Home · Blog

Blog

What we learn from putting agents into production.

Notes from real projects: where process context decides whether AI works, what governance costs and what it saves and which questions keep coming up in procurement.

Who Was Driving?

A fine for a company vehicle and no driver named on it. The authority has six months, you have fourteen days — and they start at a post room, not at a desk.

  • Fine management
  • Deadlines
  • Fleet

Read the article →

Which Contract Applies?

Four files in one supplier folder, three of them called final. Why a notice period attached to a document instead of a process renews the contract for you.

  • Contract management
  • Deadlines
  • Audit trail

Read the article →

One Wrong Digit.

Two digits change places in a VAT ID. Nobody notices until it is on an invoice — and for intra-Community supplies the exemption can depend on it.

  • Customer registration
  • Process context
  • Audit trail

Read the article →

Stop Running Pilots. Start Running Processes.

Most AI pilots never reach production. The reason is rarely the model — it is that nobody ever told it how the business actually works. What it takes to move from a pilot to a process that runs every day.

  • Process context
  • Data sovereignty
  • Open LLM integration
  • Identity and access

Read the article →

Notes from production. Three of these pieces form one series. Each takes a workflow that runs in production and follows a single small object through it — a transposed digit, a folder of contracts, a notice with no driver on it — to the point where a process model stops it. There is no fixed schedule: we write them when a workflow gives us something worth writing about.

Have a process worth writing about?

If you are solving one of these problems, we would like to hear how. Some of the best material comes out of those conversations.

Book a conversation

DYNAMIXS.AI GmbH · Munich · info@dynamixs.ai · +49 89 452136-22