AI & Automation.
Practical AI that ships. Not demos.
Cost controls, evaluation suite, prompts under version control. An AI project without an eval pipeline isn't a project — it's a demo.
How the work actually shows up.
Anthropic-first integrations — caching, cost caps, structured outputs where they earn their keep.
Evaluation suites and prompt/version discipline before anything touches users.
MCP servers when tool reuse across clients or agents justifies the surface area.
Structured extraction into Postgres — typed records, not mystery JSON blobs.
- Prompt engineering + evaluation suite
- Anthropic / OpenAI API integration with caching + cost controls
- MCP server(s) for tool orchestration
- Structured extraction pipeline (Postgres ingest)
- Anthropic Claude (API + MCP)
- OpenAI (where justified)
- Python / FastMCP
- TypeScript
- Postgres
- Sentry
Clear anchors.
These are scoped build tiers after the Discovery Sprint (€1,500). Retainers, hand-off, and invoicing rules live on the pricing page.
One production use case — API integration, caching, cost controls.
- Scoped integration architecture
- Prompt + eval for the primary workflow
- Cost monitoring hooks
- Handoff documentation
Structured extraction → Postgres, evaluation suite, production monitoring.
- End-to-end extraction pipeline
- Schema + validation on ingest
- Eval harness + regression checks
- Dashboards for cost + error rate
MCP servers, multi-agent orchestration, custom evaluation infrastructure.
- Multi-agent or multi-tool orchestration
- Custom MCP + internal tooling
- Long-term eval + model-migration playbook
Real work, no stock.
Mockups and production screens from client and own-product work across the ai & automation discipline.
Built with this discipline.
The honest questions.
Anthropic (Claude) is my default for reasoning + tool use. OpenAI where a specific capability wins. Local models only when residency or cost forces it.
Start here.
30-minute call — honest fit on whether ai & automation is the right move.
Discovery Sprint cost is deducted from the first invoice if we proceed to build.