Most AI pitches to agents promise the wrong thing. The useful version is narrow: sort enquiries, answer routine questions from published facts, acknowledge people out of hours, and draft the boring writing. Valuing, negotiating and being accountable stay with you. This guide sets the line — and shows how MoveLane enforces it in our own platform.
No sign-up needed to read · Written for owners and directors of independent UK agencies
Every AI decision in an agency falls into one of three buckets. Routine, reversible and low risk — safe to run on its own. Commercial, contractual or emotive — drafted by AI, sent only after a person approves it. And a third bucket that should never be automated at all, whatever a vendor tells you: money movement, legal positions, and anything that publishes property detail to the public.
That is not a philosophy, it is the policy MoveLane runs on. Our AI team works to three levels: safe-auto, human approval, and prohibited in code. Sorting the work this way is what makes automation boring enough to trust.
Each one shows how far the AI goes, and where a person picks it up.
Portal and website enquiries arrive in one inbox and get sorted by whoever is free.
An enquiry lands at 21:40 and nobody replies until the morning.
The same twenty questions about fees, process and paperwork, over and over.
Descriptions written in a hurry, missing key facts, inconsistent across the stock list.
Valuation leads go cold because nobody remembered to chase them.
No single view of enquiries, viewings booked and what needs attention today.
If a supplier offers to automate any of these, that is your signal to walk away.
Valuations and asking-price advice
A number attached to someone's home is a professional judgement with real consequences. AI can prepare context, not the figure.
Fees, quotes and discounts
Commercial terms are a negotiation. An AI that improvises on price costs you money and creates disputes.
Complaints and anything emotive
Complaints need accountability and a named person. Automated sympathy makes them worse.
Legal, GDPR, AML and compliance replies
Regulated answers must come from someone who can be held responsible for them.
Offers, negotiations and progression decisions
These involve competing interests and disclosure duties that sit with the agent, not a model.
Publishing anything to a live site or portal
Misdescription risk. A person approves before a listing goes public.
| Risk | Guardrail |
|---|---|
| Invented facts (“hallucination”) in a customer reply | Answer only from a verified knowledge source, and make “I don't know, passing this to a colleague” an acceptable output. |
| Consumer Protection Regulations and misdescription | Never let AI publish property detail. Draft, then a human checks against the facts. |
| Data protection when enquiries contain personal data | Keep enquiry data inside your own systems, limit who and what can read it, and log every automated action so you can show what happened. |
| Tone that does not sound like your agency | Fix the wording once at the template level rather than editing every message. |
| Loops and over-messaging | Rate-limit automated messages per recipient and stop on any human reply. |
| No audit trail when something goes wrong | Log every automated decision with its reason, and keep a switch that stops it instantly. |
Nothing here is legal advice. Consumer protection and data protection duties stay with your agency however the message was written.
This is the pipeline behind every enquiry sent through MoveLane — described exactly as it runs, including the parts that deliberately stop and wait for a person.
1. The enquiry is read and classified
A viewing request, a valuation request, a support question or something sensitive. The classification decides where it goes, and it is stored with the enquiry.
2. It is routed with a short summary
The right team gets the enquiry plus a one-line summary and a suggested next action, so nobody has to re-read a thread to work out what is being asked.
3. Routine acknowledgements can go out immediately
A factual confirmation that the enquiry arrived and what happens next. It never quotes a price, promises a viewing slot or negotiates.
4. Anything commercial or sensitive is held
Fees, discounts, contracts, complaints, legal, GDPR and cancellations are drafted and queued for a named human in our approval centre. Nothing sends until someone approves it.
5. Every action is logged with its reason
Each automated decision records which rule allowed it, so you can audit what happened. There is also a master switch that stops all automation instantly.
6. Rate limits and loop protection
Automated messages are capped per recipient and stop on a human reply, so no one is ever chased by a machine.
Being straight about it: this handles the admin around enquiries. It does not value property, negotiate, or replace anyone on your team — and we would not sell it as though it did.
Three screens from the MoveLane admin: enquiry triage, the approval gate that holds commercial replies, and the rules that decide what may auto-send.
Enquiry triage
Enquiry triage and routing
A sample enquiry is read, summarised, scored and routed to the sales team. The annotations highlight the specific controls and AI decisions visible in the screenshot.
Approval gate
Commercial replies wait for a person
A pricing or commercial enquiry is drafted, then deliberately held as “draft awaiting approval”. The annotations explain the controls that prevent automatic sending.
Auto-reply rules
What may auto-send, and the kill switch
The live rules behind auto-replies. The annotations map the screenshot to the three autonomy levels and the emergency stop.
Real MoveLane admin screens, captured in test mode with made-up sample enquiries. No customer data is shown.
Paste a real enquiry (or pick an example) and see the decision the same policy would make: handled automatically, drafted for a human, or blocked entirely.
This runs the same policy the live system uses. It happens in your browser — nothing is sent, saved or replied to.
From £99/month after the trial · Rolling monthly · No commission
What agency owners ask us most about automating enquiry work.
No. The parts of agency work that customers pay for — valuing accurately, negotiating, and being accountable when something goes wrong — need a person. What AI reliably removes is the admin around them: sorting enquiries, acknowledging them quickly, drafting copy and chasing tasks.
Routine, reversible, low-risk work: classifying and routing an enquiry, confirming it was received, answering a question from published facts, and creating internal tasks and summaries. On MoveLane these are the only actions that run automatically.
Valuations and asking-price advice, fees and discounts, complaints, legal, GDPR and AML replies, offer negotiations, and publishing anything to a live listing. On MoveLane those are either held for a named human to approve or blocked in code entirely.
Every enquiry through MoveLane is read and classified, routed to the right team with a summary, and logged. Routine acknowledgements and published-fact answers can send automatically; anything commercial, legal or sensitive is drafted and held until a person approves it. Every automated action is recorded with the reason it was allowed.
Not for this. The enquiry handling described here runs as part of a MoveLane subscription — there is nothing to install or configure. If you want to build your own tooling on top, that is a separate project with separate risk.
Founding Agent Membership is £99 per month for the first 100 agencies (normal price £199) with a 14-day free trial, rolling monthly, and no sales commission. The AI enquiry handling is part of the platform rather than a paid add-on.
Tell us your area and branch count and we’ll walk you through what would be automated, what would stay with your team, and what it would look like in your inbox.
Published pricing £99/month, rolling monthly, 14 days free. Enquiry triage and acknowledgements are part of the platform, not an add-on.
