The Appraisal, April 2025 — The Top 50 AI Startups in Real Estate 🏆

I’m excited to release the inaugural Real Estate x AI List, highlighting the top startups building AI applications within the real estate sector. Congratulations to the companies selected this year! A high-resolution version of the graphic is available here. The complete list and methodology is included below.

In 1968, an unassuming device quietly reshaped our relationship with computers: the mouse. Introduced by Doug Engelbart during SRI’s landmark “Mother of All Demos,” the mouse was unveiled alongside a suite of innovations that would shape modern computing — including word processing, hypertext, shared screen collaboration, and teleconferencing. This small, handheld device served as a vital bridge between human intention and digital execution, making computers more intuitive and accessible. In the decades that followed, the personal computer — propelled in part by this humble tool — reshaped virtually every industry and profession.

Today, large language models (LLMs) are delivering the next transformative leap. Where the mouse lets us point and click, LLMs now enable us to converse, create, and collaborate with machines. It broadens the scale of problems software can address and the industries it can touch—from healthcare and law, to accounting and education. We believe real estate is poised for a similar transformation.

Why Real Estate?

Despite a decade-plus of venture capital flowing into real estate technology startups, significant inefficiencies persist across the asset class. Thomvest’s early investment thesis in real estate touched on this:

Much of the inefficiency associated with a real estate transaction can be traced back to coordination hurdles. Multiple stakeholders are required to get a transaction across the finish line, including agents, lawyers, title companies, inspectors, appraisers, and notaries (the list goes on). In total, intermediaries capture more than $80B of transaction value annually.

Of course, some of this is inherent to the sector, driven by structural barriers to software adoption, including:

• Real estate is a multi-player game — every real estate transaction or development involves dozens of stakeholders. This creates coordination friction and slows down tech adoption.

• Real estate is highly localized and heterogeneous — real estate is not one uniform industry but thousands of hyper-local ones, each with unique regulations, processes, and data formats. This lack of standardization creates challenges in scaling across markets.

• Real estate is cyclical — capital markets dictate the pace of activity in the sector. Today, high interest rates and rising labor costs have significantly cooled transaction volume; during these times, companies often postpone or reduce spending on technology.

The recent wave of AI advancements may finally create openings to tackle some of the sector’s most persistent inefficiencies. Unlike earlier software tools — which digitized only isolated tasks — modern AI systems can learn from messy, siloed data and even coordinate multi-party, multi-modal processes. This creates an opening for new use cases, new business models and new control points, sidestepping some of the adoption hurdles that stymied earlier startups.

The 2025 List — Early Winners & Themes

By our count, there are more than 750 startups building AI tools for real estate practitioners. This includes dozens of AI startups in every sub-category of real estate. Our goal in compiling the “Top 50” list is to present a snapshot of promising companies based on feedback from end-users and investors. Given the flurry of activity in this space, I expect the list to change meaningfully year to year, reflecting the period of innovation and iteration we’re in.

The pace of innovation both at the infrastructure layer and app layer of AI is astounding — it’s difficult to overstate the flood of smart technical talent and entrepreneurial energy into the category. ChatGPT launched publicly just over two years ago, acting as the starting gun for startups building AI apps. Today, foundational models are simultaneously more powerful and more affordable to deploy. As cost per unit of knowledge declines, new use cases and industries at the app layer are unlocked. My graphic above gets at this dynamic — AI systems are evolving to handle increasingly complex and creative use cases for end-users.

While we’re seeing entrepreneurs build AI startups across every corner of real estate, several key themes are emerging that help describe the nature of products being built:

Tackling Recurring, Lower Skill Work (The “Junior Analyst”)

AI streamlines labor-intensive administrative tasks—like scheduling, document drafting, and repetitive follow-up—allowing real estate teams to focus on higher-value work. I’d classify these companies as “sell the work” models that have become popular over the last year. This concept is particularly salient in real estate, where cyclicality leads to constant periods of understaffing and overstaffing. Interesting companies in the category include:

• EliseAI provides a virtual leasing assistant that automatically responds to renter inquiries, schedules property tours, and follows up with prospects​ — automating roughly 90% of a leasing team’s routine workflow​s.

• GreenLite is an AI-driven permitting platform that standardizes and automates the once-manual construction permit process across jurisdictions​. It can reduce the time and labor needed for plan review by up to 75% per project​, allowing developers to obtain permits faster while easing the burden on municipal building departments.

• Rexera develops specialized AI agents to handle real estate closing workflows, from gathering HOA documents and lien searches to sending routine updates​.

• In the mortgage space, Maxwell (a Thomvest portfolio company) has built an AI-enabled “mortgage fulfillment-as-a-service” offering. This means lenders can outsource and automate underwriting, processing, and post-closing tasks.

Design & Pre-Construction Automation (“Generative First Drafts”)

Generative AI tools let architects, developers, and even non-experts rapidly produce and refine floor plans, building models, and interior layouts. By eliminating guesswork and manual drafting, they not only speed up design cycles but also expand the range of viable options for each project (here’s a fun recent example shared on X). Interesting companies in the category include:

• Bobyard interprets landscaping and site plans, producing instant material takeoffs and cost estimates that replace hours of manual counting. This automation saves contractors time, improves bidding accuracy and speeds up client proposals.

• TestFit applies generative design to produce optimized building layouts in seconds, accounting for zoning, density, and financial constraints. This turns multi-week feasibility studies into a more instant and iterative review process for real estate developers.

• Cedar unifies public datasets with AI-powered modeling to generate multiple early-stage building configurations for any parcel, giving developers quick, data-driven insights into project feasibility and potential returns.

Data Capture & Analysis (“Instant Understanding”)

We are seeing broad adoption of tools designed to better capture and process both visual data (e.g., property photos, construction drawings) and textual data (e.g., sales contracts, loan agreements). Although it's still early days, the major AI labs have recently released impressive new models for OCR (Mistral) and image recognition (Gemini 2.5) that significantly advance the state of the art. I view this category as an important enabler to digitizing manual workflows in real estate. Interesting companies in the category include:

• NavigateAI’s platform uses AI to automatically aggregate and analyze disparate real estate data, eliminating error-prone manual data gathering​. By delivering instant, unified analytics, it enables professionals to make faster, insight-driven decisions in everything from market research to portfolio strategy.

• Hover employs AI-powered computer vision to turn ordinary smartphone photos of a property into a precise 3D model, collapsing a multi-hour site inspection into a 15-minute digital workflow​. This ensures contractors and insurers get consistent, accurate measurements and documentation with a fraction of the usual effort.

• ZestyAI applies machine learning to high-resolution aerial imagery, automatically identifying property features and risk factors for over 150 million buildings​. By replacing manual inspections with instant AI-driven analytics, it helps insurers and investors make quicker, data-backed decisions on property valuations and risk management.

• DocumentCrunch uses AI to automatically analyze construction contracts and identify critical risk clauses and obligations. This streamlines contract review workflows, reducing manual review time and helping project teams ensure compliance.

Unblocking Bottlenecks (“Messy Inbox Problem”)

Real estate transactions often encounter slowdowns at specific choke points — whether it’s waiting on a mortgage underwriter, a legal review, or an appraisal. Much of the industry still runs on spreadsheets, emails, and repetitive workflows. AI is now being deployed to attack these bottlenecks and speed up the slowest parts of the process. These are often “agentic AI” models that are multi-modal and work across multiple systems. Interesting companies in the category include:

• Keyway (another Thomvest portfolio company) streamlines complex commercial real estate transactions with an AI-driven platform that automates labor-intensive tasks — from extracting critical deal terms out of dense documents​ to instantly generating rental comps and pricing recommendations​.

• Valuebase can instantly appraise properties; they’ve built workflows for tax assessors offices to streamline the valuation process for local jurisdictions — a process that has historically been manual, highly subjective, and error-prone.

• TurboHome is a fixed-fee real estate brokerage. Their platform automates transaction workflows to help agents close faster, reducing friction and overhead to support a low-cost, high-volume brokerage model.

Looking Ahead — Obstacles & Opportunities

We’re so early into the AI era — the contours of how machine intelligence will weave into every physical and financial workflow of real estate are only just beginning to come into focus. Yet turning that vision into tangible value across real estate will mean grappling with a host of challenges:

• The tools and use cases in real estate are nascent, leading to an “enthusiasm vs. implementation” gap across most firms. This is to be expected — every major technical shift experiences a productivity paradox where where measurable productivity gains lag early adoption and excitement (for example, it took nearly fifteen years from the mouse’s unveiling in 1968 before it evolved into the everyday staple of personal computing). To help overcome this implementation gap, AI startups must prioritize tangible, immediate value creation for their customers — showcasing quick wins that demystify AI.

• Another important question for startups in the category is how to out-compete incumbents with a distribution advantage and a mandate to integrate AI into their existing products. Procore, Rocket, Appfolio have leaned in to the AI narrative over the last year, as have many other public companies in our real estate index.

• And finally, it’s important to consider the long-term value of verticalization in AI — by that, I mean the role sector-specific AI tools will play vs. general purpose AI (e.g. ChatGPT). There have been an influx of “horizontal” AI startups over the last year, for example Writer and Hebbia are making inroads selling into enterprises across multiple sectors. And last month, Chinese AI startup Monica launched Manus, “the first general AI agent,” capable of doing a wide range of tasks using a computer like a human would. In my view, vertical AI startups will build the models, integrations and workflows to win-out over general-purpose competitors, particularly in sectors with deep mid-market and SMB segments (like real estate).

In addition to the themes highlighted above, I’m tracking a few areas this year where I expect an influx of startup activity:

• Physical AI: A common refrain when discussing the proliferation of AI within the sector is the physical nature of real estate that limits startups’ potential impact. That’s certainly a valid point, which speaks to why we’ve seen more immediate impacts of AI on purely white-collar industries like law, accounting and software development. We are in the early days of startups innovating at the intersection of artificial intelligence and the physical world, enabling robots and machines to perceive, understand, and interact with their surroundings autonomously. This is critical to breaking the five-plus decade streak of declining productivity in the construction industry. There are a handful of robotics startups targeting specific trades, like painting, landscaping and drywall. Expect robots to be deployed more broadly over the next decade, particularly in factory settings (e.g. modular construction) where physical conditions can be better controlled.

• Better Generative Design Tools: The state of the art in generative design is rapidly improving — I expect an evolution from “first draft” tools to higher-fidelity outputs that can be relied on for more accurate renderings, schematic designs, interior staging, and even construction drawings. This has the potential to meaningfully decrease soft costs in a construction project.

• Better Agentic AI Tools: I highlighted agentic AI earlier, but it’s worth emphasizing the opportunity here, particularly as these tools embed more deeply into existing systems (e.g., property management systems, loan origination systems). Much of the industry continues to run on spreadsheets, email, and repetitive workflows. I’m especially intrigued by agentic AI’s potential to address challenges point solutions could not—automating processes end-to-end, replacing human “middleware,” and seamlessly navigating across fragmented data systems and platforms (here is a helpful demo from Kairos to see this action). Recent releases of Model Context Protocol, Computer Use and Operator are just now getting implemented by startups. I expect additional improvements to agentic AI systems in the near future, including longer context windows, model interoperability, and better data security.

Thanks for reading! As always, please feel free to share feedback or thoughts on this newsletter — you can respond to this email directly or shoot me a note at nima@thomvest.com.

---

Complete 2025 List

Methodology: Thomvest selected companies for inclusion on this list based on proprietary criteria, taking into consideration nominations from dozens of venture capital firms investing in the real estate technology vertical, real estate operators and industry experts. Certain companies selected (as indicated above) are Thomvest portfolio companies.

Note: Charts and graphs provided within are for informational purposes solely and should not be relied upon when making any investment decision. The content speaks only as of the date indicated. Any projections, estimates, forecasts, targets, prospects, and/or opinions expressed in these materials are subject to change without notice and may differ or be contrary to opinions expressed by others.