Real estate has digitized its data. The next transformation is digitizing work itself.
New research from ProptechOS maps 289 operational processes across the real estate value chain and finds that AI agents can already perform approximately 40% of the total operational workload — not in theory, but in production, today.
Intro
Digitizing work: How AI agents will perform 40% of operational work in Real Estate is a research paper by Per Karlberg, CEO of ProptechOS, presenting the most granular mapping to date of where autonomous software can replace, augment, or support human work in building operations. The finding is independently corroborated by Morgan Stanley (37% of sector tasks automatable) and McKinsey ($430–550 billion in annual value at stake globally).
By the numbers
40%
of total operational workload in real estate can be executed by AI agents
289
operational processes mapped and classified across the real estate value chain
53.3%
of processes are fully or largely automatable
82%
of atomic steps across alarm handling, work order lifecycle, and error/fault management are fully or largely automatable
37%
Morgan Stanley’s independent estimate of automatable tasks across 162 REITs and CRE firms, $92 billion in combined labor costs
$430–550 billion
McKinsey’s estimate of annual value at stake from automation across real estate, construction, and development
36%
reduction in district cooling and heating costs at Vasakronan’s Kista Science Tower — 200,000 SEK per year at a single asset, 6+ MSEK projected across the portfolio
31%
reduction in energy consumption at Eufemia (292,000 kWh/year)
What’s inside
- The 40% finding — how 289 processes were classified into fully, largely, partially, and minimally automatable, and what that means weighted by effort
- The digital-physical divide — why AI agents can own the entire workflow before and after physical intervention, and what that leaves for humans
- Six high-value domains — where automation potential is highest: energy management, predictive maintenance, alarm triage, vendor and financial management, compliance, and tenant services
- Business impact, quantified — productivity, cost reduction, response time, and scalability, with production numbers from Vasakronan and Locum
- Why platform architecture matters — what separates the fewer than 10% of organizations that have scaled AI agents from everyone still stuck in pilots
- The operating model of the future — three tiers of human-AI collaboration, and what changes for the facilities professional’s role
The technology is ready. The question is how fast you can capture it.
Get the full research paper — methodology, the complete 289-process breakdown, domain-by-domain findings, and the production results behind the numbers.
