“When everyone can code, the competitive advantage shifts from having developers, to knowing which problems are worth solving.”

EXECUTIVE SUMMARY

Why Real Estate Is Looking in the Wrong Direction The industry obsesses over where we work (office, home, hybrid) whilst missing the more fundamental shift: what work we do.

Location is the output, not the input.

The microsoftware revolution changes everything. AI tools like Lovable and Cursor now let anyone create bespoke applications using plain English. The economics flip when software development drops from thousands of pounds to one hour of professional time plus minimal API costs.

Power shifts from IT to end users. When domain experts can solve their own micro-problems without coding or procurement, organisational dynamics transform. The "long tail" of previously uneconomical workflow needs becomes instantly addressable.

For real estate, this is seismic. As work becomes defined by orchestrating AI agents and custom microsoftware rather than location-dependent tasks, the link between productivity and physical space weakens further. Future demand will be shaped less by hybrid policies and more by the fundamental reimagining of work itself.

IT’S THE WORK WE DO THAT MATTERS

In real estate everyone is fixated on ‘where we work’ as a signal to the future nature of demand. In the office, at home, in 3rd places, or a hybrid of them all. The mix matters. Or so everyone seems to think.

There is though a much more important driver of the future of demand, and that is ‘the work we do’. The nature of the tasks we have to perform, to achieve the basket of goals we need to satisfy, to fulfil the requirements of our jobs, is a massively more important variable than where we do them. Because where we do something is a function of what it is we have to do. Where we work is an output, not an input. It is correlation, not causation.

And of course, it is technology that changes the ‘work we do’. Always has done. Electricity changed the nature of factories, and the computer, first the mainframe, then the PC, then the laptop, and then mobile devices all led to profound changes in offices and beyond.

And the Internet and then WiFi did so again.

And as hardware, and software, developed we all changed the ‘work we do’, which in turn led to the breaking of the link between work and space. We no longer need to be in place A to do task B.

So fixating on where we work is to miss the point. To be looking in the wrong direction.

ALL CHANGE, ALL CHANGE

AI is set to change things again, probably quite dramatically.

A few weeks ago, in ​‘Intelligent ‘AI’ Agents and Real Estate Deman​d’​ ​we looked at the coming era of Agents and Agent Bosses, where each of us will be primarily working with, orchestrating, a slew of virtual assistants. But something else is also brewing.

The age of “Microsoftware” or “Workflowware.”

And this will be as significant as the introduction of spreadsheets was.

THE ECONOMIC CONSTRAINTS OF THE PAST

Historically, software development followed a capital-intensive model. High fixed costs meant only problems shared by many got solved. The traditional software market serves the "head" of the demand curve - problems common enough to justify building a commercial product (e.g., a CRM).

So ‘the long tail’ never got serviced. An almost endless set of micro requirements that would be super useful, but perhaps only for a limited time, or for a small group of people.

‍AI-guided automation now makes it feasible to serve this long tail. And this opens up an enormous marketplace of heterogeneous unfulfilled needs.

LOVING LOVABLE

Let me give you an example.

On my website I have an index page of all my blog posts going back to 2013. Hundreds of thousands of words. I’ve long wanted to be able to extract the content of each of these posts, save them as pdfs, and move them into a folder on my Google Drive.

But how?

Until now I had no idea. I could do it manually but that would:

1. Take ages
2. I’d die of boredom doing it.

So my drive remained empty.

This weekend though I used an AI powered software development tool called ​Lovable​ (Lovable.dev) and simply told it what I wanted to do, in plain written English.

From there it produced a baby application I could use to fulfil this task.

We went back and forth a bit (mostly with me saying ‘What do I do now?’) and I had to set up a new account with a ‘scraping’ service called ‘Firecrawl’, as well as set up a ‘Zap’ on my Zapier account. Both of which are cheap, non-tech friendly, and can be ‘mastered’ in not a lot of time.

But essentially the whole process was via ‘natural language computing’. I did not write any software. I just asked the existing software to write it for me.