Lumber wants to fix the analog trade.

The Scene

It is 6:47 a.m. on a Tuesday and a foreman in a hi-vis vest is squinting at his phone in the cab of a pickup. He is not checking the weather. He is approving 38 timecards before the concrete pour starts at seven. In a different decade he would have done this with a clipboard, a pencil, and a small fire of cursing. Today he taps four times in an app called Lumber, the green checkmarks line up, and the crew clocks in by walking through a geofence drawn around the staging area.

That little moment - thirteen minutes saved, an audit-ready certified payroll quietly building itself in the background - is the entire pitch.

Lumber is a San Jose-based software company, three years old, that has decided the construction trades deserve software that doesn't make people want to lie down. It builds payroll, time tracking, HR, safety, compliance, and field productivity tools for contractors - and in 2025 it began stacking AI agents on top of them.

The Problem They Saw

The American construction industry runs on paper. Not exclusively, but enough that paper still wins. Foremen scribble hours into tri-fold cards. Payroll clerks re-type them. Auditors arrive every quarter and ask for them. Somewhere a union rep is also asking for them, in a different format, for the same week.

Underneath the paperwork is a workforce in trouble. About 41% of the construction workforce is expected to retire within seven years. The replacements - to put it gently - are not lining up. Young workers who grew up tapping glass are not enchanted by jobs that ask them to fax. Contractors are losing both labor and the institutional memory of how to push it through a 1970s back office.

Meanwhile a typical mid-sized contractor is running 20-odd disconnected systems. Procore for the project. Sage Intacct or CMiC for the books. QuickBooks for the small subs. Gusto for the office. ADP for the field. Paper for the rest. The data doesn't reconcile. Nobody is sure which timecard is canonical. The whole thing is held together by the patience of one extremely tired payroll administrator named Sharon.

The Founders' Bet

Shreesha Ramdas is, by his own count, on startup number four. He sold his last one - Strikedeck, a customer success platform - to Medallia and helped take Medallia public. After that exit he did what serial founders always do: he started taking meetings about what to build next. The meetings kept landing on the same industry.

Construction had everything he wanted in a target. Vast spend, terrible software, a labor cliff, and almost no AI being applied at the ground level. He recruited Manish Kumar as CTO and pulled in a bench of operators from his Yodlee days - people who knew how to ship financial infrastructure and not break it.

The bet is a specific one. It is not that AI will replace tradespeople. It is that AI will replace the spreadsheets, PDFs, and email threads that surround tradespeople - the soft tissue of construction, which is currently load-bearing.

The Product

Lumber sells four overlapping things under one login. Payroll handles the genuinely brutal stuff: certified payroll, prevailing wages, union reporting, multi-state tax filing. HR runs applicant tracking, onboarding, benefits, and credential management - which matters more than it sounds, because a welder without a current certificate cannot legally weld. Field Productivity covers time tracking with geofencing, scheduling, equipment tracking, job costing, and safety incident workflows. And then, sitting on top of all of it, the AI Agents.

There are two agents in production today. The Certified Payroll Reporting Agent stitches together job data and pushes federal WH-347 forms without a human pasting between tabs. The Paper Timecard Agent reads, yes, photographs of paper timecards, and turns them into structured hours that flow into the same payroll run as the digital ones. Lumber says it has thirteen more on the roadmap by year-end.

Fig. 2 - The numbers Lumber prefers to put on slides. Sharon prefers the first one.

The Proof

Money is the easiest signal to read, so let's start there. Lumber closed a $15.5M Series A in March 2025, led by Foundation Capital, with Tishman Speyer - yes, the firm behind Rockefeller Center - participating. 8VC, Carbide Ventures, Sure Ventures, and Firsthand VC came along. Total raised to date: about $21M.

Lumber doesn't publish a customer list, but third-party data providers peg revenue near $43M annually and the company sells primarily to specialty contractors and mid-market construction firms across the United States and Canada. The product reads as built for them, not for Bechtel.

On partnerships: Lumber has two-way integrations into Procore, Sage Intacct, Sage 300, QuickBooks, CMiC, and Gusto. That is the boring sentence. The interesting sentence is that those six systems collectively touch the back office of nearly every contractor in North America - and Lumber sits politely between them.

The Mission

Lumber's stated mission is to make construction workforces more productive, compliant, and engaged - all under one AI-powered platform. The unstated version, which you can pick up from any founder interview, is closer to this: build the back office a 22-year-old apprentice would download without complaining.

That means multilingual by default, because a significant share of the trades work in Spanish. It means mobile-first, because nobody on a job site is opening a laptop. It means gamification - actual rewards and recognition built into the same product that pays you. And it means absorbing compliance burden quietly, so the company hiring the 22-year-old isn't passing their headache down to them.

Why It Matters Tomorrow

Construction is roughly 4% of US GDP. The industry's productivity has been famously flat for decades - one of the few sectors where economists actively argue about whether output per worker has gone down. If Lumber's wager is right, AI agents won't fix that by replacing labor. They'll fix it by absorbing the administrative drag that sits behind every billable hour - the compliance paperwork, the certified payroll, the timecard reconciliation, the credential renewal nag.

That's a less glamorous AI story than the ones currently dominating tech headlines. There is no chatbot girlfriend. There is no model that writes screenplays. There is only a payroll administrator named Sharon getting her Fridays back, and a foreman approving 38 timecards in four taps.

It is also, quietly, the kind of AI bet that funds itself. If Lumber's agents save a 200-person contractor even one full-time admin role, the math works. If they save more than that, the math gets embarrassing in the contractor's favor.

Back to the Pickup

The foreman is still in the pickup. Concrete pour is still at seven. But the 38 timecards he just approved have already triggered a certified payroll report his back office will not have to assemble by hand. The Spanish-speaking apprentice who just walked through the geofence is being onboarded into a benefits plan in his own language. The union rep, three counties away, will get a report tomorrow that matches the rep's format exactly. Sharon, in payroll, will run the Friday cycle without canceling lunch.

None of that is the revolution AI is supposed to look like. There are no robots. No autonomous excavators. No model that draws skyscrapers. There is only a small, quiet rearrangement of the soft tissue around a job site, so that the actual job can get done.

Which, if Lumber is right, is the actual revolution.