From Pilots to Performance

TL;DR

• Nine in ten real estate organisations are piloting AI, but only 5% of occupiers say they have met all of their programme goals (JLL, October 2025).
• The share of CRE companies actually scaling AI in production rose six-fold in a year, from 1.5% to 9.7%, while the share stuck in pilots barely moved (Kolena, 2026).
• The bottleneck is not the technology: it is 6 to 24 month sales cycles, integration with incumbent systems such as Yardi and MRI, and buyers who never test a product on their own data.

PropTech Connect Europe 2026 lists "Adoption at Scale: Turning PropTech Pilots into Quantifiable Business Performance" as one of its six themes, promising to "examine what it takes to move from isolated innovation to large-scale deployment", and opened the Arora Stage on 9 September with "Build or Buy: How to Engage Technology?" (10:00 to 10:40). The framing is a tell. Ten years in, the industry is still debating how, not what, to buy.

Where the pilots stall

JLL's 2025 Global Real Estate Technology Survey of more than 1,500 senior decision-makers across 16 markets found 88% of investors, owners and landlords piloting AI, and 92% of occupiers, running five use cases at once on average. Only 5% of occupiers had achieved all of their programme goals. JLL's diagnosis was organisational, not technical: data quality, infrastructure and change management.

Kolena's State of AI in CRE 2026, built on 667 buyer conversations with 277 companies, adds a twist. Companies actively scaling AI in production grew from 1.5% to 9.7% year on year; the share merely piloting moved from 13.0% to 13.7%. The winners are pulling away and the middle is not moving. Two details explain why. Only 14% of buyers had tested a product on their own documents, and 60% had watched a vendor demo without validating it on internal files. The tools that fail, in Kolena's words, fail because "the work lives in Excel, SharePoint, and Yardi, and a general chatbot does not go there".

The RICS 2026 survey, covered in depth in AI across the development lifecycle, agrees: regular use grows, embedded use stays marginal, and integration with existing systems is a top barrier.

Why scale is a procurement problem

What decides whether a product crosses from one building to a portfolio comes down to procurement, not engineering:

• Sales cycles. Enterprise proptech deals run 6 to 24 months and are judged against net operating income, not feature lists (SaaS Hero, 2026). Few startups can fund many of them.
• Incumbent systems. Deloitte's 2026 outlook, surveying more than 850 executives at owners and investors, found 22% running industry-specific platforms such as integrated workplace management systems, and 27% reporting implementation challenges. Yardi, MRI and CoStar "serve fundamentally different functions", one integration architect notes, and treating them as interchangeable is "the root cause of most failed integration attempts".
• Build, buy or partner. Owners build on platforms they already license, buy where the vendor owns the integration, and partner where the data is the asset. None of the three is a default, which is why the Arora Stage session exists.

"Portfolio-wide" therefore means a product that works inside the existing property-management and ERP stack, survives procurement, and reports a number the asset manager already tracks. which is why where UK proptech money goes is tilting towards companies that sell results, not tools.

An evidence ladder for buyers

Vendors and owners talk past each other because "adopted" means something different at each rung. Name the rung first.

• Pilot: one asset, one team, vendor-run, success defined loosely.
• Proof of value: tested on the buyer's own data, against a baseline the finance team accepts, with integration into at least one live system.
• Multi-asset rollout: repeated across several assets or funds with the same configuration, owned by the buyer's operations team, not the vendor.
• Portfolio standard: written into the operating model, procurement templates and reporting, so a new acquisition inherits it by default.

On Kolena's figures fewer than one in ten CRE companies sit on the top two rungs. The programme decoded shows how much of London's agenda is aimed at that climb.

<!-- SOURCES

• Theme wording "Adoption at Scale: Turning PropTech Pilots into Quantifiable Business Performance", the description "what it takes to move from isolated innovation to large-scale deployment", and the session "Build or Buy: How to Engage Technology?" (Arora Stage, 9 September, 10:00-10:40): https://www.proptechconnect.com/event-europe/agenda/
• JLL 2025 Global Real Estate Technology Survey (28 October 2025): 1,500+ senior CRE decision-makers, 16 markets; 88% of investors/owners/landlords and 92% of occupiers piloting AI; average five use cases; 5% of occupiers achieved all programme goals; 47% achieved two to three; readiness diagnosis and 60%+ of investors lacking readiness: https://www.jll.com/en-us/newsroom/real-estates-ai-reality-check-companies-piloting-only-achieved-all-ai-goals
• Kolena, The State of AI in CRE 2026: 667 conversations, 277 companies (September 2025 to July 2026); scaling in production 1.5% to 9.7%; piloting 13.0% to 13.7%; active execution 14.9% to 24.5%; 14% tested on own documents; 60% saw vendor demos only; quotation on Excel, SharePoint and Yardi: https://www.kolena.com/real-estate/report-the-state-of-ai-in-cre-2026/
• RICS AI in Commercial Property and Construction Report 2026 (18 August 2026), integration with existing systems as a leading barrier: https://www.rics.org/news-insights/ai-in-commercial-property-and-construction-report-2026
• Deloitte 2026 Commercial Real Estate Outlook: 850+ C-suite executives and direct reports at owners and investors, fielded June and July 2025; 22% using industry-specific platforms such as IWMS or CAFM; 27% experiencing implementation challenges (technical issues, lack of expertise, resistance to change): https://www.deloitte.com/us/en/insights/industry/financial-services/commercial-real-estate-outlook.html
• 6 to 24 month PropTech sales cycles, NOI-driven buyer evaluation, integration with Yardi and RealPage (SaaS Hero, Aaron Rovner, May 2026): https://www.saashero.net/strategy/proptech-gtm-strategy-2026/
• Yardi, MRI Software and CoStar "serve fundamentally different functions"; treating them as interchangeable is "the root cause of most failed integration attempts" (TFSF Ventures): https://www.tfsfventures.com/blog/proptech-integration-architecture-for-agents-consuming-yardi-mri-and-costar
• The four-rung evidence ladder (Pilot, Proof of value, Multi-asset rollout, Portfolio standard) and the build/buy/partner characterisation are editorial synthesis, not quotations.

-->