Two channels, reliably served.
The team barely touches it.
From May to August 2026 a law firm in the DACH region published every week without anyone touching the tools. Before, that meant manual work spread across tools and people. We built one system, wired it into their Teams and channels, and we ran it. The team approved with a click, the rest happened on its own. This page describes that operation, not a demo of what an AI can write.
73 seconds, the real workflow.
A real recording from the running system: the lawyer picks a topic in the chat, the finished post appears for approval. Shown with a fictional demo firm, the workflow is the real one.
The work that disappeared.
Same output, on two channels, every week. What changed is who does the work: nobody, by hand, anymore.
The AI part is the easy part.
A strong model writes a decent draft in seconds, and everyone has one now. That is not what we sell. The value is everything around it: that it was wired into their systems, that a human approved, and that it ran on its own for three months because we operated it and were responsible for it. The intelligence is a component. An operation someone stands behind is not.
Measured, not estimated.
Wired in once, then it runs.
Connected to their tools
The system sits inside their existing Teams and talks to their channels directly. No new tool for the team to learn, no extra login, no data leaving their world.
Draft and approval
Drafts are produced against a fixed rule set (voice, mandatory and forbidden terms) and land where the team already works. One click approves or corrects.
Published automatically
Approved posts go out across the channels on their own, reliably, on schedule. Nobody copies anything between tools anymore.
Operated by us
We monitor it, fix it, and keep it working as models and APIs change. It is our job to keep it running, not theirs.
What firms ask us first.
Does this really work, or is it a nice demo?
What you see above came out of a real operation: 19 posts went out through it between May and August 2026, not a concept. The video is a recording from it. You see the actual output before you commit to anything.
Will the AI match our voice?
We fix your tone, your mandatory and forbidden terms and your taboo topics in the system, calibrated on your own texts. And every post goes through your approval before it appears. Nothing goes out unchecked.
Who is behind this, and what if you drop out?
The value is not headcount, it is an operated, monitored system with clear responsibility. That it holds, you see from the fact that it has run at a client for months without an outage. Keeping it running is our job, not yours.
Where does the AI get the topics?
The lawyer briefly describes their field. From that, the system proposes concrete, current topics, partly with its own research on new rulings. The lawyer picks one, the rest is produced automatically and goes through approval.
What about data protection and client confidentiality?
No case or client data is accessed. The workflow runs inside your own Teams, topics come from a general description of your work, not from files. No client data leaves your house, and every post is approved by you. We provide a data processing agreement.
See it on your own firm.
Start with a live pilot: your first posts in your voice, live on your channels, within two weeks, each one approved by you. Fixed price € 2,900, fully credited against the operation, cancel anytime. No report to read, no long contract, you see the quality on your own firm. And this operation is only one example, the same approach fits any recurring work between your systems.