TL;DR
- AI use has passed three-quarters in commercial property and two-thirds in construction, which a year earlier was at just over half (RICS, August 2026).
- Both sectors run early-stage pilots at the same rate, 39%, but commercial property converts them into regular use far better: 29% against 19% in construction.
- The UK remains the most cautious region: 35% of UK construction respondents have no plans to invest in AI, down from 44%, and only 5% expect a significant increase.
For years, proptech treated AI as a point tool: a valuation model here, a chatbot there. Now AI runs through every stage of a development, and survey data published three weeks before PropTech Connect Europe shows the industry has moved from curiosity to use. The harder question, the one worth taking into the Development Stage in London, is why so little of that use has become routine.
From a point tool to a through-line
Follow a scheme from start to finish and the applications stack up:
- Site acquisition: screening land and assets against ownership, planning and viability data at a speed no analyst can match.
- Design and planning: testing massing, density and compliance scenarios before a drawing is finished.
- Construction: forecasting programme risk and detecting clashes before costs are sunk.
- Sales and operations: pricing, lead handling and building performance, the parts closest to revenue.
The organiser frames its Development Stage in exactly these terms: "address the new economics of construction by integrating data analytics and targeted workflows to shield delivery margins, fix critical execution errors, and build higher-quality schemes faster without eroding viability".
What the numbers say
The RICS AI in Commercial Property and Construction Report 2026, drawing on 3,148 responses to its Q1 2026 monitors, is the most recent large sample. Its headline is adoption: about two-thirds of construction respondents now use AI in some part of their work, up from just over half in 2025, and more than three-quarters of commercial property respondents do.
Its second finding is the one developers should read twice. Both sectors are piloting at the same rate, 39%. But commercial property reports regular use at 29%, against 19% in construction, and widespread or full integration at 6% against 4%. The industry has stopped experimenting and started struggling to convert. RICS concludes that the industry's challenge has moved from persuading organisations to try AI to embedding it in professional practice.
The UK sits at the cautious end. It recorded a 17% fall in "no AI use", the largest single movement in the RICS construction data, and regular use nearly doubled, yet 35% of UK construction respondents still have no plans to invest, the highest anywhere. In commercial property the leading barriers are data quality and integration with existing systems; in construction, skilled staff.
Where the capital is going
In the first half of 2026, three AI companies alone took 29% of all UK equity capital raised and pushed the average round to a record £5.4m (Beauhurst). Proptech competes for capital inside that pull, and where the UK's proptech money goes increasingly rewards AI-driven results over AI features.
The delivery bottleneck
The honest limit is rarely technical. A developer can compress site analysis from weeks to hours, but the system it feeds into does not move at software speed. Real Estate:UK's 2026 investor report points to delays at the Building Safety Regulator and a construction PMI of 45.6 in March 2026, the fifteenth consecutive month of contraction, as headwinds that no model removes. The sharpest tension in the "site to sale" story is that the analytical half of the lifecycle keeps accelerating while the delivery half does not.
On the agenda in London
The published programme is built for exactly this conversation. On 9 September the Arora Stage opened with "Build or Buy: How to Engage Technology?" at 10:00 and returned to the theme with "From Data to Decisions: Powering the Modern Real Estate Firm" at 14:05. Neither is a demo of what the models can do. Both ask how a real estate firm gets from pilots to performance, which, on this year's evidence, is the only question that still matters.
<!-- SOURCES
- RICS AI in Commercial Property and Construction Report 2026 (published 18 August 2026): 3,148 responses (GCM 1,883, GCPM 1,265); about two-thirds of construction respondents use AI, up from just over half; more than three-quarters in commercial property; pilots 39% in both; regular use 29% commercial property vs 19% construction; widespread or full 6% vs 4%; UK 17% reduction in no AI use, "the largest movement in the GCM data"; UK no plans to invest 35% down from 44%, significant increase 5%; commercial property barriers data quality and availability, integration; challenge shifted from experimenting to embedding: https://www.rics.org/news-insights/ai-in-commercial-property-and-construction-report-2026
- Same report as covered by BE News, including the conversion-of-pilots framing: https://benews.co.uk/ai-use-by-construction-and-property-firms-continues-to-accelerate-report-finds/
- Three AI companies took 29% of all UK equity capital in H1 2026; average round a record £5.4m: https://beauhurst.com/research/the-deal-h1-2026/
- Building Safety Regulator delays; S&P construction index 45.6 in March 2026, fifteenth month below 50 (Real Estate:UK and CoStar, Who invests in UK real estate 2026): https://realestateuk.org/media/n01dtxc5/who-invests-in-uk-real-estate-2026_costar_and_realestateuk.pdf
- Published sessions "Build or Buy: How to Engage Technology?" (Arora Stage, 9 September, 10:00-10:40) and "From Data to Decisions: Powering the Modern Real Estate Firm" (Arora Stage, 9 September, 14:05-14:45); Development Stage description; nine stages: https://www.proptechconnect.com/event-europe/agenda/
- Lifecycle framing (site acquisition, planning, construction, sales and operations) is editorial synthesis, not a quotation.
-->
