Case studies — Production systems

Production AI systems,
deconstructed.

Each study is a real system running in production right now — what it does, what it cost, what we'd do differently. No mockups, no vaporware, no "AI-powered" marketing speak.

01 AI Integration

Nova — A Personal AI Assistant That Costs <$50/mo

Replaced 6 mobile apps with one AI assistant. 90% of queries handled by free local models. In daily use since February 2026.

Read the full case study

Result: $450/mo in API costs
Stack: Node.js / React Native / llama.cpp / Claude API

02 AI-Augmented Development

Multi-Agent App Factory — 48 Apps in 10 Weeks

One person, four AI workers, parallel app development with automated QA gates and shared memory. RxLog (now in Play Store production) is the flagship output.

Read the full case study

Result: Equivalent to 3-4 contract developers
Stack: Claude Code / Bash / Docker / Custom orchestration

03 AI Cost Optimization

Local Content Pipeline — Blog Posts on Cron, Zero API Spend

Three niche sites drafted end-to-end on a local 35B MoE model. SEO + automated fact-checking + 14 reusable analysis patterns. Originally cost $200/mo on cloud APIs.

Read the full case study

Result: $200/mo + zero data leakage
Stack: llama.cpp / Bash / Cloudflare Pages / Astro

Looking for something similar in your business?

See Services & Pricing