Reading and classifying incoming email
Work out what an incoming message is about, which property and people it relates to and how urgent it is, then route it into the right workflow instead of the general inbox.
AI for property businesses
AI that handles the reading, sorting and drafting. Your team keeps the judgement.
AI has become genuinely useful for the parts of property work that involve reading, classifying, extracting and summarising information. Used carefully, inside well-designed workflows, it removes a great deal of mechanical effort. It is one tool among several, and we use it only where it earns its place.
01 The reading problem
Most of the work in a property business is reading and re-typing.
Emails from tenants, landlords, applicants, contractors and solicitors. Referencing reports, inventories, certificates, invoices and applications. Every one of them has to be read, understood, categorised and acted on, and the information in it moved somewhere else.
That reading and re-typing is exactly what modern AI models are good at. They can tell a leak from a lockout, pull the important details out of a document, summarise a long thread and draft a sensible first reply. What they should not do is make consequential decisions on their own.
So we use AI where it helps, alongside conventional automation and software, and we keep a person in the loop wherever judgement, money, tenancy rights or legal obligations are involved.
Sound familiar?
- 01
Someone reads every incoming maintenance email to work out what it is about and how urgent it is.
- 02
Details from referencing reports and applications are copied into the CRM by hand.
- 03
Long email threads have to be re-read from the top before anyone can respond.
- 04
Enquiries sit unanswered until the right person has time to write a reply.
- 05
Invoices are matched to jobs and properties one at a time.
- 06
Somebody summarises the week’s activity for a landlord or a director every Friday.
02 Where AI fits
Useful, specific and always checkable.
These are the kinds of tasks where we most often put AI to work in a property operation. They are examples rather than a package, and each one sits inside a workflow that a person can see and correct.
Extracting information from documents
Pull the useful fields from applications, referencing reports, certificates, invoices and inventories into your systems, flagged for a quick human check rather than a full re-type.
Prioritising maintenance and requests
Rank incoming repairs and requests by urgency and impact against your own rules, so a serious problem never waits behind routine ones and a person confirms anything consequential.
Summarising history and threads
Give a manager the short version of a long email chain, or a property’s recent history, before they pick up the phone.
Drafting replies for review
Prepare a first draft of routine responses using the information already in your systems, for a person to check, adjust and send. The tone is yours; the typing isn’t.
Quality and consistency checks
Spot missing information, inconsistencies between systems or a record that doesn’t look right, and raise it with someone before it becomes a problem.
Many of these sit alongside the workflows described on our property management automation and letting agency automation pages. For the connections that feed them, see property software integration.
03 Our approach to AI
Technically sophisticated. Commercially pragmatic.
Sometimes the right answer is AI. Sometimes it is conventional automation, an integration or a straightforward piece of software. You shouldn’t have to decide which. That is our job.
AI where it helps.
Reading, classifying, extracting, summarising and drafting. Tasks that involve understanding language and documents, where a good first pass by a model saves a person a great deal of mechanical effort.
Automation where it works.
Moving information between systems, scheduling, notifications and predictable steps don’t need AI at all. Conventional automation is cheaper, faster and more predictable, so that is what we use.
Software where it’s needed.
When the right tool doesn’t exist, we build it. AI may sit inside it, or it may not. The problem decides.
How we keep it safe
Guardrails that come as standard.
- A person reviews and approves anything consequential: money, tenancy matters, legal obligations, anything that affects a customer.
- Every AI-assisted step is logged, so you can see what was suggested, by what, and who confirmed it.
- When the system is unsure, it says so and hands the item to a person rather than guessing.
- It works from information you already hold, and how that information is handled is agreed before anything is built.
- Nothing is presented as magic. If a task is better done by conventional automation, or by a person, we say so.
Will AI make decisions about our tenants, landlords or clients?
No. In the systems we build, AI suggests, sorts, extracts and drafts. People decide. Anything with consequences for a customer, a tenancy, money or a legal obligation is confirmed by a member of your team, and the system is designed so that step cannot be skipped by accident.
Is our data used to train AI models?
We design systems so that your information is processed only for your own workflows, and we choose providers and settings accordingly, including options that do not use your data for training. Personal data is handled in line with UK data protection law, and the specifics of what is processed, where and by whom are agreed with you before anything is built.
How accurate is it?
It depends on the task. Classifying an email or extracting a date from a certificate is very reliable; interpreting an ambiguous message is less so. We measure performance on your real data during the build, design a human check into any step where errors would matter, and make sure the system falls back to a person when it is uncertain.
Do we actually need AI at all?
Quite possibly not, or not everywhere. A surprising amount of the friction in a property business is solved by connecting two systems or automating a predictable step, with no AI involved. We use it where it genuinely earns its place and will tell you plainly where it doesn’t.
Discovery call
Let’s find where AI would actually help.
Show us the reading, sorting and re-typing your team does each week. We’ll tell you honestly which of it AI should handle, which should be conventional automation, and which should stay with people.
- No preparation needed
- No presentation, no pitch deck
- Honest view on whether we can help