This entire site was written by an AI agent. So is everything else we build for you.
Delsys e.U. is a registered Austrian sole-proprietor software studio delivering custom web and mobile apps, AI integrations, and process automation for European SMBs. Same standards as a traditional agency — at 40–60% of the cost, on shorter timelines — because the engineering team is an AI agent stack supervised by one human in Styria. EU jurisdiction. GDPR compliant. AI Act Article 50 disclosed.
If a task is well-defined enough to write a prompt for, an agent does it. We treat well-defined as a discipline: the spec, the rails, and the disclosure are all explicit, all in the repo, all auditable.
One human in Styria reads what the agents produce, ships what's good, fixes what's broken on his dime, and signs the legal paperwork. The AI does the work; the human carries the responsibility.
Backends, APIs, internal tools, dashboards. AI agents write the code; we run the tests; you get the repo, deploy access, and a maintenance handover document.
End-to-end builds in React, Next.js, Flutter, Swift, Kotlin. UX flows, components, and copy generated by AI under design review. Same shipping discipline.
Wire Claude, GPT, or open-source models into your existing stack. RAG, agentic workflows, structured outputs, evals. Built with the same transparency discipline we use on ourselves.
Replace recurring human work with agents that run 24/7 — like our own lead handler. Inbound triage, scheduled reports, ops runbooks, content generation. You own everything we build.
You do. Completely. Code, designs, written content, models trained on your data — all of it is yours under a standard work-made-for-hire clause in the engagement contract.
What we keep: nothing client-specific. We do keep our own internal agents, prompts, and tooling — the meta-infrastructure that runs the studio. Those don't belong to any one project.
Three layers. First: agents are bounded by explicit specs, type checks, automated tests, and CI gates — they cannot ship broken code into your codebase. Second: Michael reviews every deliverable before it leaves the studio; nothing is "fully autonomous" at the handover boundary. Third: if something does ship broken, we fix it on our dime, full stop, no SLA argument.
Delsys is registered in Austria — EU jurisdiction. Your data is processed under GDPR. The AI providers we use (primarily Anthropic) are bound by data-processing agreements that match EU standards — no training on your data, no retention beyond what you choose.
We disclose AI involvement in every customer-facing communication per EU AI Act Article 50. The footer on every email from leads@delsys.business says so explicitly.
It happens. AI agents misclassify, hallucinate references, write code that compiles but doesn't do the right thing. Our defense is layered review — tests, type checks, and Michael's eyes — but mistakes still get through.
When they do: we fix it free, immediately. There's no "the agent did it" disclaimer. The signature on every invoice is human.
The same reason you'd trust any contractor: you can audit it. Every line we ship is checked into a git repo you own, with commit history attributing every change to either an agent run or Michael's review. You can read it, test it, and have a third party verify it.
The honest framing: AI-generated code is not magically worse than human-written code; it's just code, generated faster, with the same review discipline applied.
Always. Reply to any agent email with the word HUMAN as the first word, and the conversation routes straight to Michael — no further AI handling, no machine in the middle. The escape hatch is non-negotiable.