In March 2026 McKinsey published the report How agentic AI can reshape real estate's operating model. It describes how AI moves from answering questions to doing the work – and what it takes for that to work in a property company. Here is what matters most, summarised by us.
Read the full report at McKinsey →
From understanding to getting things done
In recent years property companies have tried AI to summarise leases, draft memos and speed up reporting. It helps, but according to McKinsey it rarely changes the outcome. The next step is AI that carries out several steps in a row, inside the company's systems, with approvals and logging.
The value
The McKinsey Global Institute estimates that automation can free up roughly 430–550 billion dollars a year globally across real estate, construction and development. The figure includes both practical work and work with documents, numbers and follow-up. The estimate is based on 48 countries.
It all rests on a fact layer
The report describes five technical layers. The first is the fact layer: data and documents must be usable, properties, premises and tenants must be the same across systems, information must be retrievable reliably from documents – and there must be a source of truth when the systems disagree. Without that, the rest becomes demonstrations that cannot scale.
Agents take the steps, people the judgment
McKinsey divides the work into steps and judgment. The steps – gather the material, send it for approval, update the status – can be automated. The judgment – exceptions, negotiations, decisions that carry risk – stays with people. Trust is built with clear approvals, traceability and visible uncertainty.
Four areas with the greatest value
Maintenance and operations, leasing and renegotiation, investment and asset management, and construction projects. In asset management the report points out that contract terms and important dates are often locked in documents.
Who owns the data and the learning?
In a few years the most successful owners, according to McKinsey, may resemble operating systems more than collections of properties: a shared layer of workflows, data and control. The hunt for documents and status shrinks. Those who succeed are those who own the workflows and the learning.
Source: McKinsey & Company – Alex Wolkomir, Ankit Kapoor, Vaibhav Gujral, Andrei Stoica, How agentic AI can reshape real estate's operating model, 4 March 2026. Read the full report →
What it means
For what McKinsey describes to work, documents and data must be usable and the owner must keep them. nextopia gathers the property's documents and data, shows what is there and what is missing and makes the information available to people and AI.