Software is finally connecting, and the way it arrived is not what any of us expected.
In Australia, the single biggest complaint across every form of real estate business is the same one: the systems don’t talk to each other. Every office runs a CRM, a trust accounting package, a forms platform, a marketing tool, a portal uploader and a phone system, and almost none of them share a brain.
More and more, I am encouraging property professionals to supply that brain themselves. To stop thinking of their chosen LLM as a writing tool and start treating it as the thing that sits above all the software, reads across it and helps make sense of information that individual systems cannot see on their own.
Very few property professionals are using it that way. I estimate that the average property professional is using less than 1% of what the tool in front of them can already do, and that gap has nothing to do with the technology. Asking it to write a listing description is using a brain to hold a pen.
What has changed in the last twelve months is that the brain can now reach the rest of your business. From sales prospecting to property management, the opportunities are opening up across almost every part of an agency. Admin data entry, forms and contract completion are among the harder areas to connect, largely because they sit inside closed compliance-grade systems that are not easily accessible to outside technology.
Everything else is going.
And I suspect loyalty to a software provider will count for very little if a competing product can do more simply because it connects. I know that because I did it myself. I recently ditched video software I had used for twenty years and moved to a product I had never heard of, purely because it supported MCP.
What MCP actually is
MCP, or Model Context Protocol, is an open standard that lets an AI connect to software, data and tools without a developer building a custom bridge every time. It was released by Anthropic in late 2024 and has since been picked up across the industry, so it is no longer a one-vendor idea.
Think of it as a universal adaptor. Before MCP, connecting your AI to your CRM meant a project. With MCP, the software publishes what it can do and what it holds, and an AI that supports the standard can access the functions and information the provider makes available, subject to the necessary permissions. The software provider does the work once, not once per customer and not once per AI vendor.
That is the important part. It shifts integration from a negotiation between two companies to a standard your provider either supports or doesn’t.
Only a year ago, the gold standard for software buyers was an open API. Today, a product without a properly functioning, accessible API is not just behind the times; it is making it harder for customers to use the software they are already paying for. If software providers believe their customers are only using a fraction of their capability, charging them extra for the mechanism that unlocks the rest is counterproductive.
An API is now the floor, not the ceiling.
The 95% you’ve never used
Every software provider in this industry has told you the same thing at some point: you’re using maybe 5% of what you’re paying for. Whether that figure is accurate or not, the underlying problem is real. The reason has never been laziness – it’s that the other 95% required a human to sit down and learn it, and nobody in a real estate office has ever had that afternoon spare.
Connected AI could change that. The LLM reads what the software can do and works out how to use it. You describe the outcome you want in plain English, and it can help find the function you’ve never opened, in the menu you’ve never clicked. You are no longer buying capability you have to learn; you are buying capability your brain can reach.
That is the whole shift, and it is why the API question matters. A model can only reach the functions a provider chooses to expose. If they expose everything, you finally get the software you actually paid for. If they expose nothing, you are stuck at 5% forever.
How the work actually gets done
A connector is the piece that plugs a system into the model. Once your CRM, calendar, email, phone system and portal data are connected to the same assistant, the assistant stops being a chatbot and starts being an operator. It doesn’t just tell you what to do – it reads across the systems, identifies what needs doing and, where authorised, can take action.
Practically, that looks like:
- A sales team whose activity is in the CRM at the end of the day without anyone typing it in. Calls, texts, appraisal appointments and inspection conversations, logged against the right contact, with the next action set.
- Lead detection and nurture inside your own database. Not a dumb drip campaign – a model reading contact history, ownership length, previous appraisal activity and market movement, then surfacing owners who warrant a closer look and drafting the approach for each one.
- Prospecting outside your database, using publicly available information, to identify target owners and streets you have no relationship with yet, subject to privacy and marketing requirements.
- Listing presentation prep, comparables, and suburb commentary assembled from live data instead of a Friday afternoon of copy-and-paste.
- Property management arrears, routine inspections and lease expiries triaged and prioritised before a property manager starts working through the list.
None of that is theoretical. I am doing these things myself, and I have seen them operating inside Australian real estate businesses.
What this does to the competitive field
This is levelling the playing field in a way we have not seen before. The operational advantage large franchises have built over the last two decades – centralised marketing teams, data teams and systems teams – is being compressed into something a two-person independent can access for a monthly subscription. The gap that used to be bought with scale is now bought with comprehension.
It is also reshaping offshore staffing. A trained, connected model can take on some of the repetitive work traditionally handled by offshore administration teams, potentially changing the economics of those arrangements. That is a real commercial shift, and it deserves to be said plainly rather than dressed up.
The risks nobody is pricing in
Three of them matter.
Dependence on a single model. If your entire operation is hard-wired to one provider and that provider has an outage, a price change or a policy shift, you could find yourself unable to carry out critical work. The ability to move between models needs to be designed in from the start.
Data governance. Connectors give a model access to your business and, increasingly, the ability to act within it. Who can see what? What can the model change? What requires human approval? What leaves the country, what gets retained, and what is recorded in an audit trail? Your obligations under applicable Australian privacy laws and industry regulations are now operational questions, not IT questions.
Internal comprehension. The technology is moving faster than the average office’s ability to understand it.
That is the real one.
Calling it
2027 is the year proptech providers sink or swim on this. The ones without open, connected architecture will not be outcompeted on features. They will simply be unreachable, and the market will route around them.
The real estate community should be excited. Connected software is what we have been demanding for twenty years. I suspect, though, that the way it has arrived is not the way most of us anticipated, and the old problem of comprehension and change management will be what stops some offices from moving at all.
We are entering a time when the tools are no longer the bottleneck.
We are.