If there is one place where the invisible work of a property management firm piles up, it is the inbox. Hundreds of emails a day: supplier invoices, claim reports, questions from residents, requests from insurers, quotes, advertising and, mixed in with all of it, the things that matter.
The property manager spends a considerable part of the working day on a task that adds nothing: reading an email, deciding what it is about, working out which community it belongs to and, if it carries an invoice, opening it, reading the amounts and typing them into that community’s accounts.
It is mechanical, repetitive, error-prone work, and it is exactly the kind of work a person should not be doing.
Connect the email you already use
The starting point matters: there is no need to change email address or ask anyone to write somewhere else. FixrOS connects to the firm’s email account, the one it has always had, and works on top of it. The emails stay where they were; what changes is that they arrive sorted.
Two levels of classification, and there is a reason
The system does not classify in one way, but in two, and the design is deliberate.
First, rules
The first layer is a rule-based classifier, not a language model. It analyses the subject, the body and the sender, and looks for vocabulary specific to the sector to decide the category. The categories are closed: invoice, supplier, insurer, community, enquiry and other.
It may seem less sophisticated, and that is on purpose. For a closed set of categories with very distinctive vocabulary, well-made rules are more reliable and more predictable than a badly configured AI. And if nothing fits, the email falls into «other» instead of a category being made up.
Then, real understanding
When the email calls for it, a second layer with artificial intelligence comes in that does actually read: the body of the message, the attached PDFs and the images. From there it draws three things: the category, a summary of what the email is asking for and a draft reply written in the tone of a property manager.
The draft is not sent on its own: it is proposed. The property manager reads it, adjusts it if needed and sends it. The difference between writing an email from scratch and reviewing one that is already written is, in volume, several hours a week.
And if the AI fails or is unavailable, the system does not grind to a halt: it falls back to the rule-based classifier and keeps working. No email is lost because an external service is having a bad day.
The invoice that posts itself
This is the feature that gives back the most time. When an email carries an attached invoice, the system detects it, reads it and posts it to the community it belongs to.
It is not just extracting an amount. It is identifying the supplier, recognising the invoice details, determining which community in the portfolio that expense belongs to and generating the entry in its accounts with the right account.
Think about what that means in a firm with twenty communities: the lift company, the cleaning company, the gardening company, the boiler maintenance, the insurance policies. Dozens of invoices a month that until now were opened, read and typed in one by one.
The email that carries an incident
The other automation is the one for incidents. When a resident writes to report a breakdown, or a supplier reports a problem detected during an inspection, that email is synchronised with the incidents module and stops living only in the inbox.
It matters because an email cannot be tracked: it has no status, no owner, no resolution date. An incident does. Turning the one into the other is what prevents a legitimate notice from sinking beneath two hundred later emails.
A product decision that deserves an explanation
There is a deliberate absence in the list of categories: «spam» is not a category of the classifier. It is a conscious decision. An email automatically marked as junk is an email nobody is going to read, and in a property management firm the cost of losing a legitimate email (a formal request, a claim report, a message from a resident) is far higher than the cost of seeing advertising.
The system’s general criterion is the same: when in doubt, a false positive is preferable to a false negative. The property manager decides in the end; the tool proposes.
Frequently asked questions about the AI email inbox
Do we have to change email address?
No. The system connects to the account the firm already uses. The emails keep arriving where they used to arrive.
Does the AI reply to residents automatically?
No. It prepares a draft and leaves it ready for review. Sending is always decided by a person.
What happens if it misclassifies an email?
The property manager corrects the category. The system is designed on the assumption that it will get it wrong some of the time, and that is why no relevant decision (sending, posting definitively to the accounts) is taken without human review.
Can it read invoices in scanned PDF?
The second layer processes both PDF documents and attached images, which covers the scans and the photos of invoices that some small suppliers send.
What about the confidentiality of the firm’s email?
The content is processed in order to provide the classification and capture service, within the applicable data protection framework and the confidentiality commitments of the contract. It is worth reviewing in detail during onboarding, because it is a legitimate question.
Does it replace the accounting software?
It does not replace it: it feeds it. What disappears is the manual step of reading an invoice and typing it in.
You may also be interested in
- FixrOS for property managers
- Electronic invoicing and Verifactu for property managers
- AI in property management: real cases
- FixrOS accounting
Invoice capture is one of the features that gives back the most time and one of those you need to check when comparing property management software.
Conclusion
The inbox is where a firm loses the hours it does not bill. Classifying, summarising, writing the draft and posting the attached invoice are mechanical tasks with clear rules. Automating them does not take judgement away from anyone: it gives it back, because the time that today goes into typing amounts is the time that should be devoted to the decisions that do require a professional.