural intake → reflected understanding → dependency diagnosis → author gates → accepted story model → scene contracts → controlled prose → structural and line revision → continuity → manuscript export

Finished prose is in scope. Premature prose is not the default method for discovering unresolved story structure.

Governing requirements

• Never begin from a blank-page demand when conversational intake is possible.
• Keep source, system understanding, and proposed story use separate.
• Ask one high-leverage author-gate question at a time.
• Support free answer, decision help, inspiration, deferral, and correction.
• Propagate only author-accepted decisions.
• Show concrete story progress, not invented completion percentages.
• Require inspectable scene contracts for load-bearing prose.
• Learn voice from choices, corrections, positive examples, and dislikes.
• Separate prose residue from deeper narrative residue.
• Return newly exposed structural gaps from Composition to Workshop.
• Preserve provenance, decision history, continuity, and exact resumption state.
• Keep Markdown or another open export as the durable source of truth.
• Optimize for completed, author-approved work rather than requiring the author to manually perform every stage or master writing craft first.

Two flagship engines

Workshop Engine

Diagnose the current story, identify the most consequential missing information, conduct adaptive interview/workshop sessions, record author decisions, and propagate their consequences through the project.

Composition Engine

Translate approved story state into scene contracts and finished prose, then run separate developmental, continuity, narrative-residue, voice, prose-residue, and line-editing passes without inventing missing canon.

Demonstration surfaces

• Project Explorer: explains and demonstrates the vault, author decisions, provenance, workshops, engine state, continuity, and progress toward a finished story.
• Docs site: presents the resulting story as a reader-facing narrative experience, with development machinery excluded unless intentionally exposed.

Current constraint

Seeds remains development-first at its present stage. This direction expands the destination to finished prose but does not bypass current story gates or authorize bulk chapter generation.

Next validation milestone

Complete the four audits, then test one vertical slice:

intake → one author gate → accepted decision record → dependency propagation → scene contract → scene draft → deterministic continuity check → proposed vault updates

Do not install or rewrite the final skill system until the audits and the author-gate architecture are reviewed.

The proposed architecture and staged validation plan are indexed at Conversational Authoring System.