Founders
Inbox and calendar
- The agent
- Sorts incoming email, drafts replies, and prepares a short briefing before each meeting.
- You
- Approve the replies and decide what matters.
Services
What I can do
for you.
I help people and companies go from asking AI questions to letting it do real work. Here are the three ways I work, examples of what gets built, and how a project runs.
Your own AI setup
For founders, freelancers, managers, and anyone whose week is full of small jobs that repeat.
- What happens
- We start with your actual week: your inbox, calendar, documents, and the tools you use. Then we set up agents and automations for the repeating parts, in accounts you own. I explain each step as we go, so you know what runs and why.
- What you end up with
- A working setup with a handful of automations you use every week, short notes on how each one works, and enough practice to change them or build the next one yourself.
- How long
- Two to four sessions, in person in Frankfurt or on a video call.
For example
- Your inbox sorted every morning, with draft replies waiting for your OK
- Meeting notes turned into tasks and follow-up emails
- Receipts and invoices filed and ready for your accountant
- A research agent that collects and summarizes sources on a topic
- Your week planned from your calendar and task list
Automation sprint
For teams that lose hours every week to work that follows the same pattern.
- What happens
- I spend the first days with the people who do the work and write down what actually happens, including the exceptions. Together we pick the jobs where automation pays off and a mistake is easy to catch. I build them in the tools your team already uses. Then we run them next to the old way until they’re reliable.
- What you end up with
- A few automations running in daily use, documentation in plain language, and a person on your team I have trained to look after them.
- How long
- Two to four weeks.
Forward-deployed engineer
For companies that want AI inside their operations: production, inventory, purchasing, sales, and service.
- What happens
- I work with your people, systems, and data for a longer stretch. There, agents can do much more than office work. They can watch stock and supplier delays, prepare orders, sort service requests, and connect systems that never talked to each other. Where your data supports it, I also build custom machine learning models, for example to forecast demand or to spot production problems early.
- What you end up with
- Agents and software built into how your company runs, a team that knows how to work with them, and a clear list of what to automate next.
- How long
- Three months or more, on site or remote.
“Forward-deployed” means I work inside your business, next to the people who will use what gets built, instead of building it somewhere else and handing it over.
Examples
What this can look like
Automations that can be built with today’s tools, sorted by role. They are examples, not past client projects.
Sales
New leads
- The agent
- Researches each lead, updates the CRM, and drafts a first reply and the follow-ups.
- You
- Approve outreach and have the conversations.
Operations
Orders and exceptions
- The agent
- Watches orders across systems, spots delays or missing data, and opens the right ticket.
- You
- Handle what it can’t resolve.
Finance and admin
Invoices and receipts
- The agent
- Reads incoming invoices, matches them to orders, and prepares payments and reminders.
- You
- Approve every payment.
Customer service
Customer emails
- The agent
- Answers routine questions within your rules and drafts replies for the rest.
- You
- Handle complaints and unusual cases.
Inventory and purchasing
Stock and suppliers
- The agent
- Tracks stock and supplier lead times, flags likely shortages from expected deliveries, and drafts purchase orders.
- You
- Approve orders and look after supplier relationships.
Production
Problems that keep coming back
- The agent
- Connects production, quality, and maintenance records and flags patterns that came before earlier problems.
- You
- Investigate and decide what to change.
Management
Weekly and monthly numbers
- The agent
- Pulls figures from your tools, explains what changed, and posts the update.
- You
- Read it and answer the open questions.
Research
Market and competitor news
- The agent
- Collects sources on a schedule, summarizes what’s new, and keeps the links.
- You
- Decide what to act on.
Try it
Is this job ready for an agent?
Think of one job that comes back every week, then answer three questions.
Answer all three to see the result.
How a project runs
Four steps, in this order
Watch the work
I sit with the people who do the work and write down what really happens, including the exceptions nobody mentions in a meeting.
Pick the right jobs
Not everything should be automated. We pick work that repeats, follows rules you can explain, and where a mistake is easy to catch before it does harm.
Build it in your tools
Agents work in what you already use: email, calendar, CRM, accounting, spreadsheets, and your own databases. Your team learns a new tool only when it’s worth it.
Run it, then hand it over
We run it next to the old way until it’s reliable. Then I document it and train the person on your team who will look after it.
Ground rules
Your data, your rules
- Agents get the access they need for the job and nothing more.
- Anything that sends, pays, deletes, or publishes waits for a person until you decide otherwise.
- I use business plans whose providers don’t train their models on your data, and I tell you where your data goes.
- Every run leaves a record, so you can see what an agent did.
- If a job isn’t a good fit for AI, I’ll tell you.
Questions
Common questions
Do I need to be technical?
No. You need to know your work. I handle the technical side and explain it in plain language.
Which tools do you use?
Mostly tools from Anthropic and OpenAI, such as Claude, Claude Code, ChatGPT, and Codex, plus the APIs and connectors that link them to your systems. I choose them to fit your setup and your data rules.
What does it cost?
It depends on the scope. After a first conversation, I send a proposal with what I’ll build, how long it takes, and a fixed price.
What if the AI makes a mistake?
It will, sometimes. That’s why people approve the steps that matter, every run is recorded, and we start with work where a mistake is easy to catch.
Will this replace people on my team?
The aim is to take repetitive work off people, so they can spend their time on the parts that need judgment. What you do with that time is your decision.
What happens after you hand it over?
You own everything: the accounts, the automations, and the documentation. If you like, I stay available for fixes and changes. If not, your team takes over.
Will I need new subscriptions?
Often not. Many automations run on the ChatGPT or Claude plan you already have. Some need a business plan, an API account, or a connector, and I tell you before we start.
Can we start small?
Yes. One automation that saves a few hours a week is a good first project. It also shows you what working with agents is like.