Modern tools, accountable people.

1. Purpose of this statement

This statement describes MNessa Limited's general approach to artificial intelligence, machine-learning services, code assistants, automated analysis and similar tools. It is a public statement of principles, not a technical specification or a promise that every project will use AI.

Project-specific requirements, prohibited tools, data locations, confidentiality restrictions and review standards should be recorded in the relevant proposal, contract or data-processing arrangement.

2. How AI may be used

Depending on the task and client agreement, AI-assisted tools may support:

• early-stage research, ideation and comparison of technical approaches;
• drafting, summarising or restructuring non-final text;
• code suggestions, refactoring, test generation, documentation and debugging;
• data classification, extraction or analysis where the inputs and risks are appropriate;
• prototyping interfaces, workflows and automation concepts; and
• quality checks that complement, rather than replace, suitable human review and testing.

AI output can be incomplete, inaccurate, biased, insecure or inconsistent. It is treated as material to be evaluated, not as an authority merely because it was generated quickly or presented confidently.

3. Human accountability

A responsible person remains accountable for work released by MNessa. The level of review is proportionate to the impact of the task and may include checking sources, inspecting code, running tests, reviewing security, validating calculations and confirming that the output meets the agreed requirement.

MNessa will not describe unreviewed machine-generated material as verified professional work. Use of AI does not transfer responsibility for scope, quality, confidentiality or legal compliance to the tool provider.

4. Confidentiality and data protection

MNessa applies data minimisation when using AI services. Confidential, personal, credential, production or commercially sensitive information should not be entered into a general-purpose public AI service unless there is a legitimate need, an appropriate legal basis, suitable contractual and security controls, and any required client approval.

Where AI use involves client or personal information, MNessa should consider:

• whether anonymised, synthetic or reduced data can achieve the purpose;
• whether the provider uses inputs or outputs for model training;
• retention, access, sub-processors and international transfers;
• the client's instructions and the role of each party under data-protection law;
• security, deletion and audit options; and
• whether a data-protection impact assessment or additional agreement is appropriate.

5. Accuracy, testing, security and bias

Review should reflect the consequences of error. Public copy may need factual and source checks; code may need static analysis, dependency review, tests and manual inspection; data analysis may need validation of assumptions, completeness and reproducibility.

MNessa aims to consider whether training-data limitations, prompts, evaluation choices or deployment context could create unfair or misleading outcomes. A tool's output should not be used to conceal uncertainty or fabricate sources, evidence, credentials or results.

6. Automated decisions affecting people

MNessa does not use the public website enquiry form to make solely automated decisions that produce legal or similarly significant effects. MNessa will not deploy a high-impact automated decision process for a client without clearly defining responsibility, lawful basis, human oversight, testing, challenge routes and other safeguards appropriate to the context.

7. Intellectual property and provenance

AI-assisted work must still be assessed for ownership, licensing, attribution and third-party rights. MNessa should avoid knowingly requesting or reproducing protected material in a way that infringes rights and should not assume that generated output is automatically original, exclusive or safe to commercialise.

Client ownership or licensing of deliverables remains governed by the applicable contract. Where provenance or exclusive authorship is a material requirement, the project should define a suitable workflow and record.

8. Client choice and disclosure

MNessa will discuss material AI restrictions when they are relevant to confidentiality, regulated work, intellectual property, data location, procurement requirements or the nature of the deliverable. Clients are encouraged to state any organisation-wide AI policy at the start of an engagement.

Routine use of an AI-assisted coding or productivity tool may not always be separately itemised, just as ordinary use of a compiler, search tool or spell checker is not normally itemised. However, MNessa should be transparent where AI materially shapes a deliverable, performs a client-facing function or creates a risk that a reasonable client would expect to understand.

9. Technology suppliers

AI and automation providers change rapidly. MNessa may review supplier terms, security information, model behaviour, administrative controls, sub-processors and data-location options before approving use. Approval for one task does not mean a tool is suitable for every project or data category.

10. Governance and review

This statement will be reviewed as MNessa's services, client profile, technology and applicable guidance develop. Material incidents, recurring errors or supplier changes should trigger an earlier review. Questions or project-specific AI requirements can be sent to contact@mnessa.co.uk.