Berlin · AI systems for commerce, distribution and operations

We build AI systems that run in the business.

Most AI projects stop at a demo. Ours end up in the working day, wired into the systems you already run, with your people approving anything that matters.

Bring us a case How we work 30 minutes. If we can't help, we say so.
Two systems exchanging information across a controlled handover point SYSTEM A SYSTEM B HANDOVER

Your own system

HELYOS technology, set up under your name and on your rules. We build it and we run it.

HELYOS

Custom Intelligence for commerce, distribution and operations. HELYOS owns the product. We handle it in German-speaking markets.

ROVE

Market intelligence for expansion decisions. Rove Global Limited owns the product. We bring it to German-speaking markets.

01What we do

Not every problem needs something built from scratch.

White label on HELYOS

Your own system

The engine is HELYOS, already proven in commerce and distribution. Everything above it is yours: your data, your rules, your thresholds, your name on the interface, on your own domain. We do the setup, the connections and the day-to-day operation. Your people see one system that belongs to the company, not a tool someone bought in.

Partner product

HELYOS

A finished platform beats a new build more often than people expect. Where HELYOS fits, we handle the rollout, the adaptation and the day-to-day in German-speaking markets.

Partner product

ROVE

Before you commit to a market, someone will ask you why. ROVE gives you the answer with the numbers behind it, and we turn that into a plan your company can carry.

Every signal in your systems, boiled down to what to do next.

02Why most of it fails

The model is no longer the advantage.

You have access to the same models as the largest company in your industry. What separates the two of you is not the model. It is what the model knows about your business.

88%

of organisations report using AI. Most of it is a chat window somebody opens, describes a situation to, and closes again. The adoption is there. The embedding is not.

54%

of the recommendations a model gave across 100,000 business situations came down to two sentences: win new customers, raise revenue per product.

500

simulated days with a million dollars of capital. Most models ran the business into insolvency. A simple set of rules beat almost all of them.

2.7%

separates the leading American model from the leading Chinese one. Whatever advantage you are looking for, it is not in that gap.

Sources: Stanford HAI AI Index 2026 for adoption and model gap · study across 100,000 business situations, Udi Menkes, Intuit · Princeton simulation. None of these figures are ours, and none come from our own client data. We publish them because they can be checked.

A fluent answer built on what the internet knows about business will sound right and still be wrong. That is why we start with your margins, your exceptions and your thresholds, and why every recommendation carries a path back to the data it came from.

03Use cases

Four situations we build for.

These four come up again and again. Pick the one closest to your situation and it goes straight into the form.

Prices and listings move faster than the meeting about them

Channels are checked manually, the corrections land late, and the margin loss only shows up at month end.

  • Price position watched continuously across every channel
  • Listings ranked by how much each one is costing you
  • Reorder points that use real lead times instead of last month's average

For e-commerce leads running a shop plus two or three marketplaces

Often delivered with HELYOS

A simple customer question takes three lookups

Customers, quotes, orders and availability sit in separate systems, so nobody has the whole picture at the moment they need it.

  • One view per customer across ERP, mail and spreadsheets
  • Quotes ranked by how likely they are to close, so the old ones stop crowding the list
  • Accounts going quiet flagged while you can still do something about it

For managing directors and sales leads in B2B wholesale

Often delivered with HELYOS

Your best people spend the day on routine work

The work that follows a rule gets done by hand, and the case that genuinely needs judgement surfaces too late.

  • Recurring cases handled end to end, to your rules
  • Exceptions go to a person who can see how the system got there
  • A record of what was decided and what followed

For COOs and operations managers with more manual process than they would like

White label on HELYOS

Go. All three hold. Enter in the planned order.

The method, not a market. Your numbers replace these switches.

You know less about this market than the budget suggests

Demand, competition and channel structure are estimated from a handful of conversations and a spreadsheet.

  • Demand, competition and channel structure per market
  • Go, conditional or no-go, each with its reasoning
  • An entry order sized to the team you have today

For founders and MDs with a new market on the table

Often delivered with ROVE

04How we work

Five steps, in this order.

  1. 01
    Systems and data

    We look at what already exists: shop, ERP, CRM, spreadsheets, inboxes. Skip this and every system after it is a guess.

  2. 02
    Context and rules

    Your margins, your exceptions, your approval thresholds, and the cases where your company decides differently from the textbook.

  3. 03
    Priorities

    All that data turns into a short ranked list. It names the next sensible move and why it comes before the others. Reports you already have enough of.

  4. 04
    Approvals

    Anything that follows a rule simply runs. Anything unusual goes to one of your people with the reasoning attached, and the trail stays readable months later.

  5. 05
    Operation

    Nothing is finished at handover. We record what the system suggested and what happened next, and that record makes the following suggestion better.

Routine work runs. Anything unusual goes to one of your people.

Nobody takes responsibility for a system they cannot explain.

Approvals, exceptions and manual overrides stay with your people. Months later you can still show who decided what, and why.

Your data stays yours. It is held in your environment and exportable at any time. And after six months something exists that did not exist before: the record of what your own business decided and what followed. Not what the internet thinks about discounts — what happened when you discounted in March.

05Work

The proof happens on the call.

We have systems running in commerce and operations right now. They are not on this page yet, because the clients involved have not approved their names in writing, and we will not publish a result figure without the measurement definition, the period and the source behind it.

It makes for a thin reference page. It also means that when a number does appear here, you can check it. On a call we walk you through what is running, in detail.

Commerce and operations systems, in daily use right now.
Built in Berlin. Two partner products carried, our own systems built.

06First call

Bring us a case.
We spend 30 minutes on it.

You describe a real situation, which takes about two minutes. We read it before we talk, and on the call we tell you which part of it is solvable, which part is not, and what a first step would cost. If it is not something we should take on, we say so and point you somewhere better.

Press c anywhere on the page to jump straight here.

Read how a project runs, before you talk to anyone.
Would rather just talk? Pick a time straight away. The written case helps us prepare, it is not a condition.

Step 1 of 4 · Area

What is this about?

Pick the area closest to your case.

Step 2 of 4 · The case

Describe the situation.

A few sentences is enough. Specific helps more than complete.

Step 3 of 4 · Context

A little context.

Optional, but it lets us prepare properly.

Step 4 of 4 · Contact

Where should we reply?

You can pick a time right after sending. No automated sequence either way.

No sales sequence. You pick the time, or we reply within one working day.

Got it.

Your case in is with us. Pick a time now and we will have read it before we talk.

Pick a time

  1. Thirty minutes, whenever suits you. A confirmation is in your inbox.
  2. We read the case beforehand and work out which part of it is solvable.
  3. You get an assessment and a next step, confirmed in writing afterwards. No deck.

Questions

What you want to know first.

Do HELYOS and ROVE belong to ainulo?

No. HELYOS is the brand and product of HELYOS, and ROVE belongs to Rove Global Limited. We look after both across Germany, Austria and Switzerland, from the first conversation through rollout and day-to-day operation. When we build a system under your name, it runs on licensed HELYOS technology. Your data, your configuration and your rules stay yours, the underlying platform stays with HELYOS. We say this on the website so that nobody hears it for the first time in a contract.

Does the AI decide on its own?

No. The systems prepare work and carry out what has been approved. Approvals, exceptions and manual overrides stay with your people, and you can always retrace how a recommendation came about.

What does a project cost?

Operation starts at 499 € a month, with no VAT added and no minimum term. The setup is a fixed price, quoted after the working sessions and never billed by the hour. Everything up to the offer, including the call and the project plan, is free. Most of what we earn comes from a system that is being used, so we lose money when it goes quiet — that is deliberate. The full picture is on the project page.

Where does our data sit and who can reach it?

In your environment, exportable at any time, and never used as training material by us or anyone else. Access is limited to named people and logged. The full picture, including what happens at the end of a contract, is on the project page.

Why you and not HELYOS directly?

Because what you are buying is not only the platform. It is the setup on your data, the connection to the systems you already run, the German-language rollout with your people, the GDPR paperwork, and somebody in your own time zone who picks up when something breaks at half past four. HELYOS builds the engine. We put it in your house and keep it running.

When are you the wrong fit?

If what you want is a chatbot, or if the data behind the decision does not exist anywhere yet, this is not the right call. The same goes for work where nobody inside the company owns the process we would be changing. We would rather say that in the first ten minutes than in month three.

What happens to the details in my enquiry?

They are used to answer your enquiry and nothing else. We do not pass them on and we do not add you to a mailing list. Details are in the privacy notice.

30 min, no pitch Bring us a case