Does my data go to OpenAI or other providers?

Only if you want it to. For sensitive data, I set up a local model; for non-critical tasks, a cloud service can be cheaper. You decide per workflow, and it's documented what goes where.

Does the AI write to my customers on its own?

Not by default. The AI prepares a draft, a human approves it.

What does a pilot cost?

A fixed price after a short initial consultation. The pilot is deliberately kept small, so you can see whether it's worth it without much risk.

Do I need new software?

Usually not. I connect what you're already using.

Automation · AI assistant · Local language model · Email · Chat · Integrations

AI & Automation

Automating recurring office work, with AI that doesn't hand your data to third parties unless you want it to.

Person checking a tidy email inbox on a laptop in the office

Start with one workflow, not a strategy

Many businesses know AI can save time, but not where to start, and have concerns about data protection. The Bitkom study KI in Unternehmen 2026 names exactly this as the biggest hurdles: data protection and a lack of skilled staff.

I start with a single, measurable workflow and only expand it once it proves worthwhile. What happens to your data is defined for each workflow individually. Data protection questions are handled by your law firm or data protection officer; I take care of the technical implementation.

Service

What I set up

Pre-sorting inquiries: Incoming emails, form and chat inquiries are recognized, sorted and presented with a draft reply. Nothing is sent until you approve it.
Customer chat on website and messengers: A shared inbox for website chat, email and messengers, with AI-suggested replies on request.
Receipts and documents: Automatically reading, naming, filing and forwarding invoices and scans to bookkeeping.
Connecting programs: Transferring data between your website, calendar, spreadsheets and inventory system instead of retyping it.
AI on your own premises: A local language model on your own hardware, when data shouldn't leave the business.
Team onboarding: A short training session and clear rules on what may go into AI tools and what may not.

Process

How it works

Choosing a workflow
A process that occurs often and eats up time, for example pre-sorting inquiries.
Pilot over two to four weeks
With a measurable target, for example minutes per inquiry before and after.
Evaluation
Is it worth it? You see the numbers and decide.
Operation and expansion
The workflow runs, is monitored and, if useful, extended with the next one.

Who it's for

Who this is for

Businesses with many similar inquiries: appointments, quotes, follow-up questions
Offices with a lot of paper and receipts
Owners who want to use AI but don't want to hand their data to every cloud service
Online retailers with customer inquiries across multiple channels

Practice

Practical example: knowledge base with a local model

An AI search across your own documents that runs without cloud AI: built in a test environment, with measured response times and honestly stated limitations.

Read the practical example

Practice