Stage 14 — Applications
Five application areas where the patterns from Stages 1–13 land in real products. Each is its own ecosystem; each illustrates patterns transferable to your domain.
Prerequisites
- Stages 08–13 (you can build, ship, and operate AI features)
Articles
- Text-to-SQL — natural language to database queries
- Text-to-code — code generation, IDE agents, Claude Code patterns
- Browser agents — automating the web
- Financial reasoning — agents over financial data
- Case studies — real-world AI products and their lessons
MVU
You can:
- Recognize architectural patterns reused across application domains
- Lift a pattern from one domain (e.g. text-to-SQL retrieval) and apply it to your own
- Skim a “we built X” blog post and pull out the durable engineering lessons
Exercise
Pick one application area. Build a working v0 in a weekend. Identify three patterns from the article that you used; identify two patterns you’d add for production.
Hands-on companions
This stage maps directly to the /case-studies section — each case study is a real product on top of the /ship stack:
/case-studies/01— docs assistant with citations — RAG-heavy: MDX-aware chunking, citation-first prompting, three-bucket refusal eval. Real numbers (96% cite-coverage, 91% refusal precision)./case-studies/02— code-review agent — agent-heavy: propose-then-act tools, action-rate as the metric, 71% real-world action rate./case-studies/03— research assistant — multi-agent fan-out for cited briefs, with the cost/latency math measured on a real query./case-studies/04— customer-support bot — composes RAG + tools + escalation logic. Includes the ROI math (~$23K/month net).
Watch it interactively — applications-flavored demos:
- Text-to-SQL Playground — editable SQL with schema validator that catches hallucinated columns.
- Browser Agent — three traces (happy path / modal interrupt / layout shift) showing real failure modes.
- Tool Use Builder — the schema → call → result protocol every application uses.
