Terms of Service.

The full legal text follows below and prevails in the event of any disagreement.

1. parties

Supplier: Martin Nørmark Hansen, a sole proprietorship trading under the brand algoritma, address to be updated upon registration, CVR (Danish business registration number): to be updated once the business is registered, martin@algoritma.dk.

Customer: the business or private individual accepting a quote from algoritma, whether verbally or in writing.

These terms apply to all agreements unless otherwise agreed in writing. In the event of a conflict between these terms and a separate project contract, the project contract prevails.

2. services and process

algoritma delivers ML and automation solutions in three phases:

Spot (free): a 30-minute video call + a written report with up to 3 concrete proposals, an impact estimate and a price assessment. No obligation for either party.

Prototype (free): a working demo of one well-defined feature, delivered as a live demo, hosted preview or video walkthrough. The prototype is delivered within 1-4 hours and is not production-ready. Its purpose is to reduce the customer's decision risk.

Build (paid, fixed price): a production-ready ML solution delivered within 1-4 weeks. Scope, price and acceptance criteria are specified in the project description, which forms the contractual basis.

Services beyond the above, including ongoing support, maintenance, training and additional tasks, are agreed separately and invoiced at the prevailing hourly rate or at a fixed price following a prior quote.

3. quotes and acceptance

Quotes are valid for 30 days from dispatch unless otherwise stated. Acceptance may be given in writing (email, message) or by payment of the deposit. Verbal acceptance is binding but should be confirmed in writing.

algoritma reserves the right to decline assignments that do not match its competence profile, or where availability does not permit responsible delivery.

4. prices and payment

Fixed price: all build projects are priced at a fixed price based on the agreed scope. Price changes require written acceptance from both parties.

VAT: all prices are exclusive of Danish VAT (25%) unless explicitly stated otherwise. B2B customers in Denmark are charged VAT on the invoice. EU businesses with a valid VAT number are invoiced without VAT (reverse charge).

Deposit: for projects above €3,300 excl. VAT, a 40% advance payment is charged at project start. The remaining 60% is invoiced upon delivery and approval.

Payment term: 14 days net from the invoice date, unless otherwise agreed.

Late-payment interest: in the event of late payment, interest accrues at the official lending rate of the Danish central bank + 8 percentage points p.a. from the due date, in accordance with section 5 of the Danish Interest Act. The rate is adjusted every six months (on 1 January and 1 July).

Reminders: a first reminder is sent 7 days after the due date has passed. If payment remains outstanding, work may be suspended and the matter referred to debt collection.

5. delivery and approval

Form of delivery: code is delivered via a private git repository (GitHub/GitLab), as a zip file or directly into the customer's existing system. Hosting, deployment and operations agreements are not included unless explicitly agreed.

Acceptance period: the customer has 7 working days from the day of delivery to report errors and defects within the agreed scope. Errors must be reported in writing.

Bug-fix guarantee: errors documented within 30 days of delivery and directly related to the delivered scope are remedied free of charge. This guarantee does not cover errors arising from the customer's own changes, third-party updates or changed data patterns.

Changes to scope: changes or additions to the agreed scope (change requests) require a new written quote and may result in an adjusted price and timeline.

The customer's cooperation: the customer undertakes to make the necessary data, access and contact persons available in good time. Delays caused by a lack of cooperation on the customer's part result in a corresponding postponement of the delivery date without this constituting a breach.

6. intellectual property rights

Upon full payment, algoritma transfers all rights to the code developed specifically for the project and the models developed for it. The customer receives full ownership. There are no licence fees, no vendor lock-in and no restrictions on internal or commercial use.

Open source and third-party components: solutions may contain open source libraries and pre-trained models with their own licences (typically MIT, Apache 2.0 or similarly permissive licences). algoritma informs the customer of material licence requirements. The customer is responsible for complying with these licences in its use.

Third-party API providers: solutions may use third-party cloud APIs (e.g. OCR, language models, vision APIs) to perform specific tasks. algoritma informs the customer of the relevant API providers as part of the project description. If these APIs process personal data as part of the solution's function, the customer, as data controller, is responsible for ensuring that the necessary agreements are in place with the relevant providers. algoritma recommends and exclusively uses APIs from providers offering GDPR-compliant data processing agreements (e.g. Google Cloud, AWS, Azure, Anthropic).

algoritma's generic knowledge: algoritma retains the right to use general knowledge, methods and experience gained during the project in future work, provided no confidential content is disclosed.

Portfolio: algoritma may not mention the customer publicly, nor show code, results or details of the project, without written permission. This applies unconditionally both during and after the project period.

7. confidentiality

Both parties undertake to treat all non-public information from the other party as confidential. This applies to information about business affairs, technical systems, customer data, strategies, prices and project content.

• The confidentiality obligation applies during the project and for 3 years after its completion.
• algoritma never passes the customer's data to third parties without explicit written permission.
• algoritma is happy to sign the customer's standard NDA as a supplement to these terms, provided the NDA contains no provisions in conflict with these terms.
• The confidentiality obligation does not apply to information that is publicly available, has been obtained independently from a third party, or must be disclosed by law or by order of a public authority.

8. data processing and AI tools

algoritma processes personal data only to the extent necessary for the performance of the assignment. See the privacy policy for details on GDPR handling.

If the project involves algoritma processing personal data on the customer's behalf, a separate data processing agreement (DPA) is entered into before project start. algoritma does not use the customer's personal data to train its own or third-party AI models.

Use of AI development tools: algoritma uses AI-based development tools (including, among others, Anthropic Claude and similar) as part of its working process. These are used exclusively on anonymised, non-identifiable data or on data algoritma has generated itself. algoritma never enters the customer's personal data, trade secrets or confidential data into external AI services without the customer's prior written permission. The AI services algoritma uses in a commercial context are covered by data processing agreements (DPAs) compatible with the GDPR, including Article 28 of the General Data Protection Regulation.

9. limitation of liability

Direct loss: algoritma's total liability for a project is limited to the amount invoiced for the specific deliverable giving rise to the claim.

Indirect loss: algoritma is not liable for indirect losses, including loss of business, loss of profit, loss of data, damage to reputation or other consequential losses, whether or not such losses were foreseeable.

Third-party systems: algoritma is not liable for errors, outages or changes in third-party systems, cloud platforms, APIs or open source components forming part of the solution.

Changed conditions: the performance of ML models depends on data quality and data patterns. algoritma does not guarantee specific accuracy targets unless this is explicitly agreed, with test conditions and acceptance thresholds stated.

The limitation of liability does not apply in cases of gross negligence or intentionally harmful acts.

10. termination

Termination by agreement: either party may terminate the project with 14 days' written notice. The customer pays for work performed up to the time of termination, plus any expenses already incurred.

Termination for breach: in the event of material breach, the non-breaching party may terminate the agreement with immediate effect, subject to written notice and a reasonable remedy period of 5 working days.

Upon termination: work performed up to the time of termination is invoiced. The customer is entitled to receive all code and documentation that has been produced and paid for.

11. force majeure

Neither party is liable for delays or non-performance caused by circumstances beyond that party's reasonable control, including outages at cloud providers, natural disasters, illness, power failures or similar extraordinary events. The affected party must notify the other party in writing without delay.

12. governing law and disputes

These terms are governed by Danish law. The parties will first seek to resolve any dispute amicably. If agreement cannot be reached, the matter is settled by the Court of Odense as the agreed venue.

13. EU AI Act & compliance

algoritma works in accordance with the EU regulation on artificial intelligence (EU 2024/1689, the "AI Act"). The following clarifies the parties' obligations.

AI competences (Art. 4): algoritma holds documented competences within machine learning and artificial intelligence (MSc Data Science, SDU). These competences satisfy the AI Act's requirement for adequate AI skills among those working with AI systems on the business's behalf.

Risk classification: most ML solutions delivered by algoritma (recommendation systems, automation, forecasting, document processing) are classified as minimal or limited risk under the AI Act. algoritma informs the customer in writing if a project touches on a high-risk use case (cf. Annex III of the regulation), including credit assessment, hiring decisions, biometric identification or critical infrastructure.

Transparency towards the customer (Art. 13): as part of build projects, algoritma provides adequate documentation of the delivered AI system's purpose, limitations, expected performance and known risks, to an extent proportionate to the project's complexity.

Transparency towards end users (Art. 50): if a delivered AI system communicates directly with the customer's own users (e.g. a chatbot or automated decision support), the customer, as deployer, is responsible for informing those users that they are interacting with an AI system. algoritma informs the customer of this obligation at project start.

Data processing for AI training: algoritma does not use the customer's data to train its own or third-party models without explicit written permission. See also section 8.

14. changes to these terms

algoritma may update these terms. Material changes are notified to active customers with 30 days' written notice. For new projects, the version in force on the quote date applies.

Version 1.1 · last updated: 2026-05-05