The first two products, in pilot with companies.

Processes that stay manual consume hours every week: copying data between systems, hunting for information in different places, repeating steps that should already be solved. The larger the team, the more the cost grows without anyone being able to point to a number.

A bespoke automation project. We don't sell licences or platforms: we start by mapping the process with you, design the automation around the tools the team already uses, and build it in stages — with closed prototypes your team tests against real cases before moving on.

1. Mapping

   One or two sessions to choose the process, measure the time it consumes and define the expected result.

2. Design and proposal

   Before building, your company knows exactly what the automation does, what it does not do and what it costs.

3. Staged build

   A closed prototype first; the team tests it with real cases and whatever does not work is fixed before the next step.

4. Handover and training

   The automation is documented, and the people who will use it know how to operate it.

• Small teams with manual processes that no longer scale.
• Companies using several tools — email, spreadsheets, CRM, ERP — with data copied between them by hand.
• Operations where a manual error costs time or money to correct.

• We don't replace judgement: decisions and exceptions stay with people.
• We don't promise “AI for everything”: if a process does not justify automation, we say so.
• We don't support third-party tools outside the project scope.

One price per project, fixed after the mapping. No licences and no mandatory monthly fees during the project. If it makes sense to continue after handover, we agree that separately.

Book a mapping conversation. In that conversation we work out whether the process justifies a project — and we say honestly if it does not.

Invoices, receipts, proofs of payment, transport notes and supplier documents arrive every day — and the same data ends up being entered by hand into several systems. It is slow, repetitive work, easy to duplicate.

A tool that receives documents by email or upload, reads the relevant fields with OCR and AI, identifies possible duplicates and prepares the data for human validation and export. The system does the heavy lifting; the decision stays with people.

1. Receipt

   Documents arrive by email or upload, the way they already arrive at the company today.

2. Extraction

   OCR and AI read the relevant fields: date, amount, supplier, document number and the fields the team defines.

3. Duplicates

   The system flags possible duplicates before continuing — the same document does not enter twice.

4. Human review

   A person confirms or corrects what was extracted. The system never marks anything as validated on its own.

5. Export

   Validated data leaves ready for the systems where the team works, with the integration defined during the pilot.

• Importers and distributors.
• Transport and logistics companies.
• Accountants and accounting firms.
• Finance departments.

• Human review is mandatory: the system prepares, the person validates.
• We do not claim tax compliance for any document or process — that validation is done by those with the legal and technical competence for it.
• Evaluation documents are shared anonymised; production data is handled under a data processing agreement.
• Nothing is used to train models on client data.

In the pilot, the cost is agreed case by case. After that, the model is tailored: a monthly fee per company, by document volume, or an integration project.

We are selecting a small group of companies. The pilot measures current processing time against the time with the tool, using anonymised documents — and the decision to continue is made with numbers, not promises.