Founder & solo architect · 2026

TaleCheck

Static analysis for fiction, built solo with AI agents.

~2,300 commits in under 3 months

Context

Authors lose track of characters, timelines and set-ups across a long manuscript. TaleCheck works like ESLint for a novel: an LLM extracts the story’s world model, the author reviews it like a pull request, and deterministic rules find plot holes.

What I did

Designed the product and architecture; ran Claude Code agents in parallel and reviewed every change.
From an empty repo to a live product with paid subscriptions and beta authors in under three months.
LLM pipeline with extraction, evals against reference fixtures and cost control per run.
Telegram bot, web reports, background workers; four interface languages.