Trebellar Raises $18M for AI Corporate Real Estate

Corporate real estate decisions age while teams assemble the evidence. A lease changes, headcount moves, utilization shifts, and a portfolio plan that looked complete last month can return to the meeting missing the one fact that changes the call.

Trebellar has raised an $18M Series A to make that evidence usable before the decision expires. Blossom Capital led the September 24, 2026 financing, with Haystack, Alt Capital, 1Flourish, and Bynd Venture Capital participating.

The San Francisco company builds an AI-native platform for corporate real estate. Trebellar unifies lease, cost, headcount, space, utilization, commute, and employee-sentiment data in a corporate-real-estate-specific ontology, then gives agents a governed model from which to gather, reconcile, analyze, and report.

The broader bet is that vertical AI will reach consequential enterprise decisions through the data model underneath the interface. A generic assistant can answer a question. A real estate team still needs to know which lease record, headcount forecast, badge count, cost assumption, and policy choice produced the answer before it commits millions of dollars and years of occupancy.

What Happened

Trebellar's $18M Series A is the company's largest disclosed financing. Blossom Capital led the round, and Haystack, Alt Capital, 1Flourish, and Bynd Venture Capital joined. A $3.5M Seed was reported in 2023, although Trebellar did not disclose a post-money valuation, ownership terms, individual investor checks, or a precise cumulative funding total in the new announcement.

Trebellar said the capital will expand its engineering and go-to-market teams, deepen its AI capabilities, and bring the platform to more enterprises. Its careers page currently lists technical hiring, including an AI/ML engineering role, which fits the stated use of proceeds.

The company also named Dave Radcliffe as an advisor. Trebellar says Radcliffe led the expansion of Google's real estate footprint across the tenures of Eric Schmidt, Larry Page, and Sundar Pichai. That appointment gives the startup operating context from one of the world's most complex corporate portfolios, but it does not substitute for customer adoption or independently measured product outcomes.

How Trebellar Works

Trebellar starts with a problem corporate real estate teams know too well: the information required for one portfolio decision rarely lives in one system. Lease terms may sit in documents, utilization in badge or sensor data, headcount in planning tools, costs in finance systems, and employee preferences in surveys. Analysts spend time reconciling the record before they can evaluate the choice.

The Trebellar platform maps those inputs into a governed ontology designed around corporate real estate. Functional agents can orchestrate plans, alerts, reports, and summaries. Domain agents bring context for lease administration, portfolio strategy, transactions, and analysis. Trebellar says each result remains traceable to the source data and assumptions behind it, allowing a human operator to inspect the work rather than accept an opaque recommendation.

That distinction matters because the product participates in decisions with long tails. A lease renewal, consolidation, relocation, or workplace-policy change can affect cost, employee access, recruitment, and operations for years. Speed is useful only when the evidence stays visible enough for finance, HR, workplace, and real estate leaders to challenge the same model.

The People Behind the Product

Diego Ferreiro Val, Trebellar's co-founder and CEO, previously led engineering work at Salesforce. David Garcia Quintas, co-founder and CTO, built AI and infrastructure systems at Google and Waymo. The pair began with a home-automation side project before a conversation with Salesforce's workplace technology leadership exposed a larger enterprise problem.

Trebellar's current leadership page also lists Adam Sewall as CRO/COO. The combination is relevant to the next stage of the business: Ferreiro Val and Garcia Quintas must keep the data and agent architecture trustworthy, while Sewall's go-to-market organization has to turn a technically persuasive product into a repeatable enterprise buying decision.

The company names Meta, Uber, Merck, and Cohesity as customers in the financing announcement. Blossom Capital also identifies Workday and Okta. These relationships are material signals that large organizations are willing to test the platform, but the company has not disclosed revenue, retention, a verified customer count, or audited savings tied to those deployments.

Why Corporate Real Estate Is an AI Market

Corporate real estate is a large expense managed through a mixture of software, services, documents, negotiation, and institutional memory. Hybrid work made the portfolio more dynamic at the same time that many planning tools remained organized around static occupancy assumptions. Every change in hiring, office policy, commute behavior, lease timing, or regional strategy can alter what the company needs from its space.

That creates room for category-specific AI. The system must understand leases, buildings, headcount, utilization, geography, and decision rights well enough to organize the evidence without pretending to own the executive judgment. Trebellar's agent design explicitly keeps people responsible for portfolio choices, landlord negotiations, business-unit trade-offs, and long-range strategy while software handles gathering, reconciliation, and drafting.

The investor thesis is therefore larger than adding chat to a dashboard. Blossom is backing an operating layer that could make corporate real estate data reusable across planning, lease events, location research, workplace operations, and executive reporting. If the ontology remains governed and integrations remain durable, each new workflow can begin from the same portfolio record instead of another manual reconstruction.

Competitive Pressure and Market Proof

Trebellar competes against more than other AI startups. The incumbent alternative includes enterprise real estate suites, analytics tools, consulting engagements, internal data teams, spreadsheets, and the informal expertise of operators who know which number to distrust. A new platform has to integrate with that environment before it can replace any meaningful part of it.

The company also faces a familiar vertical-AI challenge: a polished answer can appear before the organization trusts the underlying permissions, definitions, and audit trail. Trebellar's emphasis on governed data, source traceability, and human-held decisions addresses that concern in product design. Commercial proof will require enterprises to keep using the system when lease deadlines, disputed inputs, executive scrutiny, and cross-functional incentives collide.

No public evidence yet establishes Trebellar's revenue scale, retention, independently audited return on investment, or defensibility against larger enterprise platforms. The named customers, active product surface, investor syndicate, and current hiring show momentum. They do not close the question of how consistently the platform changes portfolio outcomes.

What the $18M Changes

The Series A gives Trebellar time to connect more source systems, deepen its domain agents, support larger deployments, and build the go-to-market motion required for enterprise software. The company is entering a phase where product architecture and commercial execution become inseparable. A trustworthy ontology without distribution remains a technical achievement; distribution without trustworthy data creates a faster route to a bad real estate decision.

For operators and investors, Trebellar is a useful test of where enterprise AI creates durable value. The model does not need to replace the executive responsible for the portfolio. It needs to preserve the evidence, shorten the manual assembly work, and keep the decision intelligible after the meeting ends. The $18M now follows Trebellar into those ordinary, expensive moments when a lease clock is moving and the organization can no longer afford to rebuild the same picture from scratch.