Construction software · Deep dive

Rebar

AI takeoff and quoting software for commercial HVAC manufacturer reps and distributors — computer vision that reads a plan set, counts the equipment and drafts the quote in minutes — fresh off a $14M Series A led by Prudence.

emerging

The question that decides it: Rebar's wedge is a computer-vision takeoff engine trained on HVAC plan sets, sold to the estimating desks of manufacturer reps and distributors whose quote volume is capped by estimator hours. Within 90 days of its Series A, generalist AI-takeoff rivals raised two to three times its capital (Bobyard $35M, Dec 2025; Attentive.ai $30.5M, Nov 2025), and multimodal foundation models keep cutting the cost of reading a blueprint. Can Rebar convert its head start — 40 rep/distributor customers with seven invested, an AHR Innovation Award, and a basis-of-design data layer no generalist has — into the quoting system of record for HVAC distribution before plan-reading accuracy stops being a differentiator?

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

Rebar sells AI takeoff and quoting software to the sales side of commercial HVAC: the manufacturer rep firms and equipment distributors whose inside staff turn construction plan sets into equipment quotes. Its computer-vision models — trained, the company says, on millions of HVAC blueprints — find the mechanical pages, mark and count every piece of equipment with its properties, and assemble a bill of materials, making quotes 60-70% faster (company, Mar 2026). Founded in New York in October 2024 by a former Johnson Barrow estimator and an ex-Goldman product builder, it won the 2026 AHR Expo Innovation Award for software, reached 40 customers and thousands of projects a month within a year of commercialization, and closed a $14M Series A led by Prudence on March 10, 2026. It matters now because estimator hours, not demand, cap how many jobs a supplier can bid — and because better-funded generalists are converging on the same blueprint-reading problem.

Founding story

Evan Brown is the rare founder who did the job his software replaces. He grew up in his uncle’s commercial HVAC distributorship watching estimators spend full weeks on quotes, studied mechanical engineering, then spent more than five years as an estimator and sales engineer at Johnson Barrow/DMG Corp — a major West Coast HVAC manufacturer rep — building inventory lists off printed blueprints with highlighters and rulers, then PDF editors, across work touching 100+ supplier offices (Crunchbase News, Mar 2026; company site). When ICG-backed roll-up Ambient Enterprises acquired Johnson Barrow and DMG around the turn of 2024 (Baird/MergerLinks, Jan 2024), Brown got a mandate to explore AI in day-to-day operations and led research into visual search for automated takeoffs. The conclusion — that the tooling should exist and didn’t — became the company: he left and founded Rebar in October 2024 with Andrew Schwartz, a Goldman Sachs product developer with earlier stints at MycoWorks, Lasso Capital and PwC and a co-founded researcher social network behind him. The pair say they interviewed thousands of operators before settling on takeoff as the wedge (company site, 2026). It is textbook vertical AI pairing — one founder who has done the manual workflow at scale, one who has shipped regulated production software — though notably neither is a deep-learning researcher, a gap the competition does not share.

How it works

Start with what a quote request actually is. A general contractor or mechanical contractor bidding a commercial project sends the local HVAC rep or distributor an invitation to bid: a plan set that can run hundreds of pages plus a spec book. Somewhere in there are the mechanical sheets — equipment schedules and drawings specifying rooftop units, air handlers, VAV boxes, fans, diffusers, each with properties like tonnage and CFM. A human estimator finds those pages, counts and tags every unit, checks the spec book for the engineer’s basis of design (which manufacturers are approved), maps the equipment to the lines their firm represents, prices it, and assembles a proposal. Done by hand this takes hours to days per bid — up to a week for large ones — and firms win only 5-10% of the proposals they submit (Brown to Crunchbase News, Mar 2026), so estimator hours directly cap revenue.

Rebar automates the middle of that pipeline. A user uploads the plan set; the computer-vision models identify the mechanical pages, then mark equipment on the drawings with properties attached — “AI Equipment Marking” — supported by automatic room search and configurable markup symbols, colors and equipment preferences so output matches each firm’s house style. When the architect issues an addendum, an AI change summary diffs the revised set against the original, a genuinely painful manual chore. The result is a structured bill of materials the estimator reviews and turns into a quote in minutes. The platform extracts “tens of thousands of data points” per project from plans and spec books (BusinessWire, Mar 2026), feeding an analytics layer: what engineers in a market are specifying, which product lines local jobs contain, how the estimating team performs. Positionally, Rebar sits upstream of the distributor’s ERP and the manufacturers’ configurators: it does not yet price or book the order; it manufactures the takeoff those systems consume. Per-customer data isolation and SOC 2 Type I are advertised — table stakes for firms whose bid pipelines are competitively sensitive.

Product and business overview

The shipped product is AI Takeoff — equipment marking, room search, addendum summaries — wrapped in Customizations (symbols, colors, equipment and property preferences per firm) and a Analytics suite (Basis of Design, Estimating Performance, Key Lines) that turns quoting exhaust into market intelligence. A Security layer (SOC 2 Type I, NDA-project handling, per-org data isolation) rounds out the platform. The stated roadmap widens the aperture: spec review, submittals and bid management next; then “additional agents” for plumbing and electrical suppliers; then geographic expansion in the US and internationally (BusinessWire, Mar 2026). The framing throughout is “AI operating system for the trades” — takeoff is the wedge, not the product. Customers are commercial HVAC manufacturer reps and distributors first, with contractors named in the press release as well; seven of its 40 customers liked the product enough to invest in the company (Crunchbase News, Mar 2026), which doubles as a channel strategy in a clubby industry where rep firms all know each other through HARDI and AHR.

Business model and pricing

Rebar sells a usage-based subscription (Brown to Crunchbase News, Mar 2026) — no public price list, no self-serve tier; the website’s pricing section is a demo-booking button. That is deliberate: pricing indexed to project volume rather than seats matches an estimating desk where headcount is small but throughput is the product, and it lets Rebar participate in the upside when a customer’s bid volume rises. The ROI arithmetic it sells against: proposals that took up to a week now take minutes; firms bid more jobs at the same headcount; and early customer conversations suggest win rates may be rising 2-3x, though Brown himself flags that construction contracts take years to finalize, so that number is anecdotal (Crunchbase News, Mar 2026). Prudence’s Viniar claims task-time compression over 90% on some workflows (Mar 2026). What is not public: ACV, gross margin (inference on hundred-page plan sets is not free), contract length, or any absolute revenue figure. Every economic claim in the record is company- or investor-supplied.

Traction over time

The slope is steep but the y-axis is hidden: doubling ARR in six weeks off an undisclosed, certainly small base is a seed-stage growth fact, not a scale fact. The more convincing signals are structural — thousands of projects monthly implies real production usage rather than pilots, and customers writing checks into the cap table is the strongest available proxy for satisfaction at this stage.

Market analysis

The end market is enormous relative to the software spend it currently supports. North American HVAC distribution was an estimated $65.9B market in 2024, projected to grow at ~6.2% annually through 2030 (Grand View Research); HARDI counts 570+ distributor members across roughly 5,000 branches, and contractors make ~70% of purchases through wholesale distribution (HARDI Voice of Contractor survey). Layer on the manufacturer rep firms that quote engineered equipment and the mechanical contractors who buy from both, and Rebar’s stated arena — HVAC, then electrical and plumbing — covers a meaningful slice of the roughly $150B+ US HVAC economy (2025 industry reporting). The structural tailwinds are real: an aging estimator workforce with no replacement pipeline, a data-center and electrification construction boom pushing mechanical scope, and near-zero purpose-built software history among suppliers — Viniar’s “leapfrogging software” thesis. The honest caveats: the software TAM is a small fraction of the goods TAM — estimating desks run 2-10 people even at large firms, which is why Rebar prices on usage and needs the workflow expansion to work — and the cycle is soft: HARDI distributor sales grew just 2.8% in 2025, mostly on price, and fell 5.9% year-over-year in January 2026 (Distribution Strategy Group, Mar 2026). A volume recession kills discretionary distributor software spend — though a tool sold as “bid more with the same staff” can cut the other way.

Competitive intel

See the competitor table. The category is crowded and freshly capitalized. Bobyard ($35M Series A, 8VC, Dec 2025) and Beam AI/Attentive.ai ($30.5M Series B, Insight, Nov 2025; $48M total) are generalist takeoff platforms with two to three times Rebar’s funding, attacking from breadth. Togal.AI (~$17M) proved the category early but skews to GCs and commoditized scopes. Against all three, Rebar’s argument is the same: HVAC equipment takeoff is schedule- and spec-driven, not area-measurement-driven, and a model trained solely on mechanical plans plus a rep-firm workflow (basis-of-design analytics, line-card mapping) beats a horizontal tool retrofitted to the trade. The legacy mechanical estimators — Trimble AutoBid, FastEST — own contractor install-bases but automate little and ignore the distributor desk. The AHR finalists AGILE-MEP and AlphaTakeoff show the HVAC-native lane is contested too; the incumbent that wins most deals today is Bluebeam plus habit. Adjacent threat: contractor platforms (ServiceTitan, BuildOps) consolidating commercial HVAC from the service side could reach into equipment quoting. Rebar’s edge is focus and founder-market fit; its weakness is that focus caps the data advantage at one trade while rivals’ capital compounds across many.

History and evolution

• Dec 2023 - Jan 2024 — ICG-backed Ambient Enterprises acquires Johnson Barrow and DMG Corp; Brown, an estimator/sales engineer there, is put on AI implementation and researches visual search for takeoffs.
• Oct 2024 — Brown and Schwartz found Rebar in New York; undisclosed early capital follows from seed investors including Founder Collective, Villain Capital and Optimist Ventures (participation confirmed at Series A; terms never published).
• ~Mid 2025 — Commercial launch; first enterprise rep/distributor customers; SOC 2 Type I obtained.
• Nov 2025 — Named 2026 AHR Expo Innovation Award winner in Software — the HVAC industry’s main trade-show prize — over finalists AGILE-MEP and AlphaTakeoff (ACHR News).
• Nov-Dec 2025 — The competitive board resets: Attentive.ai raises $30.5M (Insight Partners), Bobyard raises $35M (8VC).
• Feb 2026 — Exhibits at AHR Expo 2026; ARR doubles in the first six weeks of the year.
• Mar 10, 2026 — $14M Series A led by Prudence closes; expansion announced into spec review, submittals, bid management, plus electrical and plumbing agents.

No pivots, no leadership changes, no public stumbles yet — which mostly reflects how short the record is.

What people say

The case for. There are no G2, Capterra or Glassdoor reviews yet — unsurprising at 16 months old — so the honest proxies are industry juries, investors and revealed customer behavior, labeled as such. The AHR Expo Innovation Award (Nov 2025) is peer recognition from the HVAC trade itself, and Brown’s pitch — multi-day takeoffs to minutes, with higher accuracy — survived judges who do this work. The strongest customer signal is financial: seven of 40 customers invested in the company (Crunchbase News, Mar 2026); supplier executives do not put money into vendor software they merely tolerate. Prudence, a construction-vertical specialist (Trunk Tools, Clearstory), led the round after watching enterprise accounts run thousands of projects monthly. Hiring posture — a $5K referral bounty for a customer-success lead, engineering roles on Ashby, meals and late-night rides at a NYC office — reads like a team straining against demand rather than searching for it.

The complaints. Direct criticism is essentially absent from the public record, which is itself the caveat: every number — 60-70% faster, 2-3x win rates, doubled ARR — is company-supplied, unaudited, and off undisclosed bases, and Brown concedes the win-rate lift is early anecdote. Category-level skepticism is easy to find: takeoff vendors have overpromised since Togal, accuracy on messy real-world plan sets is every tool’s recurring weak point (takeoff-tool comparisons, 2025-26), and Bricks & Bytes has noted YC keeps minting near-identical AI takeoff startups — evidence the moat, if any, is workflow and data, not the CV itself. The site’s copy wobbles between “remote team” and NYC in-office perks — trivial, but a reminder this is an ~11-person company selling to national enterprises. The deepest structural gripe: Rebar’s customers are intermediaries. If manufacturers quote direct with their own AI configurators, or generalist platforms give contractors takeoff for free, the rep-firm estimating desk Rebar serves shrinks beneath it.

Outlook: the open question

Can a trade-specialist takeoff engine become the quoting system of record for HVAC distribution before blueprint-reading becomes a commodity? The bull case requires three things to be true: that HVAC-specific accuracy — schedules, spec books, addenda, line-card mapping — stays measurably ahead of what Bobyard, Beam AI or a foundation model with a PDF can do, for at least the two to three years it takes to ship spec review, submittals and bid management; that the basis-of-design data layer compounds into something distributors and even manufacturers will pay for on its own, converting quoting exhaust into a market-intelligence franchise no horizontal rival can replicate; and that usage-based revenue holds up through the current HVAC volume slump because customers buy it to bid more, not to cut staff. If those hold, Rebar owns the revenue-critical desk in a $66B distribution channel with almost no software incumbency, and the electrical and plumbing agents triple the surface. The bear case requires only two: that plan-reading converges — multimodal models make accurate takeoff a bundled free feature of every platform and eventually every manufacturer configurator — and that Rebar’s one-trade data advantage compounds slower than rivals’ cross-trade capital, leaving a well-liked point tool acquired cheaply by Trimble, ServiceTitan, Procore or a distributor roll-up like Ambient, the very company its founders left. The tell through 2026-27: whether Rebar’s next disclosed numbers are absolute — ARR, named logos, net retention — rather than multiples off hidden bases, and whether any of the seven customer-investors expands to an enterprise-wide, multi-branch deployment. Vertical AI history says the workflow-and-data company beats the model company; Rebar is betting five years of highlighter time is the harder thing to copy.

How a challenger would attack it

The wedge. Rebar’s moat claim is HVAC-specific plan-reading accuracy, and that is precisely the layer multimodal foundation models are commoditizing quarter by quarter. A challenger with Bobyard’s or Beam AI’s capital doesn’t need to out-train Rebar on mechanical drawings — it needs to get close enough, then bundle: takeoff free inside a broader preconstruction platform sold to GCs and mechanical contractors, the parties who send the bid invitations Rebar’s customers receive. If the contractor’s platform already produces the equipment list, the rep firm’s takeoff desk becomes a checker, not a producer, and usage-based pricing collapses with the usage. Second vector: go around the intermediary. Rebar’s buyers are manufacturer reps and distributors; a challenger that partners with the manufacturers themselves — wiring takeoff directly into their selection configurators so an engineer’s basis of design flows straight to a factory quote — shrinks the desk Rebar serves. Third, the blunt one: Rebar is ~11 people with $14M against rivals holding $35-48M, no deep-learning researcher on the founding team, and every public metric a multiple off a hidden base. A rival that simply publishes audited accuracy benchmarks on messy real-world plan sets — the category’s known weak point — forces Rebar into a proof fight it may not have the eval infrastructure to win.

Same playbook, new buyer

The playbook — a founder who did the manual estimating job, CV trained on one trade’s document conventions, sold to the supplier’s quoting desk on usage pricing — replicates cleanly wherever equipment schedules and spec books drive the quote. Electrical distribution is the obvious first: switchgear and lighting packages are quoted off schedules by rep firms exactly like HVAC, the market is comparably sized, and Rebar has publicly pre-announced electrical and plumbing “agents” without shipping them — an invitation for a focused team to own that trade first while Rebar’s 11 people are consumed by HVAC expansion. The subtler shift is the buyer within the chain: Rebar sells to reps and distributors, but the engineer-side version — helping mechanical engineers produce and check the schedules themselves, and monetizing the basis-of-design data upstream where specs are written rather than downstream where they’re counted — captures the decision point that determines whose equipment gets quoted at all. Rebar can’t easily pivot there; its cap table now includes seven rep/distributor customers whose interests sit squarely on the sales side of the counter, and serving engineers would put it crosswise with the very channel that funds it.

Sources and further reading

• Crunchbase News — “Exclusive: Rebar Lands $14M To Help HVAC Suppliers Generate Quotes Faster With AI” (Mary Ann Azevedo, Mar 10, 2026)
• BusinessWire — “Rebar Closes $14M Series A Led by Prudence” (Mar 10, 2026)
• Pulse 2.0 — “Rebar: $14 Million Raised For HVAC AI Operating System” (Amit Chowdhry, Mar 10, 2026)
• ACHR News — “AHR Expo Announces 2026 Innovation Awards Winners” (Nov 2025)
• Rebar — About page and product site (2026)
• Grand View Research — North America HVAC Distribution Market report (2024 base year, 2025)
• Distribution Strategy Group — “HVACR Distributor Sales Fall 5.9% in January, HARDI Reports” (Mar 2026)
• BusinessWire — “Bobyard Raises $35 Million Series A” (Dec 10, 2025)
• BusinessWire — “Attentive.ai Secures $30.5 Million Series B” (Nov 12, 2025)
• Baird — “DMG and JB Are Acquired by Ambient” (Jan 2024)