- Landlord News
- Homeflow’s AI website chat could win more lettings leads — but one privacy mistake will undo the gains
Homeflow’s AI website chat could win more lettings leads — but one privacy mistake will undo the gains
Homeflow’s new AI website chat may help agencies capture more landlord, tenant, buyer and seller enquiries out of hours, but the compliance risk sits in the setup. Before launch, agencies need clear privacy wording, valid cookie consent where required, supplier due diligence, visible bot disclosure and testing to stop inaccurate or discriminatory answers reaching the public.
A website chatbot that captures names, phone numbers, email addresses or tenancy details is not a harmless website add-on; it is regulated data processing from the moment it goes live.
That is why Homeflow’s new AI live chat tool, reported by Property Industry Eye, should be treated first as a compliance project and only then as a lead-generation tool. The commercial case is obvious: 24/7 chat can turn anonymous traffic into valuation requests, viewing enquiries and landlord instructions. But if the setup is wrong, the same tool can create avoidable UK GDPR, PECR, consumer protection and discrimination risk.
Privacy notice and lawful basis come first
If the chatbot collects enquiry details, stores transcripts or pushes leads into a CRM, agencies need a lawful basis for that processing and a privacy notice that actually reflects what the tool does. A notice drafted for a simple contact form will rarely be enough once an AI tool is asking follow-up questions, recording conversations and passing data to third-party systems.
Before launch, your notice should explain what the chat captures, why it is collected, who receives it, how long it is kept and whether any processor or sub-processor is involved. If you cannot describe that data flow clearly, you are not ready to switch the tool on.
Cookie consent is often the first failure point
Many agency sites still get this wrong. If the chat product uses non-essential cookies, analytics tags or similar tracking tools, PECR generally requires consent before those technologies are set. A banner that covers only your existing analytics stack is not enough if the chatbot introduces new tracking.
The practical test is simple: does the tool fire before the visitor has accepted non-essential cookies? If it does, that is a compliance problem, not a web design quirk. Agencies should update the cookie policy, list chat-related trackers where possible, and make sure the consent platform blocks them until the user opts in.
Tell visitors clearly they are speaking to AI
Bot disclosure should appear in or next to the chat window, not in a footer link. Users need to know they are interacting with an AI system, what it can help with and when a human will take over.
That matters particularly in lettings, where queries quickly stray into deposits, notice periods, repairs, affordability, guarantors and licensing. A bot that sounds authoritative while getting the law wrong creates a complaint first and a compliance issue shortly afterwards. The safer approach is to stop the tool from giving legal, financial or tenancy advice and route those queries to staff.
Check whether a DPIA is needed
Not every chatbot deployment will require a Data Protection Impact Assessment, but every agency should make and record that decision before launch. The reporting on Homeflow’s tool does not show that it carries out profiling or automated decision-making with legal or similarly significant effects. Agencies should not assume that it does.
But if your configuration scores leads, steers applicants, profiles users or processes sensitive information at scale, the case for a DPIA becomes much stronger. Larger multi-branch operators should be especially careful because high volumes and shared systems make it easier for risky data flows to go unnoticed.
Supplier due diligence matters more than the sales demo
If the provider is processing personal data on your behalf, a signed data processing agreement should be in place before the first lead is captured. Agencies should get four points confirmed in writing: where the data is stored, what security measures apply, how breach notification works and how deletion is handled.
Also ask whether transcripts are retained for service improvement or model training, whether sub-processors are used and whether any data is stored outside the UK. If a supplier cannot answer those questions clearly, that is a warning sign.
Inaccurate answers can become regulatory risk
The biggest operational weakness in AI chat is confidence without accuracy. Wrong answers about fees, availability, licensing, deposit handling, rent levels or landlord obligations can quickly become misleading statements.
That is not just a customer-service issue. It can engage consumer protection law, advertising rules and, in some cases, fair treatment obligations. In London especially, where licensing rules vary sharply between boroughs, a generic answer can be wrong in a way that costs a landlord or tenant real money.
Test for discrimination, not just functionality
Agencies should also test whether the chatbot responds differently based on names, family status, benefit status, language or other protected characteristics. If it does, the risk is not only reputational. It may point to Equality Act problems as well.
This testing should happen before launch and again after any significant update. Keep records of the scenarios tested, the answers produced and the escalations added as a result.
Do not treat enquiries as marketing consent
A viewing request or valuation enquiry is not automatic consent for property alerts, landlord newsletters or valuation campaigns. If chatbot leads are pushed into a CRM, the marketing basis and channel permissions need to transfer with them.
This is where firms often trip up: staff see a fresh lead in the database and assume it can be used for wider campaigns. It cannot, unless the compliance basis is there.
AI chat is not an AML shortcut
For agencies that move quickly from website enquiry to instruction or onboarding, the chatbot should never be treated as identity verification. If a lead progresses into a tenancy, sale or landlord instruction, the usual AML and identity checks still need to happen where relevant.
Three checks before rollout
- Get the supplier’s data flow, storage, security and sub-processor position in writing and decide whether a DPIA is required.
- Update the privacy notice, cookie controls and on-screen bot disclosure before the tool captures a single lead.
- Run scripted testing across high-risk topics such as deposits, licensing, affordability, repairs, complaints and discrimination, with clear escalation to human staff.
R&S can help your agency review AI chat wording, cookie setup and supplier paperwork before launch so the first leads it captures do not become the first complaints you have to defend.
This article is general information, not legal or financial advice. Rules can change and may apply differently to each property. Check the dated source and seek appropriate professional advice before acting.
Need help reviewing your property?
Arrange a conversation with our team about your property, tenancy and the local requirements that may apply.
Request a compliance conversationStay informed
Get our landlord updates by email
Receive selected landlord updates and links to source material.
