AI, Valuation and Agent Terms

These terms apply to TrueValue's AI-supported research, extraction, analysis, valuation, drafting and Agent features where enabled for your subscription. They supplement the Terms of Service and DPA; they do not replace our obligation to provide the contracted software with reasonable care and skill. TrueValue Technologies Ltd is the supplier identified in those Terms.

An AI output is material generated or assisted by a model, such as a summary, classification, draft or research result. A calculation is a computation using specified data and assumptions. An Agent action is a software operation performed following permitted instructions. These are different functions with different risks; describing a calculation as deterministic does not establish that its inputs or commercial assumptions are correct.

Indicative valuations depend on the supplied financial information, dates, currency, normalisations, debt and cash treatment, working capital assumptions, ownership interest and selected method. Revenue, EBITDA and seller's discretionary earnings are not interchangeable. Enterprise value is not automatically the amount payable for equity or the cash received on completion.

You must check source periods, units, accounting policies, exceptional items, add-backs, comparable transactions and any estimated or inferred figures before material reliance. A calculated range is not a firm offer or a guarantee of market demand, financing availability or the final price a counterparty will accept.

Reports, sensitivities, risk flags, due-diligence checklists and scores are aids to review. They may omit matters, classify information incorrectly or depend on incomplete evidence. They do not constitute an audit, fairness opinion, formal tax valuation, legal due diligence certificate or proof of compliance unless a separate properly authorised professional engagement expressly provides that service.

AI can generate inaccurate statements, unsupported references, incomplete summaries, outdated information and plausible but incorrect interpretations. Documents can be misread or extracted incorrectly. Similar prompts may produce different results, and the same or similar output may be produced for another user.

Check important statements against the underlying evidence, open source references, verify figures and assumptions, and obtain appropriate qualified advice before a transaction, binding communication or other material decision. Do not present model confidence, a score or a polished report as independently verified fact. Where a source, extraction or action fails, do not treat a missing warning as confirmation that it succeeded.

The need for human review does not excuse a defect in our contracted software or a failure to apply agreed security or approval controls. Our liability is governed by the Terms of Service, not by a blanket statement that every AI-related loss is your responsibility.

The ordinary TrueValue service supplies software and information, not an engagement to advise on the legal, tax, investment or financing suitability of a particular transaction. We do not accept a transaction mandate merely because the product stores a buy box or proposes a task list.

You must not use output as an unauthorised financial promotion, represent TrueValue as an appointed professional adviser, or use a disclaimer to disguise an activity requiring authorisation. Any separately proposed regulated or professional service needs an appropriate legal assessment and express agreement. A product disclaimer alone does not determine regulatory status.

Do not use the ordinary Service for solely automated decisions about individuals with legal or similarly significant effects, including employment, personal credit or eligibility decisions, without an expressly agreed and lawfully assessed arrangement. General business research is not permission to infer sensitive personal characteristics or conduct prohibited profiling.

You may authorise the Agent to perform available internal research, analysis, monitoring and preparation within the workspace, subject to its permissions and configured limits. Define the objective, relevant data, scope and completion conditions. An instruction to prepare or research does not authorise an external disclosure or commercial commitment.

An authorised user may approve a clearly bounded set of actions. Approval must identify the affected records or recipients, final content or tightly defined action, scope, timing and any cost limit. Open-ended phrases such as “handle everything” or the mere existence of an integration token are not sufficient authority for an unbounded external campaign or transaction commitment.

We may stop or request clarification of an ambiguous, unsafe, unlawful or out-of-scope instruction. We do not guarantee that an autonomous task will always complete, detect every relevant event or operate at a particular frequency unless that frequency is expressly included in the agreed service description.

Before sending an external email or enquiry, disclosing a document, publishing information, changing an external record or carrying out another consequential external action, the relevant authorised user must approve the specific action or clearly identified batch. The approval must cover the destination, final content or payload, attachments, affected record and any disclosed charge. A material change requires renewed approval.

We retain appropriate evidence of the approved action, approving user, scope and execution result as specified in our retention arrangements. The customer is responsible for whether the approving person has authority on its behalf and whether the approved communication is lawful. TrueValue is responsible for implementing the agreed approval boundary and executing the approved action with reasonable care and skill.

The Agent may not sign contracts, submit binding acquisition offers, transfer transaction funds, give a personal guarantee or accept transaction liability on your behalf merely because you ask it in a prompt. Preparing a draft is different from executing it. A separate properly designed and authorised workflow would be needed for any activity outside this scope.

An external action may be irreversible. Delivered email, published content, recipient downloads and third-party responses cannot reliably be recalled. Any undo function applies only to the records or operation expressly supported; it is not a promise to undo real-world consequences.

Before connecting a service, check the requested permissions, organisation account and information that can be read or changed. Keep credentials confidential, use the least access reasonably needed and revoke connections when no longer required. A connection selected by a customer can expose data to a separate provider under that customer's relationship with it.

Where we select a provider to process Customer Personal Data, our DPA and Subprocessor Register apply. We do not avoid responsibility for an appointed subprocessor by describing its model or API as external. Customer-selected providers remain subject to the legal duties applicable to each party's role.

Subscription fees and included usage are defined by the Order. We do not incur undisclosed paid database purchases, third-party services or metered charges on your behalf without an approved price or budget. A technical safeguard or quota is not permission to charge an arbitrary overage. Independently purchased third-party services are governed by their own accepted commercial arrangements.

Only provide data you are entitled to process using the selected feature and provider arrangements. Redact unnecessary personal and confidential information, and check whether a seller NDA or client instruction limits AI or cross-border processing. Avoid including credentials or sensitive personal information not needed for the task.

We process Customer Data for the agreed service purposes and do not use confidential Customer Data to train general-purpose AI models. Any optional improvement use requires a separate specific agreement and appropriate lawful basis. Provider-side retention, abuse monitoring and access are described in the completed Subprocessor Register; we do not promise zero retention without evidence that it applies.

Outputs and action logs may themselves contain personal or confidential information. Their access, sharing, export and deletion must be controlled accordingly. Exporting an output to another provider does not automatically extend TrueValue's control to that provider's systems.

Source licences, confidentiality duties and third-party intellectual property rights continue to apply. Public availability does not create unrestricted scraping, redistribution or marketing rights. Do not remove required attribution or misrepresent a licensed extract as exclusively owned material.

Output rights are addressed in the Terms of Service. AI output may not be unique or eligible for intellectual property protection, and embedded third-party material remains subject to its rights. You may use permitted output in business and client work but must preserve material assumptions, limitations and source notices when relevant.

Report a significant error, unexpected action or suspected disclosure promptly through the support route or security@truevalue.business where security is involved. Include safe identifying information and preserve the relevant output or action record. Pause the relevant task where necessary to prevent further harm.

We may change models or technical implementation without materially reducing agreed service or data protection, subject to the DPA's supplier-change process and applicable notice requirements. New purposes, permissions or consequential action types require the appropriate assessment and user-facing information before use. These terms do not expand an Agent's actual available capabilities.