Terms & Conditions

The terms that govern a ZenAI engagement: who owns what is built, what is warranted, what we need from you, and what a proof of concept does and does not commit either side to.

Effective date · to be set at publication Superseded by any signed engagement agreement

Draft for legal review

The AI-specific clauses below are drafted for review. General clauses covering site use, liability limits, governing law, and dispute resolution are inherited from the Zendynamix terms and must be reconciled with this document before publication.

01

Model and IP ownership

Ownership of models built for a client, including fine-tuned weights and derived artefacts, is defined in the engagement agreement and settled before work starts.

Client data, and any model whose value derives solely from that data, remain the client's. Pre-existing ZenAI and Zendynamix platform components, tooling, and frameworks used to deliver the work remain ours, and the client receives a licence to use them as part of the delivered system.

Where an engagement produces a general improvement to our own tooling that contains no client data and no client-specific logic, that improvement remains ours.

02

Accuracy disclaimers

No model is warranted to be accurate. Statistical systems make errors, and their error rate changes as the data and the operation change.

Unless the engagement agreement says otherwise, systems we deliver are decision-support: they inform a decision that a person or a client-controlled process remains responsible for. Where a system is contracted to act autonomously, the boundaries of that autonomy, the accuracy targets, and the monitoring obligations are set out in that agreement.

Nothing we deliver is a medical device, a financial advice service, or a certified safety system unless expressly contracted and classified as such.

03

Client responsibilities

A model is a function of the data and the rules it is given. The client is responsible for:

  • the quality and completeness of data provided;
  • having a lawful basis for providing that data to us, including any personal data it contains;
  • the accuracy of the operational rules, constraints, and business logic supplied to us; and
  • decisions taken on the output of a delivered system, within the autonomy boundaries agreed.

04

Proof-of-concept terms

A proof of concept is an evaluation. It does not constitute a production commitment on either side: the client is under no obligation to proceed to a build, and we are under no obligation to take a system to production if the proof of concept shows it shouldn't be.

A proof of concept is not a production system. It carries no availability commitment, no support obligation, and no warranty of fitness for operational use, and it should not be relied on to make live business decisions unless separately agreed.

Data handling during a proof of concept is covered in the Privacy Policy.

05

Contacting us

Questions about these terms can be sent to [email protected].

Book a solutioning session

Tell us the problem you want AI to solve.

A solutioning lead replies within one business day. No obligation attached.

We'll only use your details to respond to this request.