Construction · Deep dive

Buildots

AI construction-progress tracking — 360° cameras on hardhats capture the whole site, computer vision aligns every element to the BIM model and schedule, and the platform forecasts delays weeks before they land.

emerging

The question that decides it: Buildots sells the heaviest version of jobsite reality capture — full-site 360° scans aligned element-by-element to a BIM model and schedule, which is what makes its delay forecasts possible but also restricts it to large, BIM-mature projects at custom per-project prices. Does element-level, forecast-grade tracking command a durable premium once OpenSpace ($199M raised, Disperse acquired November 2025) sells 'good enough' progress tracking down-market and Procore/Autodesk bundle AI into the systems that already host the schedule — or does the BIM prerequisite cap Buildots at the megaproject niche its ~$300M valuation already prices in?

Snapshot

Buildots straps 360° cameras to hardhats, has someone walk the site weekly, and lets computer vision do what superintendents do imperfectly by eye: compare what is actually built — element by element — against the BIM model and the schedule. The output is objective progress data, pace analysis and delay forecasts weeks ahead of impact; the company claims deployments cut project delays by up to 50%, roughly two to three months on an average large project (company materials, 2025-26). Founded in Tel Aviv in August 2018 by three graduates of Israel’s elite Talpiot program with zero construction background, it has raised $166M through a $45M Series D led by Qumra Capital (May 2025, ~$300M estimated valuation) and counts Turner, JE Dunn, Mortenson, VINCI, HOCHTIEF, Sir Robert McAlpine, Intel and Digital Realty among customers. It is the deepest — and heaviest — implementation of jobsite reality capture, and the question is whether depth wins against cheaper, lighter rivals.

Founding story

The founding is unusual in that the industry came last. Roy Danon, Aviv Leibovici and Yakir Sudry met as cadets in Talpiot, the IDF’s hyper-selective academic-military program, and stayed close through roughly a decade in technology units — taking an annual ski trip on which, by their own telling, they perennially discussed the company they would someday start. In late 2017 they finally committed. They had no thesis, no product, and no industry; they went looking for the largest, least-digitized sector they could find and landed on construction — a business none of the three had ever worked in.

They then did the thing most outsider-founders skip: six months embedded on construction projects, walking sites and interviewing contractors before building anything. What they observed was that nobody actually knows the true status of a large project — progress reporting is manual, sampled, political, and weeks stale — and that this ignorance, not bad planning, is why projects slide. Buildots was incorporated in August 2018 to make site status a measured fact rather than an argument. The outsider origin cuts both ways: a first-principles product no contractor would have designed, but founder credibility with a 60-year-old superintendent that is borrowed from data, not scar tissue.

How it works

The physical loop is simple. A site engineer (or a Buildots-certified capture technician, sold as a service) clips a 360° camera to a hardhat and walks the project on a fixed cadence — typically weekly, scheduled in advance to align with the plan. No dedicated hardware stations or robots; the walk piggybacks on rounds someone was doing anyway, and since July 2026 superstructure work can also be captured by drone.

The hard part is what happens next. The footage has no GPS indoors, so Buildots first localizes every frame inside the building geometry. Its vision models then classify the installation state of thousands of individual elements — a partition framed, boarded one side, boarded both, taped, painted; sockets in; sprinkler heads fitted — and write that status onto the corresponding objects in the BIM model. Comparing as-built state to the schedule yields progress that is measured rather than reported, pace per trade and per zone, and a delay forecast that flags activities trending late weeks before the master schedule shows it, with root-cause analysis pointing at the trade, location or sequence responsible. The company says its threshold for shipping a detection is around 99% accuracy, with anything below routed through human verification (“integrated tracking”) — a materially different posture from photo-documentation tools that show you imagery and let you judge.

The dependency to notice: all of it requires a current BIM model and a maintained schedule to align against. No model, no dots. That prerequisite defines who can buy.

Product and business overview

What began as a progress dashboard now sells as a platform Buildots repositioned in April 2026 as “construction intelligence”:

• Progress tracking — the core: element-level completion status against BIM and schedule, refreshed each capture.
• Delay forecast and pace analysis — predictive metrics on where the plan breaks, weeks ahead, with root-cause drill-down.
• Plan Tracker — weekly work planning checked against what the cameras actually saw, closing the loop on short-interval planning.
• Explorer — visual navigation of the captured site, the closest thing to the OpenSpace-style documentation use case.
• Superstructure tracking (July 1, 2026) — extends analysis from interior fit-out into the structural phase, via drone or 360° capture, after a year-plus beta.
• Buildots Field (announced July 15, 2026; GA slated Q4 2026) — workforce, safety and logistics management, the commercialization of Genda, acquired October 2025, pairing “what got built” with “who was on site.”
• Intelligence Lab (June 25, 2026) — cross-project benchmarking research arm, leveraging its captured dataset.
• Capture services — certified technicians who do the walks for you, removing the adoption burden at a price.

Buyers are large general contractors and, increasingly, owners running repeat capital programs — Intel uses it across its global fab construction program (Intel Capital, 2024), Digital Realty on data centers (company, 2026).

Business model and pricing

Revenue is SaaS, priced per project by quote — scope, square meterage and duration drive the number; hardware and AI processing are bundled. No price list exists publicly; Capterra, Software Finder and similar catalogs (accessed 2026) all show custom pricing only, and reviewers grumble about exactly that opacity. The commercial shape is land-per-project, expand-per-portfolio: Sir Robert McAlpine went from one job to five projects covering 260,000+ sqm over three years (company case study, 2025); Intel converted from customer to strategic investor (July 2024).

What can be triangulated: Calcalist reported (May 2025) annual revenue in the “tens of millions” of dollars with triple-digit growth over the prior year. A Latka page pegs 2023 ARR at $38.8M, which sits awkwardly against the Calcalist framing two years later and should be treated as unreliable. Nothing public on gross margin — worth pausing on, since human capture services and human verification of sub-threshold detections both sit in cost of revenue — or on net revenue retention. At a ~$300M estimated valuation (Calcalist, May 2025) on tens of millions of revenue, the Series D was priced as a solid but not frothy multiple; the flat-ish step-up from $121M raised implies earlier rounds absorbed most of the optimism.

Traction over time

The geographic arc is the story: Israel and the UK first (McAlpine, Wates, Kier, Multiplex), Europe next (VINCI, NCC, HOCHTIEF, Bouygues), and North America — the largest market — only becoming the stated growth engine in 2025, with the company claiming it would quadruple its NA presence that year. That claim is the company’s own and has not been externally verified. The named-logo quality is high; the count of active projects, unlike OpenSpace’s boasted site totals, is not published — a telling asymmetry for a per-project business.

Market analysis

Construction is a ~$13T global output industry with famously flat productivity; McKinsey has long estimated $1.6T of annual value available from digitization. The directly addressable slice — AI in construction — was roughly $4.86B in 2025, projected to ~$35.5B by 2034 at a ~24.8% CAGR (Fortune Business Insights, 2025), with North America the largest region. Structural drivers favor Buildots specifically: data centers and fabs are the fastest-growing construction categories (schedule-obsessed owners like Intel and Digital Realty are its ideal buyers), BIM mandates keep spreading across public procurement in the UK and EU, and skilled-superintendent scarcity raises the value of automated oversight. The counterweight: the majority of world construction is not BIM-modeled to the standard Buildots needs, which shrinks its true serviceable market well below the headline TAM — and its cheapest competitors serve exactly the projects it cannot.

Competitive intel

The field splits by depth of interpretation. OpenSpace ($199M raised; Tracxn 2026) owns the lightweight end — passive 360° photo documentation across a far larger number of jobsites — and moved directly onto Buildots’ turf by acquiring Disperse, the London-based progress-tracking specialist, in November 2025 (after a June 2025 Track partnership) and shipping OpenSpace Track (capture plus human-verified progress). That acquisition removed Buildots’ closest methodological rival and armed its biggest-distribution competitor with an interpretation layer. Doxel ($56.5M; a16z, Insight) pursues nearly the identical BIM-aligned thesis in the US with lidar-heavy capture, though it has been publicly quiet since its 2021 Series B. DroneDeploy (bought StructionSite, Nov 2022) bundles aerial plus interior capture into a large installed base. Track3D (~$14.3M) undercuts from below. And Procore and Autodesk host the schedules and models everyone syncs against; both are layering AI onto that gravity. Buildots’ edge is genuine: element-level accuracy, forecasting rather than documentation, and owner-grade credibility on megaprojects. Its exposure: everyone else is cheaper, lighter, or already installed.

History and evolution

• Aug 2018 — Founded in Tel Aviv by Danon, Leibovici, Sudry after six months of on-site research; early pilots in Israel (Tidhar) and the UK.
• Jul 30, 2020 — Discloses $16M (seed + Series A, TLV Partners); ~80 staff; London office established, UK becomes the beachhead market.
• Aug 4, 2021 — $30M Series B led by Lightspeed; TechCrunch frames it “eyes on construction sites.”
• May 17, 2022 — $60M Series C led by Viola Growth and O.G. Tech; total ~$106M; European expansion (VINCI, HOCHTIEF, NCC).
• Jul 11, 2024 — $15M strategic round led by Intel Capital; Intel, already deploying Buildots on its fab program, becomes an investor; total $121M.
• May 29, 2025 — $45M Series D led by Qumra Capital at a ~$300M estimated valuation; total $166M; North America declared the priority.
• Nov 2025 — OpenSpace acquires Disperse (following a June 2025 partnership powering OpenSpace Track) — the competitive landscape’s most consequential event, not Buildots’ own.
• Oct 2025 — Acquires Genda (workforce/safety management; clients incl. DPR, Clark, Hensel Phelps); first M&A.
• Apr 15, 2026 — Repositions the platform as “construction intelligence”; June-July 2026 brings Intelligence Lab, superstructure tracking, and Buildots Field in rapid succession.

No layoffs or down rounds are publicly recorded; the visible stumble is subtler — seven years in, the company still does not publish project counts or revenue, and needed until 2026 to cover the structural phase of a building.

What people say

The case for. Customer references are the strongest in the category’s deep end. Sir Robert McAlpine has used it for three-plus years across five projects and 260,000+ sqm, crediting early delay identification and objective progress reporting (SRM/company, 2025); a McAlpine-VINCI JV on the Royal Bournemouth Hospital documented improved collaboration from shared ground truth. Software directories (Software Finder, Capterra ecosystem, 2026) echo the same praise: visibility no manual process matches, delays caught early, an interface built for construction people. Intel putting money in (July 2024) after using the product is the most credible endorsement available. Glassdoor (56 reviews, 2026) runs 4.0/5.0 with 73% recommending — transparency and monthly all-hands get specific praise.

The complaints. Directory reviewers flag long initial setup, the hard requirement for BIM models and maintained schedules, opaque quote-only pricing, limited dashboard customization, and suitability only for large complex projects — no trial, no self-serve. There is no meaningful G2 review base, so field-level sentiment (do superintendents resent the extra walk? do trades game the camera?) is largely invisible — itself a gap. The generic-but-real objection on sites is surveillance: cameras on hardhats intersect with worker-privacy law and union agreements, and academic surveys of construction AI list fear of monitoring among the top adoption barriers (GAO/Berkeley Labor Center, 2024). A minority of Glassdoor reviews (2025-26) describe declining management quality, unclear expectations, favoritism and low morale on some teams — hyper-growth strain, not crisis, but recurring. And the structural critique: the company’s headline “up to 50% fewer delays” is a company-published best case, not an audited average.

Outlook: the open question

Buildots works if element-level, forecast-grade progress intelligence proves valuable enough that large contractors and repeat owners pay a heavy premium over lightweight capture — and if the BIM-mature megaproject segment (data centers, fabs, hospitals, infrastructure) is big and sticky enough to build a durable company on. It stalls if OpenSpace-plus-Disperse delivers 80% of the insight at a fraction of the setup burden, or if the BIM prerequisite permanently walls Buildots off from the majority of construction.

The bull case: the deepest interpretation in the market, seven years of construction-specific training data, blue-chip references on three continents, an owner-side wedge (Intel, Digital Realty) rivals lack, and revenue in the tens of millions growing triple-digit on $166M raised with no down round (Calcalist, May 2025). Consolidation also favors depth: documentation is commoditizing; forecasting is where budget defensibly sits. The bear case: the serviceable market is a fraction of the TAM headline; its best-funded rival just bought the closest thing to its brain and has far more jobsites to sell it into; Procore and Autodesk can bundle from above; and the human verification and capture services behind 99%-grade accuracy may cap margins in ways no public number reveals. Watch three things: whether Buildots starts publishing project counts and NRR (confidence signal), whether OpenSpace Track lands with enterprise GCs on BIM projects (commoditization signal), and whether the next round is led by a growth investor at a real step-up above $300M or looks like another insider extension (verdict signal). The company has built the best answer to “what is actually happening on this site.” The open question is how many sites the answer is worth paying this much for.

Sources and further reading

• Buildots raises $45M at $300M valuation to modernize global construction with AI (Calcalist/CTech, May 2025)
• Buildots raises $45M to help companies track construction progress (TechCrunch, May 2025)
• How three IDF veterans are fixing construction’s biggest problem (Calcalist/CTech)
• Buildots raises $16M to bring computer vision to construction management (TechCrunch, July 2020)
• Buildots raises $30M to put eyes on construction sites (TechCrunch, Aug 2021)
• Intel Capital leads $15M round in AI construction tech startup Buildots (SiliconANGLE, July 2024)
• Buildots Buy Expands Its Platform (Genda acquisition) (Engineering News-Record, Oct 2025)
• OpenSpace Acquires Disperse (OpenSpace, November 2025)
• Case study: How Sir Robert McAlpine tracked 26,000 sqm with AI (Buildots, 2025)
• Buildots launches superstructure tracking (PR Newswire, July 2026)
• AI in Construction Market Size & Industry Report (Fortune Business Insights, 2025)
• Buildots Reviews (Glassdoor, accessed 2026)