In this episode, I speak with Yogesh Ghaturle, Co-founder and CEO at Origin. Origin builds general-purpose construction robots, starting with drywall finishing and painting, and its robots work on live job sites in New York. The company has raised $20mn in total funding, co-led by Accel and Zacua Ventures. Before Origin, Yogesh founded Perpule, a cloud point-of-sale company that Amazon acquired.
What I love the most about this conversation is that Yogesh thought deeply about what he learned from his first startup, and applied those lessons to an industry that is global and personally meaningful to him. His thoughtfulness extends beyond that, into building a team through 5,000 interviews, and always surrounding himself with the best people.
In this episode we cover
(00:23) Three lessons from selling Perpule to Amazon
(03:13) Why own the full stack
(04:28) Why construction, a 4-month renovation that took 14 months
(10:35) What full stack means for a robot company
(12:09) Three robot generations in 18 months
(14:36) Why start with wall finishing?
(16:01) First job sites and what customers care about
(19:06) How Origin measures robot speed against a human shift
(19:54) First customers, wall finishing subcontractors in Brooklyn
(20:57) 6 robots, 100,000 sq ft and $20,000 a month per robot
(26:09) Tasks are model evals, workflows are customer evals
(28:31) Sun glare on the 19th floor
(30:18) The next trades after wall finishing, from taping to MEP
(32:12) The construction action model and 80% carryover
(34:43) 5,000 interviews to build the team
(36:57) India's role in robotics
(38:24) What people misunderstand about robotics companies
(39:36) Advice for an engineer in a small town in India
Below are the 7 ideas that stayed with me. You can read this without watching.
1. Customers buy delivered work, not robots
Origin's customers are subcontractors who deliver work for general contractors. Yogesh says that sets the bar.
"A customer does not care about a robot. Customer cares about their work getting delivered, right? Because they are functioning in a live setting, there is a job to be delivered to their own customers."
He names 3 things a robot has to deliver together. Those are very high completion rates, human quality, and a throughput that makes the ROI work for the customer.
My take is that robotics founders that stand out in 2027 will be like Yogesh, someone who thinks from the customer's perspective. Robots are just a tool, like a phone, or a computer. What makes customers open their wallet is the results, the jobs that get done.
2. Tasks are model evals, workflows are customer evals
Yogesh wrote this line on X in August, and on the show he explains it with a wall finisher's day. The finisher takes the elevator, finds the right apartment, identifies windows, doors and seams, climbs the ladder, does the ceiling and handles the edge cases.
"Just folding a tablecloth is a task, right? But getting that tablecloth out of a washing machine, then folding it, then putting it into a basket, then moving it to an area where it is going to be useful. That is a complete workflow."
"So that's why I say that the tasks are, like, robotic companies' evals and then workflows are customer evals."
I really like the framing between tasks versus workflow. Dyna's founder Lindon Gao said similar things at Dirty Jobs conference, where customers care about autonomy. They don't want to babysit the robot. His example was that even though the robot was able to fold napkins, someone needs to feed the dirty stack and remove the clean, folded ones. A complete workflow includes before/after tasks for the customer, so the value is fully realized.
3. Robots priced against the crew
Origin rents its robots to the customers. "Every robot is like around $20,000 a month," and the contracts it has signed run for a year. He expects pricing in the US and Europe to sit around $20,000 to $30,000 a month per robot. In his words, customers do simple math. Would they have paid the same for a human crew to do that work?
He expects each robot to last 4 to 5 years, and Origin is working on reliability and serviceability, including HALT testing, to reach a 5-year life.
This sounds so simple. But the more I think about it, I realize it is because of the workflow framing. Only if the robots are able to complete a workflow (instead of piecemeal tasks), can the customer work out the math between human crew vs. robots.
4. The field finds the edge cases
On Origin's first day in New York, the robot was finishing walls on the 19th floor of a west-facing building. Toward sunset it was working right in front of a window.
"There was a lot of glare that was coming from the sun itself, right? And when the robot was evaluating the wall, and figuring out which part of the wall it should work on, the models just stopped working."


