The Next Real Estate CRM Will Not Just Store Data. It Will Run Work

Deloitte’s latest Australian AI survey caught my eye. Sixty-nine per cent of Australian organisations say they are already using autonomous AI agents, but only 22 per cent have advanced governance, and only 30 per cent say AI is deeply changing how work gets done. This sounds less like transformation, and more like a market still testing tools around the edges.

That feels relevant to real estate because a lot of agencies are in much the same place. AI is turning up in listing copy, prospecting, follow-up, search, note-taking and marketing, but the office underneath it still works much as it did before. The CRM holds the records, the agent or administrator still carries the process. The software remembers things, but people still have to push the work through the machine.

What seems to be changing now is the definition of what software is actually there to do. Salesforce says the main blockers to scaling AI agents are visibility and control, not access to the model, and Agentforce 3 adds a command centre, interoperability support and more than 100 prebuilt actions. ServiceNow is making a similar point, describing EmployeeWorks as a front door that turns natural-language requests into governed, end-to-end execution. That is a long way from the older software model where the system mostly stored data, showed fields on a screen, and waited for a person to do the actual work.

That is why I do not really see this as a CRM story in the old sense. It is more a service delivery story. A lot of the work sitting around a sale is structured enough that software can now start carrying pieces of it. Not the parts that rely heavily on judgement or negotiation, but the more repeatable parts of agency life... a new listing enquiry that needs to be qualified properly, an appraisal request that should trigger the right follow-up, a contract request that should update the timeline, a buyer call that should become a clean note and next step, open home feedback that should flow into vendor reporting, or property facts that should move once through the system instead of being rewritten three or four times.

There is also a broader market conversation sitting behind this. The language around software is shifting. People are increasingly talking about 'service as software', where software starts delivering more of the actual service rather than simply helping staff do it. Forbes contributors have been writing about that shift directly, and Reuters reported sharp selloffs in software, data and professional-services stocks after Anthropic released new automation tools, because investors started reassessing businesses built around seat growth, labour intensity, or the assumption that software stays as a passive layer underneath human work.

That lines up with something real estate has probably under-appreciated for years. Most agency CRM systems still behave like memory banks with reminders attached. They store contacts, notes, dates, tasks and fragments of communication. Useful, yes, but still mostly passive. They tell you what happened, or what should happen next, but they do not really carry much of the process. At the same time, a lot of intelligence is already locked away inside those databases. Domain’s LeadScope was built around exactly that point, using CRM and property data to identify properties in an agent’s database that are more likely to come to market within the next 12 months. That is a useful signal, because it shows the data has been sitting there all along. The difference now is that AI gives software far more ways to act on it.

That might mean a system that captures a new enquiry, recognises the likely intent, drafts the first reply, sets the follow-up sequence, updates the timeline, alerts the right person, and prepares a clean internal summary without anyone needing to push five buttons in three different systems. It might mean vendor reporting that builds itself from the actual campaign data, buyer activity and communication history. It might mean marketing workflows where the same verified property facts feed listing copy, brochure copy, social copy and internal notes without being re-entered or reinterpreted each time. None of that is especially glamorous, but it is exactly the sort of work that fills a lot of the day in real estate offices.

You can see the same pressure in professional services more broadly. Thomson Reuters says organisation-wide AI use has risen* to 40%, more than 80% of current users engage with it weekly, more than 90% expect it to become central to workflow within five years, and only 18% say their organisations are tracking ROI. What stands out there is the mismatch between adoption and measurement. A lot of businesses are clearly using the tools, but they are still working out where the value actually sits and how much of the surrounding service model starts to change when the software carries more of the work.

That feels familiar in real estate. A lot of value still leaks out through the same old gaps... information entered twice, tasks carried manually, updates dependent on memory, fragmented communication, and support layers built around weak handoffs between systems. Once software starts taking on more of that structured work, the question stops being which CRM has nicer fields or a better mobile app. The question becomes which system is closest to the actual flow of work, and which business has structured itself to let the system carry more of it.

And if that is where things are heading, then the next real estate CRM story will not mainly be about contact and property management, it will be about service delivery.

That is where this gets commercially interesting for smaller agencies and independent operators as well. If more of the structured support layer can be delivered through the software, then the amount of labour needed around the core human parts of the job starts to change. Trust, judgement and negotiation still matter but a lot of the work wrapped around those strengths is much more structured than the industry often admits, and that is exactly the layer software is getting better at carrying.

So when I look at where Real Estate CRMs are heading, I do not really see a database story anymore. I see a service story. And I suspect that over the next few years, that will matter more than most people in real estate currently realise.

• Download report: https://www.thomsonreuters.com/en/reports/2026-ai-in-professional-services-report