“The future of real estate will be written in prompts.”

A Shared Set of Problems and Processes

We believe the global real estate industry shares a surprising truth: beneath the surface-level diversity of buildings, markets, and geographies, the problems we face—and the workflows we use to solve them—are strikingly similar.

Whether you’re leasing an office in London, developing a logistics park in Jakarta, or managing a data centre in São Paulo, 80% of your work is driven by a common 20% of processes. These high-impact workflows—leasing analysis, capex planning, asset marketing, tenant onboarding—repeat across the world, across asset classes.

Our hypothesis: If we can clearly define those workflows, and develop expertly structured AI prompts to execute their key tasks, we can massively accelerate how real estate work gets done.

Enter Prompts, and the Rise of AI-Accelerated Workflows

By ‘prompt’, we mean a structured set of instructions that tells an AI model what to do, what data to consider, how to process it, and how to present the results. Prompts are not mere queries—they are decision frameworks.

Now, with this in mind, and taking into account the rapidly developing capabilities of LLMs and AI, we believe that many, perhaps most, of these workflows that are common to all, can be greatly expedited by the development of great AI prompts.

Let’s say one of these workflows contains the Goal of Finding the best tenants for a retail space. To achieve this goal you’d need to complete the following five tasks:

1. Identify the ideal tenant profile
2. Analyse foot traffic data
3. Compare market rents
4. Generate outreach emails
5. Summarise findings for stakeholders

From Tasks to Structure: Inputs, Processing, Outputs

When you dive into how to achieve these tasks, you realise they all follow a pattern.

• First, what inputs are required?
• Then, how should those inputs be processed?
• Finally, what should the output look like?

This is the decision-making structure underpinning much of our industry: Inputs → Processing → Outputs. And it’s exactly how effective prompts work too.

Prompts Are Structured Intelligence

A good prompt provides clarity, context, and constraints. It defines:

• What to do
• What information to use
• How to process that information
• What format the output should take

And as LLMs gain increasingly agentic capabilities—like those emerging from OpenAI’s o3 ‘reasoning’ model—we can now execute complex, multi-step, multi-tool workflows from a single prompt.

Codifying Domain Knowledge in Prompts

If we can define the most important goals and write suitably advanced prompts for the tasks needed to fulfil them, we end up with something enormously valuable:

• Reusable solutions to industry-wide problems
• Transparent workflows that reveal the underlying logic and nuance
• Blueprints for how to operationalise real estate knowledge in AI systems

For example, a leasing manager could use a prompt to analyse tenant mix risk, process it through market comparables, and generate a decision memo—all in one connected, logically consistent manner.

This is the foundation of ​The Built World Challenge​.