Curve Zero to shipped: one-person AI pipeline behind Curve Clash rewrites studio cost math
Summary
Jonathan Bjerk, a Norwegian accountant, shipped Curve Clash across web, iOS, Android, and macOS in seven months using only ChatGPT, Cursor, and Claude
Zero engineering hires compressed a multi-specialist production into a single-person workflow
Engineering execution commoditizes; game feel and product judgment hold the remaining premium
"AI was my engineering team, not my creative director"
01
Seven months from curve zero to cross-platform launch
Bjerk started May 28, 2025 with no Dart or Flutter experience
Entire codebase generated through conversational prompts across three AI tools
ChatGPT handled early prototyping; Cursor added multi-file coherence; Claude managed architecture
Flutter and Firebase, both AI-suggested, delivered four-platform coverage at near-zero marginal cost
Each tool switch carried real transition cost in context and momentum
02
Real-time multiplayer broke the zero-code ceiling
AI produced correct networking code, but live edge cases required two months of human debugging
Latency variance, state desync, and race conditions defied AI's deterministic strengths
"AI is phenomenal at building systems, but you still need to deeply understand what you've built when things go wrong"
Current boundary: stateless or near-stateless games clear zero-code viability; real-time networked games do not
03
Game feel remains the non-automatable asset
Turn speed, trail thickness, gap duration required hours of manual playtesting with no AI input
"Tiny parameters completely change whether the game feels fun or frustrating, and AI has no intuition for that"
Product prioritization - modes, Arena tournament structure - was entirely human-driven
Post-launch iteration compounded: features that took a month early now ship in days
04
1,500 players validate engagement, not scale
Over 15,000 minutes of total gameplay logged - high average engagement per user
Peaked at #4 in Norway's Family Games chart on launch weekend, a temporary spike
Word-of-mouth driven by local multiplayer at schools and events, not the AI development story
The AI narrative functions as a press hook for developers, not a consumer value proposition
05
Cost structure shifts from engineering headcount to creative direction
Bjerk identifies marketing as harder than building the game itself
Simple, well-understood mechanics are the prerequisite for AI-viable production
Complex physics, narrative systems, or novel interaction models still require conventional engineering
"The studios that adapt will thrive. The ones that dismiss AI-assisted development will be competing against an army of passionate solo creators who can move faster than any committee"
For publishers evaluating outsourcing: the pipeline eliminates generalist engineering staffing, not senior creative judgment
Events
Companies
—
Games
Locations
—
SYNTHESIZED FROM:
COMPILED BY GAMES ATLAS EDITORIAL
RELATED ARTICLES
EXPLORE MORE
Mission Control Games tests whether AI game development can replace headcount in casual mobileExternal dev confidence holds but risk tolerance tightens as communication and AI governance emerge as top friction pointsAI game development reality check: only 6 of 500+ tools actually workoutsourcing and co-development
Comments (0)
No comments yet. Be the first to share your thoughts!