Y Combinator's 2026 real estate and construction cohort bets big on AI agents and construction intelligence

Y Combinator's real estate and construction portfolio has reached 126 funded companies, according to the accelerator's June 2026 directory, with the most recent cohorts reflecting a pronounced shift toward AI-native tools that embed directly into industry workflows rather than simply digitizing existing ones.

The range of problems being attacked — from blueprint analysis and concrete takeoffs to landlord banking and brokerage paperwork — points to an industry where manual, fragmented processes remain widespread despite decades of software investment. Across the cohort, the dominant architectural pattern is the AI agent: software that does not merely surface information but acts on it, chasing tickets, filling contracts, and generating estimates without human handoffs.

Construction estimation draws concentrated attention

Several of the newest entrants focus specifically on the estimation and pre-construction phase, where errors are costly and speed directly translates to revenue. PLAN0 AI, a P2026 batch company, describes itself as building the Bloomberg of construction, analyzing architectural plans using vision models to produce cost estimates and analytics in minutes. The company reports that $20 billion in projects are already running through its platform, and it aggregates historical and real-time project data across major geographies to power machine-learning-driven pricing predictions up to 24 months out, according to Y Combinator's directory.

Rudus, also from the P2026 batch, takes a narrower vertical approach, targeting concrete contractors specifically. Its platform automatically identifies footings, walls, columns, and slabs on plans, with the company claiming estimation time drops by 70% and that contractors can pursue three times more projects annually as a result. The concrete focus is deliberate — Rudus notes that concrete is the most-used material on earth after water, with demand accelerating alongside AI data center construction.

Foreman, from the W2026 batch, attacks the broader contractor workflow. The platform covers the full project lifecycle from first estimate to final invoice, replacing what it describes as a patchwork of spreadsheets, emails, and copied Word documents. Contractors upload plans, photos, or a project description, and the system generates structured takeoffs, scoped estimates, and professional proposals.

Drawing quality control emerges as a distinct category

Two F2025 companies — Helonic and Structured AI — have staked out construction drawing review as a standalone problem worth solving with AI. Helonic analyzes PDF plans, detects clashes and inconsistencies across disciplines, and generates draft RFIs so teams can surface issues before construction begins, integrating with Procore and Autodesk. Structured AI, operating with a team of 10 from New York, builds AI agents that perform quality control on technical documents, applying building codes and project-specific standards to mechanical, electrical, and structural drawings.

Both companies frame their value proposition around cost avoidance rather than productivity — catching a design clash in a drawing review is orders of magnitude cheaper than addressing it in the field. Structured AI's stated mission is to free engineers to focus on creative design rather than repetitive administrative review.

Property-side ventures target transaction friction and landlord finance

On the real estate operations side, RealPact (S2026) is building AI agents that handle paperwork for real estate transactions. The system retrieves property records — deeds, tax records, permits, parcel data, and MLS information — then uses that data to populate contracts, organize documents, and track deadlines. The company is currently selling to brokerages in New Hampshire and describes its expansion as rapid.

CentralComs (P2026) applies a similar agent model to property management, tracking maintenance calls, chasing open tickets, and keeping owners, tenants, and vendors updated. The platform integrates directly with AppFolio, Buildium, and Yardi, operating inside existing property management stacks rather than replacing them. Brickwise, a London-based F2025 company, offers comparable AI property management functionality and reports securing a contract exceeding $1 million in committed annual recurring revenue within six months of formation, alongside more than $3 million raised in total funding, per its YC profile.

Goldbridge (F2025) frames its opportunity around a structural financial inefficiency in real estate ownership. The company cites more than $1 trillion in annual rent flowing through landlord bank accounts, with roughly a quarter sitting in idle reserves and security deposits.

With $2.5T in real estate loans about to mature in 2027/28, property owners are desperate to boost their income ASAP. — Goldbridge, via Y Combinator company profile

Goldbridge is positioning its AI-powered banking platform as the financial operating system for real estate owners, targeting both the idle-cash problem and broader expense management. The founding team includes a former White House advisor and a 100-unit owner-operator.

Data infrastructure underpins multiple business models

Several companies in the cohort compete on the quality and breadth of their proprietary data rather than on AI model architecture alone. Travo (W2026) is building a real estate data set spanning rental comps, ownership records, zoning, and financials for any property or parcel, then uses that foundation to serve private equity firms, developers, and brokers. PLAN0 AI similarly differentiates through large-scale industry partnerships that feed its cost prediction models with historical and real-time project data.

Automax.ai (F2025) takes a hardware-assisted approach to data capture in property appraisal, using LiDAR and computer vision on a mobile device to collect property details automatically. The company states its complete appraisal reports are Fannie Mae and Freddie Mac compliant and can be produced in under 20 minutes.

The cohort's breadth — from a publicly traded equipment rental platform in EquipmentShare (W2015, now 5,400 employees) to two-person pre-seed teams — illustrates the full lifecycle of construction and real estate technology investment that YC has supported. The current concentration of AI agent companies in the 2025–2026 batches suggests the accelerator sees the sector's administrative overhead as one of the more tractable near-term targets for applied AI.