agents
AI Agents & Workflows
Custom agents that take over a real workflow: research, evaluation, preparation, or handover into the systems you already run.
AI that actually takes work off your desk.
AI agents, LLM integrations, automation, and custom applications developed for one concrete workflow in your company, not sold off the shelf. Short paths, running results instead of concept phases. And when your own team wants to develop with AI, I help them get there.
Services
from the idea to a running systemDo you have a process that costs too much manual work, or an idea with no product that fits it? I build the solution and I will also tell you when a standard tool is the better answer.
integration
LLM Integration
Connecting language models to your data and systems. From RAG through tool use to a clean fit into existing applications and APIs.
processes
Automation
Recurring manual work becomes an automated flow: documents, tickets, reports, data handovers. With explicit error handling instead of a black box.
tangible fast
Prototypes & Internal Tools
A working prototype instead of months of concept work: built quickly, tested directly, and expanded only once it proves itself in daily use.
enablement
AI in the Development Lifecycle
I help development teams actually use AI tooling: tool selection, prompting and review practice, guardrails, and anchoring it across the whole software lifecycle.
made to fit
Custom Software Solutions
Applications for requirements no off-the-shelf product covers. Small enough to stay maintainable, and a fit for your existing landscape.
both worlds
Secure AI Adoption
AI solutions built securely from the start: prompt injection, data leakage, permissions, and logging. The point where my two areas of work meet.
How I work
three principles 01
Understand, don't collect features
The workflow first, the technology second. It often turns out that a smaller intervention beats the solution originally requested.
02
Something running, early
A usable prototype takes days, not quarters. What works shows itself in the finished thing, not in a concept document.
03
Hand over maintainable
Documented, transparent, and without a dependency on me. You should be able to keep developing the solution on your own.
Build notes
how it actually comes together ai · development
AI & Software
Open notes from ongoing projects: what works well with AI tooling, where it fails, and which patterns hold up in real applications.
latest: How I Built This Website with AI AssistanceGot a process in mind that could run better?
Describe it to me briefly. In a no-obligation call we work out whether a custom solution is worth it and what a first working prototype could look like.
Discuss a project →