Yorick Writing Studio

An AI-assisted writing system built to preserve the integrity of a writer's voice. The writer stays in control. AI assists without overwriting.

Status: Private alpha

Overview

The Problem

As generative AI tools entered creative writing, most optimized for speed and output at the expense of voice. Readers began reacting negatively not just to AI-generated prose, but to the suspicion that a piece might be machine-written. Writers lost trust in the tools and in the work they produced.

The challenge was to design an AI system that assists without overwriting authorial intent or encouraging generic text.

Design

What we designed

An editor-first workspace where the manuscript is the center of gravity, not a chat window
Selection-driven AI actions: tighten, expand, clarify, increase tension, rewrite dialogue, check continuity
A proposal tray that holds AI-generated alternatives without silently altering the draft
Separation of analysis (diagnostics, continuity checks) from revision (text rewrites)
Writer control over AI intrusiveness: quiet, helpful, or active modes
Story Bible, canon tracking, and scene-level contextual intelligence

Decisions

Key design decisions

Preservation over generation Designed the system around preserving the writer's voice rather than text generation. Avoided default prompt-to-prose flows that encourage copy-paste authorship.

Structure and revision Built the experience around structure, revision, and decision-making rather than output volume.

Transparency and control Emphasized transparency and user control over "magic" outputs. AI suggestions are proposals, not edits.

AI as collaborator Treated AI as a collaborator within a workspace, not an authority. The writer always has final say on what enters the draft.

Relevance

Why it matters professionally

Yorick is directly relevant to the broader challenge of designing AI-enabled products. The problems it addresses — trust, transparency, user control, workflow integration, the boundary between human judgment and machine output — are the same problems facing any team building AI into their product.

This project forces working through those problems at the interaction design level, not just in theory. The technical implementation involves designing UX and system behavior in parallel, defining how context and state persist across a writing project, and building production frontend code to validate interaction models.