Responsibly implemented
Safety guardrails, source citations and transparency are part of it: chatbots are labelled as AI (EU AI Act), it stays traceable which data goes where, and sensitive data stays in-house where necessary.
AI Agents & Automation
Assistants that genuinely help
We build AI assistants, knowledge bots and automations that are tested on your data – with source citations, an eye on data protection and an honest assessment of strengths and limits.
What we build
From knowledge research to automating recurring workflows – matched to the IT you already have.
- Assistants that answer from your own documents and cite the source (technical term: RAG)
- Chatbots for the web and phone assistants (voicebots)
- Microsoft 365 Copilot & your own Copilot assistants (Copilot Studio)
- Automating recurring workflows (e.g. with Power Automate)
- AI run locally in-house (on-prem) for especially sensitive data
First a trial, then operation
We start with a working first version (MVP) as a proof of value – tested on real user questions, with an honest recommendation on whether productive use is worth it.
Chatbot & Voicebot
Ready to integrate – responsibly and transparently
Chatbots, voicebots and knowledge bots can be added on this basis. They are disabled by default; no data is sent to third parties until you consciously start a function and consent.
Chatbot / knowledge bot
Answers questions from your own content and cites the source – labelled as AI (EU AI Act).
- Entered messages / questions
- technical connection data if applicable
- no processing without conscious start + consent
Voicebot
Phone/voice assistance for recurring matters – with a recording and AI notice.
- Voice/audio data during the call
- telephony/connection data if applicable
- consent and notice before the call
Before activation we clarify data flows, data processing agreements, third-country questions and the EU AI Act labelling obligation (from 2 Aug 2026). Details in the integration concept.
Matching packages
How we can work together
Delivery On request
AI Use-Case Lab
Turn a good idea into a ready-to-build concept: data check, solution sketch, effort/value, go/no-go.
2–3 weeks Learn moreEnablement On request
Microsoft 365 Copilot Enablement
A structured, responsible Copilot rollout – from readiness and use cases to team enablement.
3–6 weeks Learn moreDelivery On request
AI Agent / Knowledge Bot MVP
A working assistant, tested on your data – with source citations, data protection and an honest evaluation.
4–8 weeks Learn morePilot → Production On request
AI Pilot Guidance
Closes the gap from pilot to production – hands-on, with success criteria and user enablement.
6–12 weeks Learn moreFAQ
Frequently asked questions
What if the AI makes things up?
That is called "hallucinating". We counter it like this: the assistant answers from your own documents and cites the source, plus safety guardrails and a quality test with real questions. We name the limits openly – it cannot be ruled out entirely.
Does our data stay secure?
Data protection is part of the solution from the start. For sensitive data the AI can also run entirely in-house (on-prem); it is documented which data goes where.
Want to try out an AI assistant?
We build a first working version on your data – as a robust proof that it works.