Lettings AI automation: how UK letting agents handle references, deposits and renewals

Lettings is the most admin-dense part of running a UK property business. A single tenancy generates more touchpoints than most sales: referencing, deposit registration, move-in coordination, periodic inspections, renewal negotiations and deposit returns. Each stage involves a defined sequence of chasers, reminders and confirmations. Most letting agents know exactly what needs to happen. The problem is executing it consistently across a portfolio of dozens or hundreds of tenancies without anything falling through the gaps. This is where lettings AI automation for UK letting agencies is most directly useful: not replacing the judgement calls, but handling the routine follow-up that currently depends on a person remembering to send the right message on the right day.

Why lettings admin falls through

The reference stage is a useful example. When a tenant is approved in principle, the referencing process typically involves an employer reference, a previous landlord reference and a credit check, each going to a different organisation on a different timeline. The letting agent's job is to chase each one until it comes back, flag any issues to the landlord and keep the applicant informed throughout. In a branch handling thirty active lets, this is several hours of chasing per week, most of it repetitive.

Deposits bring their own sequence. Under the tenancy deposit protection rules, a landlord has thirty days from receipt of the deposit to protect it with an approved scheme and provide the tenant with the prescribed information. Missing that window is a civil offence and exposes the landlord to a penalty of up to three times the deposit amount. The agent's responsibility is to make sure registration happens. In practice, the reminder to register often sits in someone's task list rather than being tracked against the actual statutory deadline.

Renewals are the third pressure point. A standard AST approaching its fixed term expiry needs action roughly two to three months out: a conversation with the landlord about terms, a renewal offer to the tenant and a decision on whether to serve a Section 21 or proceed with a new agreement. Each of those steps has a natural deadline. Agencies that track them manually, using CRM reminders or calendar entries, find that the sequence works when nothing else is competing for attention. During busy periods, it does not.

The common thread across all three is that these workflows are entirely predictable. The dates are in the CRM. The actions are defined. The only variable is whether anyone is paying attention when the trigger arrives.

What lettings AI automation actually does for UK agencies

The architecture for any effective lettings AI agent follows a single shape: an event-triggered agent with one clear loop. Trigger, read context, reason, take action, log.

References. The trigger is a tenancy moving to the referencing stage in the CRM. The agent reads the applicant's record, identifies which reference types are outstanding and sends the first chase to the relevant employer, previous landlord or referencing agency. It logs the outreach and sets a watch on the response. If no reply arrives within the defined window, a follow-up goes out. If the reference comes back with a flag, the agent surfaces the issue to the property manager via Slack, email or WhatsApp with the relevant detail already summarised. The property manager sees a notification with context, not a buried CRM note they have to go looking for.

Deposits. The trigger is deposit receipt being logged in the CRM. The agent calculates the thirty-day protection deadline, tracks it against the clock and sends a prompt to the responsible team member with enough notice to act. Where the agency registers deposits directly, the agent can draft the confirmation letter for review. If the property manager uses the inbox copilot to handle routine correspondence, the deposit confirmation draft lands there alongside their other queued messages, ready to review and send.

Renewals. The trigger is the fixed term expiry date in the tenancy record. The agent looks ahead by a defined period, typically ninety days, and initiates the renewal sequence. It drafts a renewal conversation prompt for the property manager, checks the landlord's stated preferences from the CRM and prepares a draft renewal letter or a draft notice, depending on what the landlord wants. The decision on terms and whether to renew remains with the property manager and the landlord. The agent handles the calendar-driven prompts that typically fall through the cracks.

All three loops run without requiring a person to initiate them. The property manager's job is to review the output, edit if needed and approve. The action surface is wherever they already work: Slack, Outlook, Gmail or a lightweight approval screen built for the job. Nothing sends without a human confirming it, unless the agency has specifically requested auto-send on a defined category of routine message.

This is the same underlying pattern that Sortd applies to sales progression: the event-driven loop where the AI does the chasing and the team retains the final word. Each loop has a clear trigger, a defined action and a logged result. Nothing depends on a person remembering.

What it looks like for a letting agency

Take a two-branch independent managing one hundred and fifty properties. Four property managers handle references, renewals, deposits, inspections and day-to-day tenant contact.

In a typical month, twelve tenancies are due for renewal discussion, three deposits are awaiting registration confirmation and eight sets of references are in progress across different stages. Tracking all of this relies on a combination of CRM reminders and a shared spreadsheet that is accurate until it is not.

With a lettings AI automation build in place, the picture changes.

A new tenancy moves to the referencing stage in Alto on a Wednesday afternoon. By Thursday morning, the employer reference request has gone out, the previous landlord has received a reference request and the referencing agency has been notified. All three are logged in Alto under the applicant's record. If no response arrives by Monday, a follow-up goes out automatically. The property manager receives a Slack notification that referencing is in progress, not a task to initiate.

A deposit lands in the agency account on a Friday. The agent logs the receipt, calculates the protection deadline and sends a prompt to the property manager on day twenty-five. The prompt includes the TDS portal link and the landlord's contact details. The registration happens that afternoon.

Three renewal conversations are due in the next ten days. Each property manager starts their Monday with a drafted prompt for each case, pulled from the current tenancy record: remaining term, the landlord's last stated preference on rent and a summary of any outstanding maintenance. The conversation is ready to have. The preparation is already done.

Compliance and integration

Lettings automation touches a significant volume of personal data. Tenant identity, financial information, employment details and residential history all come through the referencing and tenancy management process. UK GDPR applies from the moment those records exist in your system.

The lawful basis for processing is typically the performance of a contract: the tenancy agreement creates the relationship, and the administrative steps to manage it are necessary to fulfil it. A data processing agreement with the AI provider is required. Data must be held in UK or EU infrastructure. No tenant or landlord data should be routed through US-based servers. Sortd uses UK and EU infrastructure by default.

On AML, letting agents handling client money and tenancies above reporting thresholds have obligations under the Money Laundering Regulations 2017. The AI does not make AML determinations. It surfaces prompts and flags missing steps. The decision and the documentation remain with the regulated individual.

Deposit protection registration itself requires a person to confirm and submit to the TDS, MyDeposits or DPS. An AI agent can draft, prepare and flag, but the final submission is a human action.

On CRM integration, Sortd builds lettings automation on top of Reapit, Alto, Jupix, Vebra, Dezrez, agentOS and similar systems used by UK independent agencies. Property Industry Eye carries regular coverage of the regulatory requirements affecting UK letting agents and is a useful reference point as the compliance landscape evolves.

How to get started

The right starting point for any lettings agency evaluating AI is to map the friction first. Where are the most consistent delays in your referencing process? Which renewals tend to get handled late? How often does a deposit registration deadline arrive closer than you would like?

Those are the workflows worth automating first. A well-built lettings AI agent is not a general-purpose tool. It is a set of specific loops, each wired to a defined trigger in your CRM, each producing a draft that your team approves before it acts. The question is which loops to build first, based on where the friction is highest in your current operation.

Sortd's starting point is a discovery call and a free working version of the agent. We build it against a test slice of your CRM, with your team's approval step already in place, and run it before anything touches a live tenancy. You see how it handles your referencing and renewal sequences before committing to a full build.

If lettings admin is the bottleneck in your property management business, start with a conversation.