Trebellar Raises $18M Series A to Bring AI Agents to Corporate Real Estate

Trebellar has raised $18 million in Series A funding as it looks to bring AI agents and automated decision-making to one of the enterprise functions that has remained unusually dependent on spreadsheets, disconnected software and outside consultants: corporate real estate.

The round was led by Blossom Capital, with participation from Haystack, Alt Capital, 1Flourish and Bynd. Trebellar says the new capital will be used to expand its engineering and go-to-market teams, deepen its AI capabilities and grow its enterprise customer base.

The San Francisco-based company has already attracted several major customers, including Meta, Uber, Merck and Cohesity, giving it an early foothold among organizations managing large and complex real estate portfolios.

For Trebellar, the opportunity is not simply to replace spreadsheets with another dashboard. The company is attempting to create an AI-native layer capable of understanding an organization’s real estate portfolio, continuously analyzing how that portfolio is being used and helping teams make decisions about everything from leases and office utilization to future locations.

Corporate Real Estate Has a Data Problem

Large companies generate enormous amounts of data about their physical footprint.

Lease agreements contain financial commitments and renewal dates. Badge systems and WiFi networks provide signals about office attendance. Room-booking platforms show how employees use meeting spaces. Human resources systems contain headcount and organizational information, while surveys can reveal whether employees actually find an office useful.

The problem is that much of this information traditionally resides in separate systems.

Before a corporate real estate team can decide whether to renew a lease, shrink an office, expand into another market or change a workplace policy, analysts may first have to collect and reconcile information from spreadsheets, PDFs and multiple software platforms.

Trebellar is building its platform around a unified data layer designed to aggregate and normalize those sources before applying AI to them. The system can incorporate lease and operating expense data, market information, occupancy data from badges, WiFi and sensors, space-reservation systems, HR platforms and employee surveys. Companies can also add information through APIs, CSV uploads and internal research or policy documents.

That data foundation is important because enterprise AI becomes considerably more useful when models can reason over structured operational information rather than a collection of isolated documents.

Moving From Dashboards to AI Agents

Once that information is connected, Trebellar’s AI platform applies a combination of AI agents, large language models and machine learning models to corporate real estate decisions.

Instead of requiring users to manually assemble dashboards, Trebellar allows teams to ask questions about their portfolios and generate reports, recommendations or analyses around specific problems.

Its task-specific agents can handle areas such as portfolio management, location strategy and planning. The platform can analyze lease and location performance, conduct location scoring, model different supply-and-demand scenarios and generate research on potential markets.

Predictive models add another layer. Trebellar says its technology can forecast office attendance, detect unusual changes in workplace patterns and help companies estimate how staffing or services should be adjusted based on expected usage.

That moves the software closer to a decision-support system than a traditional business intelligence product.

For example, a company evaluating whether it needs a particular office could potentially combine lease costs, utilization trends, employee locations, commute times and workplace sentiment rather than evaluating each variable independently.

Scenario modeling also allows teams to test how changes to their portfolio might affect cost, capacity or employee access before making a decision.

AI Built Around Real Estate Data

Vertical AI companies increasingly differentiate themselves by building systems around the data structures and workflows of a specific industry rather than adding a general-purpose chatbot to existing software.

Trebellar is taking that approach with real estate.

The company says its models are designed around occupancy, workplace and lease data, while its AI agents are trained to perform specific real estate analyses. Its location strategy technology, for instance, can combine public information with market, demographic, transportation and talent data when evaluating potential locations.

There is also an enterprise security component to the architecture.

Trebellar is SOC 2 Type II certified, and the company says machine-learning models that depend on customer data are trained independently for each organization rather than pooling customer datasets. Trebellar also operates its own machine-learning infrastructure and supports open-source large language models, while using abstraction mechanisms when external enterprise LLM providers are involved.

Those safeguards are particularly relevant for corporate real estate, where datasets can indirectly reveal sensitive information about office attendance, company expansion plans and organizational structure.

From a Side Project to Enterprise Real Estate AI

Trebellar’s origins are somewhat removed from commercial property.

The company traces its beginnings to the pandemic, when co-founders Diego Ferreiro Val and David Garcia Quintas began experimenting with a centralized home automation project. Ferreiro Val had been a VP of Engineering at Salesforce, while Garcia Quintas brought AI and infrastructure experience from Google and Waymo.

The home automation product never launched.

Instead, the project attracted the attention of Salesforce’s workplace technology organization, where teams were dealing with a larger version of the same problem: large amounts of building and workplace data spread across disconnected systems.

That realization eventually became Trebellar.

The company has since expanded the team with experience spanning Salesforce, Google, Waymo and Verkada. It has also added former Google real estate executive Dave Radcliffe as an advisor. Radcliffe helped oversee the expansion of Google’s physical footprint across the tenures of Eric Schmidt, Larry Page and Sundar Pichai.

$18M Series A Targets a Large but Overlooked Enterprise Function

The new financing arrives as investors continue searching for enterprise categories where specialized AI systems can replace manual analysis and fragmented workflows.

Legal, finance, customer support and software development have already produced a growing group of vertical AI companies. Corporate real estate presents a similar opportunity, but with an unusually complicated combination of financial, operational and physical-world data.

Trebellar CEO Diego Ferreiro Val argues that real estate teams should be able to conduct more of that analysis internally rather than depending on spreadsheets and consultants.

“We built Trebellar to give leaders the independence to make these decisions themselves,” Ferreiro Val said.

The next stage will test whether that approach can scale beyond early enterprise adopters.

With $18 million in new funding, Trebellar plans to invest further in its AI technology while expanding its engineering and commercial teams. The larger opportunity is to turn corporate real estate from a periodically analyzed collection of leases and buildings into a continuously updated dataset that AI agents can reason over.

If that transition takes hold, decisions about where companies locate employees, how much office space they maintain and what they spend on physical infrastructure could increasingly become another enterprise workflow shaped by AI.