JustPaid vs. e.on: Germany’s laws could be the secret 'System Prompts' for our energy revolution.
Stack upgrades; permission problems; Engineering State vs. Lawyerly Society; opportunity: we aren't over-regulated; we are pre-specified.
I’m currently susceptible to AI case-studies while upgrading the Stratnergy stack. A larger Hetzner instance, OpenClaw upgrades, experiments with pro-subscriptions, new agent-optimised models that I can afford to keep on.
The Wall Street Journal’s piece on how JustPaid is using agents made me wonder whether we’ll ever see an article like that for a company in the German Energy Industry: delegated software work, agentic coordination.
Pinnaka says the creation of his AI agent engineering team followed the natural evolution of Claude Code, which was dramatically improved when Anthropic released the Opus 4.6 model in early February. Then, as the OpenClaw craze began to sweep Silicon Valley, and beyond, the idea popped into his head that he could build an “AI version of an engineer.”
Articles about companies using AI and ‘agents’ are a dime a dozen these days. You’re forgiven for stifling a yawn. But I think the question of how far away we are from something like this in the German energy industry gives interesting top-spin.
I hear at least one of the optimisers is going all-in on Claude Code. Good on them, especially since I think they are set up to make best use of it. More widely, your average corporation is probably struggling to put the puzzle together.
I’ve heard of an energy organisation trying to get “German AI” to work for them. Laudable, but I wouldn’t make that bet before achieving a cruising altitude with the best options. Better chances with “French AI?” The WTO may be dead, but Ricardian principles probably apply for synthetic ‘intelligence.’ Where does that lead?
What is it that makes the WSJ story interesting? Not OpenClaw as such. The extent. In German energy, ‘extent’ is the whole question… To what extent can I use that grid connection? To what extent has the system brought all parties together to use the latest technology for the best solution.
Right now, our traditional energy systems are like old-school telephone landlines – clunky, inefficient, and expensive. But with AI, we’re moving to the smartphone era, where everything is optimised in real time.
AI can adjust energy flows based on what’s happening right now, making the grid significantly more efficient. This is especially important for integrating renewable energy sources like wind and solar which can vary in output.
Germany will get there when firms stop treating law as a brake and start treating it as part of the design. What does progress look like?
A bit of AI-chat, an internal enterprise assistant.
Progress to embedded copilots. Guarded workflow automation? Polite applause.
Development agents with autonomy and repo access…
Autonomy near production, operational technology, critical operations.
Turn the law into a resource and Germany may have a unique advantage to confidently skip ahead through these stages. For decades it has invested heavily in writing and re-writing its laws, rules, and codes. Think of these as the MD files to help agents structure and understand the solution space.
Suddenly, the quality of your specs determines the quality of your product. Ambiguity that a senior engineer used to resolve intuitively now produces bugs at machine speed. So we started investing heavily in what happens before a single line of code is written.
We spend more time understanding the domain, shaping the product, and writing precise specifications. Communication skills, product sense, and judgment matter more than ever. Vasily Geyer at Entrix
Germany has invested in writing the specs. Now the solutions can begin to grow.
