AI Disclaimer and AI Use Policy

This AI Disclaimer and AI Use Policy (the Policy) is issued by Blackwellen Limited, a company registered in England and Wales under company number 16482166, whose registered office is at 61 Bridge Street, Kington, Herefordshire, HR5 3DJ, United Kingdom (Blackwellen, we, us or our). Propvora is a product of Blackwellen.

This Policy explains the artificial intelligence features available in Propvora, the limits of those features, how data is handled when you use them, and the rules you must follow. It forms part of, and should be read together with, our Terms of Service, our Acceptable Use Policy, our Privacy Notice and, where we process personal data on your behalf, our Data Processing Agreement. If this Policy conflicts with the Terms of Service, the Terms of Service prevail, except where this Policy imposes a more specific restriction on the use of AI Features, in which case that restriction applies.

This Policy applies to every person who uses the AI Features, whether as a workspace owner, a team member, a supplier, or a user of a portal in which AI Features are made available. Workspace owners are responsible for making sure their team members are aware of it.

In this Policy, the following words have these meanings. Other capitalised terms have the meaning given in the Terms of Service.

• AI Features means any functionality in the Service that uses machine learning or generative AI to produce content, classifications, predictions or suggestions, including the Propvora Agent.
• AI Provider means a third-party supplier of AI models or inference services that we use to deliver the AI Features, as listed on our subprocessors page.
• Agent means the conversational assistant (the Propvora Agent) in the Service, currently offered as a Beta Feature.
• Beta Feature means a feature we label as beta, preview, early access or similar.
• Customer means the business or individual who holds the subscription for a workspace.
• Input means any prompt, instruction, question, file, record or other content that you or the Service submit to an AI Feature, including workspace data the AI Feature retrieves to answer your request.
• Output means any content, summary, draft, classification, score, suggestion or other result produced by an AI Feature.
• External Effect means any action that leaves the Service or changes the position of a third party, including sending an email, SMS or portal message, issuing or sharing a document, making, requesting or reconciling a payment, instructing a contractor, and creating, changing or deleting records in bulk.
• Service means the Propvora software platform, websites, portals and related services provided by us.
• Workspace Data means data uploaded to or created in a Customer's workspace.

The AI Features are designed to save time on routine property operations work. They assist people; they do not replace people's judgement. At the date of this Policy, the AI Features include:

• Propvora Agent (Beta): a conversational assistant that can answer questions about records in your workspace, explain what is on screen, and propose next steps or actions for you to review.
• Summarisation: short summaries of long records, message threads, documents, work order histories, inspection notes and similar content.
• Drafting: first drafts of messages, letters, notices, job descriptions, reports and other text, for you to edit and approve.
• Classification and extraction: suggesting categories, priorities, tags or key fields, for example classifying an incoming repair request or reading dates from an uploaded certificate.
• Suggestions inside automations: within the automations module, an AI step may draft content, propose a classification or recommend a next action. Where the automation would then take an action with an External Effect, the approval rules in section 7 (Human oversight and the approval model) apply.

Not every AI Feature is available on every plan, and we may add, change or withdraw AI Features over time (see section 14 (Beta status and availability)). Where an AI Feature is in use, we aim to make that clear in the interface, for example by labelling Outputs as AI-generated or showing an AI icon.

The AI Features work from the information they are given. They may draw on Workspace Data that you have permission to view, but they do not have access to information outside the Service unless we say so in the interface, and they do not independently check public registers, legislation or the physical condition of a property.

We do not build our own foundation models. To deliver the AI Features we send Inputs to, and receive Outputs from, AI Providers. At the date of this Policy these are:

• OpenAI, providing large language models through its API.
• Microsoft Azure OpenAI Service, providing OpenAI models hosted on Microsoft's cloud.
• NVIDIA, providing an AI inference API.

We may route a request to a different AI Provider, or a different model from the same provider, depending on the task, cost, availability and quality. The current list of AI Providers, what each one does, where it processes data and the transfer safeguards that apply is published on our subprocessors page. We will update that page, and give notice to Customers as set out in our Data Processing Agreement, before adding or replacing an AI Provider that processes personal data on a Customer's behalf.

Some AI Providers may process data outside the United Kingdom, including in the United States. Where they do, we rely on an appropriate safeguard under UK data protection law, such as the UK International Data Transfer Agreement, the UK Addendum to the EU Standard Contractual Clauses, or the UK–US Data Bridge where the recipient is certified. Further detail is in our Privacy Notice and on the subprocessors page.

Generative AI produces Outputs by predicting likely text or labels from patterns in its training data and in the Inputs it receives. It does not understand your situation in the way a person does, and it does not know whether what it produces is true. You should expect that Outputs may be:

• inaccurate: an Output may state facts, figures, dates or names that are wrong, including with apparent confidence, and may invent references, clauses or sources that do not exist;
• incomplete: an Output may leave out information that matters, such as an exception, a deadline or a record the AI Feature could not see;
• out of date: the models' general knowledge has a cut-off date and may not reflect recent changes in law, guidance, prices or practice, including changes made by the Renters' Rights Act 2025 and the regulations made under it;
• inconsistent: the same Input may produce different Outputs at different times;
• inappropriate in tone or content: despite safeguards, an Output may be unsuitable to send to a tenant, applicant or contractor without editing; and
• affected by the quality of the Input: an Output is only as good as the records and instructions it is based on.

We take reasonable steps to improve the reliability of the AI Features, including grounding the Agent's answers in your Workspace Data where possible, limiting AI Features to defined tasks, and applying content safeguards. These steps reduce, but do not remove, the risk of error.

The AI Features and their Outputs are not, and must not be treated as, professional advice of any kind. Blackwellen is not a law firm, a regulated financial adviser, a tax adviser, a chartered surveyor, a Gas Safe registered engineer, a registered electrician or a health and safety consultant, and nothing in an Output creates any professional relationship between you and us.

In particular, you must not rely on an Output as the basis for any of the following, each of which requires your own judgement and, where appropriate, advice from a suitably qualified professional:

• Legal and housing-law determinations, including whether a notice or possession ground is valid or correctly served, whether a tenancy has been created or ended correctly, what rent increase is permitted, whether a deposit has been protected and the prescribed information given in time, whether a property needs an HMO, additional or selective licence, and what duties apply to you under the Renters' Rights Act 2025, the Housing Act 2004 or any other housing legislation.
• Financial and tax matters, including the treatment of income and expenses, VAT, Making Tax Digital, service charge apportionment, arrears figures, and whether a payment is due or has been received.
• Surveying and valuation, including the condition of a building, the cause of a defect, rental valuations and dilapidations.
• Gas and electrical safety, including whether a Gas Safety record or Electrical Installation Condition Report is satisfactory, whether a defect is dangerous, and whether remedial work is complete. Only an appropriately registered and competent person can make those determinations.
• Health and safety, including fire risk assessments, hazards under the Housing Health and Safety Rating System, damp and mould, legionella, asbestos and Awaab's Law timescales where they apply.
• Energy performance, including whether a property meets the minimum energy efficiency standards or qualifies for an exemption.
• Decisions about individuals, including whether to let to an applicant, accept a guarantor, pursue arrears or end a tenancy (see section 8 (No solely automated decisions and no discriminatory screening)).

The Service includes compliance tracking that shows dates, statuses and reminders based on the information you enter. Those features, and any Output about compliance, are tools to help you organise your work. Your legal and regulatory obligations as a landlord, agent, manager or operator remain yours alone, and using the Service or the AI Features does not transfer any of them to us.

Where an Output summarises or quotes legislation, guidance or a document, you must check the original source before relying on it. Where we provide links to official guidance, those links are for convenience and the official source prevails.

The AI Features are designed around a human-in-the-loop model. AI may read, summarise, draft, classify and suggest, but a person decides what happens next.

7.1 Actions with an External Effect

By default, no AI Feature will take an action with an External Effect unless an authorised user has reviewed it and expressly confirmed it in the Service. Section 7.4 explains how a Customer can choose otherwise for some actions. In practice, unless the Customer's Agent rules say otherwise:

• AI-drafted messages, letters and notices are presented as drafts. They are not sent to tenants, applicants, landlords, contractors or anyone else until a user chooses to send them.
• AI Features do not initiate, authorise or alter payments, refunds, payouts or bank details. Any payment action requires a user to complete it through the normal payment flow.
• Where the Agent or an automation proposes changes to multiple records at once, such as updating statuses, reassigning work or archiving records, the proposed changes are shown for review and are applied only after a user confirms them.
• Where the Agent proposes a single action, such as creating a task or scheduling a reminder inside your workspace, it presents the action for confirmation rather than carrying it out silently.

7.2 Automations

Customers can build automations that run on a schedule or in response to an event. An automation that you have configured and switched on may carry out the steps you have set up without asking again each time; that is the purpose of an automation, and you are responsible for configuring it carefully. Where an automation includes an AI step whose Output would lead to an action with an External Effect, the Service requires that Output to pass through an approval step before the action is taken. You must not try to remove, bypass or automate that approval step.

7.3 Your role as reviewer

Confirming an AI-proposed action means you have read it, you are satisfied it is accurate and appropriate, and you take responsibility for it as if you had written it yourself. You should only confirm actions you are authorised to take within your organisation. Workspace owners should use roles and permissions to control which team members can approve AI-proposed actions.

7.4 Agent rules and automatic actions

A workspace owner or admin can set Agent rules (Settings > Rules > Agent rules) that decide, for each kind of action, whether the Agent may not do it, may only draft it, may do it after a person approves it, or may do it automatically within limits the Customer sets, such as quiet hours, the number of messages per person and the value of work it may award to a supplier. Before any action can run automatically, the owner or admin must confirm in the Service that automatic actions under those rules are taken on the Customer's instructions and are the Customer's responsibility. We record who agreed, when, and what was agreed; turning on further automatic actions or raising a limit requires agreement again.

An action the Agent takes automatically under rules the Customer has agreed to is treated, for the purposes of this Policy and our Terms, as an action the Customer has approved. Whatever the rules say, the Agent never moves money, signs on anyone's behalf or erases data; decisions about applicants and tenancies, and notices to tenants, always need a person to confirm each one (see section 8); and people marked as vulnerable, not to be contacted, or in a dispute, complaint or legal matter are never contacted automatically. Every automatic action is recorded in the Agent's activity and the audit log, and any rule can be moved back to approval at any time.

8.1 Significant decisions about people

UK data protection law, including Article 22 of the UK GDPR as amended by the Data (Use and Access) Act 2025, restricts decisions based solely on automated processing that produce legal or similarly significant effects on individuals, and requires safeguards where such decisions are made, including the right to make representations, to obtain human intervention and to contest the decision. Stricter rules apply where special category data is involved.

The AI Features are not designed or intended to make decisions of that kind, and you must not use them to do so. In particular, you must not use any AI Feature, alone or as the deciding factor, to decide whether to:

• accept or reject a rental application or a guarantor;
• offer, renew, vary or end a tenancy;
• set or increase a rent or charge for a particular individual;
• take arrears, enforcement or possession action against an individual; or
• refuse a repair, adjustment or service to an individual.

If you use an Output as one input to a decision of this kind, a person with the authority and competence to change the outcome must make the decision, consider all relevant information, and be able to explain it. As controller of the personal data concerned, you are responsible for meeting your transparency obligations to the individual and for handling any request for human review.

8.2 Equality and non-discrimination

The Equality Act 2010 prohibits discrimination in the disposal and management of premises because of a protected characteristic: age (in some contexts), disability, gender reassignment, marriage and civil partnership, pregnancy and maternity, race, religion or belief, sex and sexual orientation. Indirect discrimination, harassment, victimisation and a failure to make reasonable adjustments are also unlawful. The Renters' Rights Act 2025 also makes it unlawful in England to discriminate against prospective tenants because they receive benefits or have children, subject to its exceptions.

You must not use any AI Feature to screen, score, rank, filter, profile or exclude tenants or applicants, directly or indirectly, by reference to a protected characteristic, benefit status, the presence of children, or a proxy for any of them (such as name, nationality, accent, postcode used as a proxy for ethnicity, or gaps in employment linked to disability or caring responsibilities). You must not ask an AI Feature to infer a person's protected characteristics. Breach of this section is a serious breach of our Acceptable Use Policy.

9.1 Roles

When an AI Feature processes Workspace Data that includes personal data, the Customer is the controller and Blackwellen is the processor, and our Data Processing Agreement applies. The AI Providers act as our sub-processors. Where we process data about how the AI Features are used for our own purposes, such as security monitoring, abuse prevention, billing and service improvement as described in section 15 (Feedback and accuracy improvement), we do so as controller under our Privacy Notice.

9.2 What is sent to AI Providers

When you use an AI Feature, we send the AI Provider the Input needed to complete the task. That may include your prompt and relevant extracts of Workspace Data that you have permission to view, such as the record you are looking at or the message thread you asked to summarise. We aim to send no more data than the task reasonably needs. AI Features respect workspace boundaries and the permissions of the user making the request: they should not retrieve data from another workspace, or data the requesting user could not otherwise see.

9.3 Training of third-party models

We do not permit AI Providers to use your Inputs or Outputs to train or improve their general-purpose foundation models where the provider's terms or settings allow us to prevent it. At the date of this Policy, the API services we use from OpenAI and Microsoft Azure OpenAI Service do not use API data for model training by default, and we have not opted in to any such use. We will not change this position for Workspace Data without first updating this Policy and giving Customers notice.

9.4 Retention

Inputs and Outputs that are saved in your workspace, such as an approved message, a stored summary or an Agent conversation history, form part of your Workspace Data. They are retained and deleted in the same way as other Workspace Data under our Data Processing Agreement, including at the end of your subscription. AI Providers may retain Inputs and Outputs for a limited period for abuse and misuse monitoring, as permitted by their terms, after which they are deleted; where a provider offers reduced or zero retention for eligible customers, we will consider using it where it is available to us on reasonable terms. Details for each provider are on the subprocessors page.

9.5 Your choices and opt-outs

You are never required to use the AI Features to use the core Service. You can choose not to use the Agent or other AI Features, and you can edit or delete saved Outputs. A workspace owner who wants AI Features switched off for the whole workspace may do so using the workspace settings where that control is available, or by asking us at info@blackwellen.com. Individuals whose personal data is held in a Customer's workspace, such as tenants, should contact the relevant landlord or agent first, as that Customer is the controller; we will help the Customer respond as required by our Data Processing Agreement.

9.6 Sensitive information

Don't enter special category data (such as health information, ethnicity or religious beliefs), criminal offence data, full payment card numbers, bank passwords or authentication codes into an AI Feature unless it is genuinely necessary for a legitimate purpose, you have a lawful basis and any required condition for doing so, and the feature is designed for it. Where you are unsure, leave it out.

When you or your team use the AI Features, you are responsible for:

1. reviewing every Output before you rely on it, approve it or share it, and correcting any errors;
2. making sure you have a lawful basis, and have given any necessary privacy information, for the personal data you put into the AI Features, including telling tenants and other individuals, in your own privacy notices, that you use AI-assisted tools;
3. making sure you have the rights and permissions needed for the content you input, including any third-party documents;
4. configuring roles, permissions and automations so that only appropriate people can approve AI-proposed actions;
5. obtaining professional advice where section 6 (No professional advice and no reliance) applies;
6. complying with section 8 (No solely automated decisions and no discriminatory screening), the Equality Act 2010 and data protection law in any decision you make about an individual;
7. carrying out any data protection impact assessment that your own use of the AI Features requires; and
8. making sure your team members understand and follow this Policy.

Where you are a Customer acting in the course of business, you are responsible for the acts and omissions of your team members and anyone else you allow to use the AI Features in your workspace.

In addition to the general rules in our Acceptable Use Policy, you must not use, or attempt to use, the AI Features to:

• make solely automated decisions with legal or similarly significant effects on individuals, or screen people in a discriminatory way, contrary to section 8;
• produce content that is unlawful, harassing, threatening, defamatory, or intended to intimidate a tenant into leaving a property, or that could amount to unlawful eviction or harassment of an occupier;
• generate misleading communications, including fake notices, fake official letters, false statements about a tenant's legal rights or obligations, or content impersonating a court, local authority, deposit scheme, ombudsman or other official body;
• generate fake reviews, testimonials or references;
• create or spread malware, phishing content or other material designed to compromise security;
• infer or predict special category data or protected characteristics about individuals;
• carry out covert monitoring or profiling of tenants or other individuals beyond what is lawful and necessary for managing the property;
• extract, reverse engineer or attempt to discover the underlying models, system instructions or safeguards, including by prompt injection or jailbreaking;
• circumvent usage limits, rate limits, the approval model in section 7 or any other technical control;
• access data in another workspace, or data you are not authorised to view;
• build or train a competing AI model or product using Outputs; or
• breach any AI Provider's usage policies that we tell you about or that apply to the content you submit.

We may monitor use of the AI Features for abuse, and may suspend or restrict access to them, or to the Service, where we reasonably believe this section has been breached, in accordance with the Terms of Service.

People generally expect to know when they are dealing with AI. Where an AI-drafted communication is sent to a tenant, applicant, guarantor, landlord, leaseholder or contractor, you are responsible for deciding whether and how to tell them. As a minimum:

• you must not tell anyone, or imply, that a message was written personally by a named person where it was generated by AI and not reviewed by that person;
• where a communication could materially affect someone's rights or obligations, such as a notice, a demand, a decision on an application or a response to a complaint, a person must review it, and you should consider stating that it was prepared with the help of AI;
• where a portal or chat interface allows a third party to interact directly with an AI Feature, it must be made clear to them that they are interacting with AI and how to reach a person; and
• you must not use AI Features to create content that pretends to come from a person or organisation that did not produce it.

We may label AI-generated content in the Service, and in some cases in outgoing communications, so that recipients can understand how it was produced. You must not remove or obscure labels that we require for legal or safety reasons.

As between you and us, you retain all rights in your Inputs. Subject to your compliance with the Terms of Service and this Policy, and to the rights of any third party, we assign to you any rights we may have in Outputs generated for your workspace, and you may use them for your business purposes.

Because of how generative AI works, Outputs may be similar or identical to content generated for other users, and we cannot guarantee that an Output is original, that it is protected by copyright, or that it does not infringe a third party's rights. Under UK law the position on copyright in computer-generated works is evolving. You are responsible for checking Outputs before publishing them or using them commercially, particularly marketing content and anything that reproduces third-party material.

Blackwellen and its licensors retain all rights in the Service, including the AI Features, their prompts, configurations and safeguards. Nothing in this Policy transfers those rights to you.

The Agent, and any other AI Feature labelled beta or preview, is a Beta Feature. Beta Features are made available so that you can try them and give us feedback. They may be less reliable than the rest of the Service, may change significantly or be withdrawn, may have usage limits, and may be available only on some plans.

The AI Features depend on AI Providers whose services may be interrupted, rate-limited, changed or withdrawn outside our control. We do not guarantee that any AI Feature will be available at any particular time, will respond within any particular time, or will continue to be offered. Where we withdraw an AI Feature that forms a material part of a paid plan, we will give reasonable notice where we can and will deal with your subscription in accordance with the Terms of Service and our Refund and Cancellation Policy.

We may set fair-use limits on the AI Features, such as limits on the number of requests or the size of Inputs, to manage cost and protect the Service for all users. Any limits that apply to your plan will be shown in the Service or on our pricing page.

You can help us improve the AI Features by rating Outputs, flagging errors or sending us comments. If you report an inaccurate, harmful or inappropriate Output, we will review it and, where appropriate, adjust our prompts, configurations, safeguards or choice of model.

When you give feedback, we may review the Input and Output concerned to understand what went wrong. We use feedback, and aggregated or de-identified information about how the AI Features perform, to maintain, secure and improve the Service. We do not use Workspace Data to train our own or a third party's general-purpose foundation models. If we ever wish to use Workspace Data for a materially different purpose, such as fine-tuning a model, we will only do so with the Customer's prior agreement and in line with our Data Processing Agreement.

To report a serious problem with an Output, for example one that is discriminatory, unsafe or that exposed data it should not have, email info@blackwellen.com with the subject line "AI issue". You can also make a complaint under our Complaints Procedure, which covers AI Outputs.

16.1 United Kingdom

The UK regulates AI through existing laws and sector regulators rather than a single AI statute, guided by five cross-sector principles: safety, security and robustness; appropriate transparency and explainability; fairness; accountability and governance; and contestability and redress. We have designed the AI Features with those principles in mind, through the approval model in section 7, the restrictions in section 8, the transparency rules in section 12 and the redress routes in our Complaints Procedure. We also take account of the Information Commissioner's Office guidance on AI and data protection, including on fairness, transparency, data minimisation and individual rights.

16.2 European Union

Where the Service is used by organisations in the European Union, or to process data about people in the EU, Regulation (EU) 2024/1689 (the EU AI Act) may apply, with its obligations taking effect in stages. We consider the AI Features to be general-purpose assistance tools rather than high-risk AI systems as provided for in the EU AI Act. However, the EU AI Act treats some uses in employment and in access to essential services as high-risk, and some uses as prohibited. If you use the AI Features in the EU for a purpose that could be high-risk or prohibited, including the evaluation of individuals for access to housing-related services, you do so outside their intended purpose and in breach of this Policy. You are responsible for any obligations that arise for you as a deployer, including AI literacy among your staff and transparency towards people who interact with AI.

We will keep the AI Features and this Policy under review as UK and EU law and guidance develop.

The AI Features are provided as part of the Service, and our liability in connection with them is governed by the liability provisions of our Terms of Service, including the caps and exclusions that apply to business customers. In particular, and to the extent permitted by law, we are not responsible for loss arising from your reliance on an Output without the review required by this Policy, from any action you approved or that ran automatically under Agent rules you set and agreed to, or from your use of the AI Features in breach of this Policy.

Nothing in this Policy limits or excludes any liability that cannot lawfully be limited or excluded, including liability for death or personal injury caused by negligence, or for fraud or fraudulent misrepresentation.

If you use the Service as a consumer, you have legal rights in relation to digital content and services that are of satisfactory quality, fit for purpose, as described and supplied with reasonable care and skill. Nothing in this Policy affects those statutory rights.

We may update this Policy to reflect changes to the AI Features, our AI Providers, the law or regulatory guidance. We will publish the updated version on this page with a new effective date and, where a change materially affects how your data is used or what you may do with the AI Features, we will give Customers reasonable advance notice by email or in the Service.

Questions about this Policy, and requests relating to the AI Features, can be sent to our data protection lead at info@blackwellen.com or by post to Blackwellen Limited, 61 Bridge Street, Kington, Herefordshire, HR5 3DJ, United Kingdom. You also have the right to complain to the Information Commissioner's Office at ico.org.uk, although we would appreciate the chance to resolve your concern first.