Recruiting · AI
Exposés that write themselves in three minutes
Applications and CVs become a finished candidate profile — Claude writes the copy, the pipeline sets the document.
3 min
per proposal · was 45
- Year
- 2026
- Client
- Trenkwalder
- Role
- Lead developer · architecture, build, rollout
- Areas
- Full-stack development · AI integration
AI-generated
AI-generatedThe starting point
Applications came in from several sources and were processed with a bought-in SaaS tool. The tool was expensive, sat outside the company's own systems, and couldn't do exactly the things that cost the most time day to day.
AI-generatedWhat was built
A pipeline inside the company's own recruiting platform: applications in from several sources, résumés parsed into consistent candidate profiles, and the finished presentation out.
If a piece of information is missing, the system spots the gap and starts a chatbot conversation to ask for it specifically — instead of blocking the case or waving it through incomplete.
01What you touch
02Where the data sits
03How it is operated
Stack
- Interface
- API
TypeScript RAG Claude / GPT / Gemini
- Data
PostgreSQL pgvector
- Cloud
AWS (S3, Lambda) Docker
Result
The bought-in SaaS tool was switched off. The generator runs inside the company's own platform.
