Shreesha
Ramdas

The construction payroll clerk at a mid-sized specialty contractor typically handles 200 workers, three union agreements, prevailing wage compliance for 12 counties, and a timesheet system that still runs on handwritten paper cards. She is almost certainly running behind. She is almost certainly the only person doing this job.

Shreesha Ramdas spent months inside that world before he wrote a line of code. He hired a sales development rep and personally authorized the budget - roughly $200,000 - to run 200 meetings with construction firms across the country. Not pitch meetings. Discovery meetings. The kind where you shut up and listen until the pattern becomes unavoidable.

What he found was a back office that was "always thinly staffed, but always overwhelmed" - his phrase, repeated in every interview, because it captures something precise and structural about an industry that moves $2 trillion of economic activity per year while managing its workforce on spreadsheets and gut instinct.

That research became Lumber. And Lumber became one of the fastest-moving companies in construction technology - an AI-first platform that automates payroll, time tracking, HR, safety, compliance, and field productivity for specialty contractors. By early 2025, it had raised $21 million, acquired a credentials platform, and was deploying autonomous AI agents to handle the work that used to take a payroll admin two full days every cycle.

The choice to leave Silicon Valley for construction was deliberate, not accidental. Ramdas had spent years in well-served markets. He wanted a harder problem. He found one wearing a hard hat.

Ramdas's path to Lumber ran through some of the most consequential infrastructure companies in tech. His early years moved through Tata, MW2 Consulting, and Catalytic Software - organizations where he learned to navigate enterprise sales and build teams from scratch. Then came Yodlee, the fintech pioneer that helped invent account aggregation, where he led as General Manager before the company was acquired by Envestnet.

The pattern accelerated from there. He co-founded LeadFormix, a B2B marketing intelligence company, and built it to an acquisition by CallidusCloud (later SAP). He co-founded Strikedeck, a customer success automation platform, and built that to an acquisition by Medallia in 2019. At Medallia, he became SVP and GM of B2B - running one of the largest experience management companies in the enterprise market.

Along the way, he deployed capital as an angel investor and advisor into more than 40 startups across SaaS, cloud, AI, and data. His portfolio reads like a who's-who of enterprise B2B: Workato, Boostup, Promethium, Rocketlane. His portfolio exits include companies acquired by Google, Symantec, ServiceNow, and LegalZoom.

By the time Ramdas left Medallia to start Lumber, he had the pattern recognition to identify what construction was missing and the network to fill the company with talent. Lumber's founding team was partially a Yodlee reunion - colleagues who understood financial infrastructure and wanted to apply it somewhere that had never seen it.

Lumber started as a gamified time-tracking app in 2023. It is now a full-stack workforce operating system for specialty contractors - covering payroll, HR, time tracking, job costing, compliance, safety management, field productivity, and employee recognition.

The product's most counterintuitive design decision: it meets workers where they are. Construction crews have used paper timecards for decades. Forcing them onto mobile apps creates a change management problem that kills adoption. Lumber built AI that reads handwritten timecards with 94% accuracy, then gradually migrates workers to digital workflows on their own timeline. The technology adapts to human behavior. Not the other way around.

The back-office focus is intentional and differentiated. While most construction tech companies chase field productivity or project management (competing with Procore), Lumber targets CFOs, controllers, and payroll admins - the people who process certified payroll reports, manage prevailing wage compliance across multiple jurisdictions, and keep the company out of legal trouble.

The company's AI agents make this concrete. The Union Agent reads collective bargaining agreements and surfaces relevant clauses. The Prevailing Wage Agent audits certified payroll reports automatically. The Paper Timecard Agent transforms illegible handwritten records into structured data. Eight new products shipped in a single year. Another pipeline of autonomous agents in development.

In May 2025, Lumber acquired BuilderFax - a digital credential wallet that lets craft workers securely store, track, and share their certifications, training records, and qualifications. The acquisition added a missing piece: a way for workers, not just employers, to benefit from the platform. Construction workers now have portable career records they control. Employers get instantly verifiable workforce data. The integration directly addresses what Ramdas called "a national imperative to develop, certify, and retain skilled trade workers who will literally build America's future."

Before Lumber, before Strikedeck, the other side of Ramdas's career was capital deployment. Over nine years as an angel investor and advisor, he backed more than 40 companies across SaaS, cloud, AI, and data. Several went on to significant exits - a signal of pattern recognition that shaped how he approached Lumber.

Most startups in construction technology were built by outsiders with a thesis. Ramdas had a thesis too - but he stress-tested it with a method borrowed from the best customer-success playbooks he'd written at Strikedeck and Medallia.

He hired a full-time SDR before the product existed. Authorized a budget that would make most early-stage founders wince: roughly $200,000. The job was simple: get Ramdas in front of 200 construction firms. Not to pitch. To listen.

Those 200 conversations shaped every product decision Lumber made in its first two years. They revealed that the problem wasn't just payroll - it was the entire operating model of the back office. It was the certified payroll reports that had to be filed in 12 different county formats. It was the prevailing wage audits that required reading collective bargaining agreements the size of phonebooks. It was the paper timecards that were faxed in on Monday mornings and had to be manually transcribed before anyone could run payroll.

When Ramdas talks about Lumber's AI being designed for real workflows rather than ideal ones, that's where it comes from. Not a whiteboard. Two hundred conversations with the people who actually do the work.