AI-Powered Writing Platform

The founders had a beautiful Figma file and a failed first attempt from an offshore team that couldn't get real-time sync working without documents corrupting. Professional screenwriters were the target users, people who would walk the moment the editor dropped a keystroke or reformatted their dialogue. They needed multi-author collaboration, genre-aware AI suggestions, and bulletproof formatting for screenplays, novels, and short-form fiction. All of it had to work offline-tolerant on slow cafe Wi-Fi.

We rebuilt the editor from scratch in Angular with a custom rich-text engine that understood screenplay formatting as a first-class citizen (Fountain-compatible output, not markdown-in-a-trenchcoat). The backend runs Django with a GraphQL API so the frontend can fetch exactly the document shape it needs without over-fetching across slow connections. Real-time sync uses operational transforms over WebSocket channels, so concurrent edits merge cleanly instead of last-write-wins. AI suggestions route through a fine-tuned model that knows when a character is speaking, when a scene is breaking, and when to stay quiet.

• Grew to 10K+ paying writers in 6 months; platform hit profitability in month 7
• Average draft completion time dropped 40% once AI suggestions were turned on for beta cohorts
• 99.8% uptime across an 18-month window on GCP with Cloud Run auto-scaling
• Founders closed their seed round with this build as the product of record

We picked GraphQL over REST so the editor could fetch deeply nested document trees (scenes > beats > dialogue) in a single round trip. REST would have meant 4+ sequential requests on every document open. Angular was chosen over React because the team inherited a small Angular codebase and we valued founder-team continuity over our own preference. The platform runs on GCP Cloud Run with Cloud SQL for Postgres; scaling is automatic and we haven't had a paging incident since launch.