Business case
Germany runs 2.6 million battery systems holding 31.5 GWh. The average one is twelve kilowatt-hours. Almost none of them is marketed, and the reason is arithmetic: taking on one asset costs more human time than the asset earns in a year. So the flexibility sits there, and the companies whose business is marketing flexibility turn it down.
Reality Pack makes that arithmetic work. It turns the rules an asset has to obey — technical limits, grid operator requirements, contracts, the owner's own instructions — into something a machine reads and a machine can be held to. One agent then takes on a thousand assets without a human checking each one. The customer is the direct marketer. The product is the layer that lets them say yes to the assets they have to turn down today.
The market
The growth rate is the part that decides the timing. This fleet is getting larger faster than anyone is building a way to use it, and every unit added is one more that is too small to onboard by hand.
The picture
The case is a statement about size. Below is the band we are built for, and above it the end of the market we leave alone.
31.5 GWh across 2.6 million systems is 12.1 kWh each — both figures from the Marktstammdatenregister as evaluated by the Bundesnetzagentur, mid‑2026.
Where we do not play
A business case that claims the whole market describes none of it. These two ends belong to someone else, and saying so is what makes the middle ours.
Tolling and floor constructions over ten years hang on the site, the asset structure and the financing. That is individual casework, and each deal is large enough to pay for every reviewer it needs. A machine-readable boundary saves nobody there.
Deterministic rules already govern it, penalties for non-delivery among them, and the systems that enforce them work. There, Reality Pack is a layer above — orchestrating, not enabling. That is a different product and a later one.
And the band between them comes with us. From 100 kWh to 1 MWh sit the larger commercial sites — hotels, light industry, depots. They need the same rulebook, the same five agent jobs and the same portfolio machinery, so a direct marketer who has them takes them along at no extra cost to us. They are simply not where the number is: Germany holds 20,600 commercial storage systems in all, from 30 kWh to 1 MWh, against 2.6 million at home. The segment line sits at 100 kWh because that is where the population is, not because the product stops there.
The customer
A direct marketer or aggregator markets assets it does not own. That is the whole business: thousands of other people's installations, brought into one portfolio and sold into the market.
The effort per asset barely falls with size, and the margin does. Every small installation needs the same registrations, the same grid-area checks, the same chasing of a connection that does not answer — for a fraction of the revenue. Below a certain size the answer has to be no, and today that line sits far above 100 kWh.
Because here the agent is foreign and the asset is foreign. An operator running its own fleet with its own software answered the permission question once, for itself, and it stays answered. A marketer answers it again for every asset it takes on, from a different owner, under a different grid operator, with different contracts.
The work
None of this is trading. Trading is solved and automated. What is left is admission, communication and settlement — and that is where the people are.
The agent triggers the registrations, checks what the particular grid area demands, and decides whether the asset can enter the portfolio at all.
It clears the communication link — is there a line, is the asset reachable — and writes to the owner and the line provider until there is an answer.
It sorts an assorted fleet into the portfolios that fit and decides, per asset, how it can be controlled and marketed.
It notices when something is off and opens the clarification or the warranty case with the manufacturer.
It estimates how much longer an asset will run and when a replacement has to be bought.
The economics
That single sentence is the business. A clarification on a small asset — one call, one letter, one follow-up — runs to €20 to €200, and an onboarding is several of them. Against the annual contribution of a twelve-kilowatt-hour battery, that is the whole margin and more.
The consequence
There is no version of this business with a human in the loop per asset. Either onboarding and operation run end to end without one, or the segment stays shut. That is not an aspiration, it is the entry condition.
What the pack delivers
A machine can only be left to it if the boundaries are machine-readable, and that is what the pack is. Every verdict carries the rule it rests on and the source behind it, so the marketer can still answer for what happened on an asset they do not own.
The external blockers stay external. A missing internet line, an owner who does not reply, a grid operator who takes six weeks — no software removes those. What an agent does is pursue them: track every open thread, ask again on schedule, and never drop one because a person went on holiday. That is most of what the chasing actually is.
The product
A proposed action comes back allow, deny or unknown. The third is what makes the first two usable: an open question stays open and is counted, instead of being guessed at by whichever system asked.
Each reason carries the rule, its layer, the explanation, and the source with provision, retrieval date, document version and quotation. That is what a marketer needs when an owner or a grid operator asks why an asset was not controlled.
The asset is described once, across five layers, and every agent works against the same boundaries without knowing them in advance. That is how one onboarding scales to a thousand without a thousand integrations. See the agents at work.
What happens next
A direct marketer or aggregator with small assets they currently turn down. We bring the layer, the gate and the agents; what we need is one real portfolio and the grid areas it sits in.