Origin raises $20 million for OG-1, a robot that applies joint compound, sands and primes drywall

Origin, a San Francisco robotics company founded in early 2025, came out of stealth on October 15, 2026, with $20 million in funding co-led by Accel and Zacua Ventures. Its robot, OG-1, does drywall finishing. It applies joint compound, sands and primes interior walls. Origin says the money will go mainly toward deploying OG-1 and building its fleet, and will also fund pilots with what it calls two of the largest drywall finishing contractors in the US.

What it does

OG-1 works on walls after the board is hung. According to Origin’s announcement, the robot sprays and spreads compound, sands, and primes to a professional finish, including corners, pillars, soffits and edges. Sanding uses real-time force control, meaning the arm senses its pressure against the wall and adjusts it as it moves. The robot reaches walls up to 14 feet high.

Origin’s product page lists the hardware: a wheeled base with a 30-by-35-inch footprint, cameras and LiDAR (a laser sensor that measures distance to build a 3D picture of the room), and a tool station with swappable end effectors. The announcement describes a three-wheeled platform with a 360-degree view of its surroundings. Origin says the robot works alongside other trades without a cordoned-off area.

The crew still has work to do around it. Per Origin, people handle site mapping and material replenishment, and the robot does the finishing passes. In a finishing contractor’s organization, that puts OG-1 under the foreman or superintendent who assigns rooms and keeps it supplied with mud and primer. Someone in operations would then compare its output with the crew’s.

The software is what Origin calls its Construction Action Model, a foundation model, which is a large AI model trained on broad data and then adapted to specific tasks. Origin describes three parts:

1. A reasoning system that plans the sub-tasks in a workflow.
2. An action policy that navigates, picks where to work and which tool to use, and checks the quality of the result.
3. A controller that turns those decisions into precise, feedback-driven motion.

Origin trains the model in simulation and on data from its live deployments. The company says the same hardware will later take on taping, drywall installation, drilling and conduit installation. The announcement presents those trades as plans. Today the robot finishes and primes walls.

Evidence so far

The evidence so far is Origin’s announcement and its own website. These figures come from Origin:

The production figures need context. The homepage reports more than 100,000 square feet completed and 5,000 hours worked. Dividing one by the other gives about 20 square feet per robot hour. That is a fleet-wide average across every job, and the page does not say whether the square footage counts wall area or floor area, how many coats each square foot received, or what finish level the work reached. Those three details decide how 20 square feet an hour compares with a finishing crew on your jobs.

The speed claims come in two versions. In the announcement, CEO Yogesh Ghaturle says the robots work at three times human speed, and the homepage lists 5x time saved. The human baseline behind each figure, along with the jobs and finish levels measured, is not yet public. The same applies to the 20% cost reduction and the 300% margin expansion on the homepage.

Origin says it put its first robots on New York job sites within 10 months of founding. It says OG-1 now works in commercial space, multifamily buildings, data centers, hotels and warehouses. The homepage lists Bouygues Construction, Performance Contracting Group, JM, Aroundtown, Nesma & Partners and MAREK as customers. Ghaturle explains the company’s timing this way: “We spent 18 months building before coming out of stealth because we wanted to prove what our robots could do on real job sites.”

The outside voices in the announcement belong to investors from Accel and Zacua, which led the round. The other backers include Unilazer Ventures, Schema Ventures, Z21, ISAI and Imad Ventures, plus angel investors from Anduril, Skild AI, 1X and Tesla. Zacua founding partner Vivin Hegde has joined Origin’s board.

How to measure it on your projects

Track finishing labor hours per 1,000 square feet of board, split by finish level. Divide the crew hours spent on finishing by the board square footage from your takeoff, then multiply by 1,000. OG-1 does not tape yet, so separate taping hours from the hours spent on compound coats, sanding and prime. Compare the robot only against the second group.

Capture the baseline before a pilot:

1. Pull timecards from the last three jobs your crews finished, and code each hour as taping or as later passes.
2. Take the board square footage for each job from the estimate or the takeoff.
3. Note the finish level the specs called for on each area.
4. Count drywall finish items on the punch list per 1,000 square feet.

During the pilot, log robot hours, the crew hours spent mapping and refilling, and touch-up hours after the robot’s passes. The crew’s support time belongs in the robot’s column. If the robot’s hours per 1,000 square feet look good only after you leave that time out, the comparison favors the robot unfairly. The labor productivity method built from daily logs and timecards uses a drywall example and fits this calculation. For background on which trades robots can reach, see Anthropic’s robot exposure index for construction occupations.

A first step is to have the foreman on one job in finishing split tomorrow’s timecards between taping and the later passes.

Availability and cost

Origin is preparing customer pilots in the US and Europe, with more scheduled in Germany, Sweden, France and Saudi Arabia. It is targeting 150 robots deployed by the end of 2027. Pricing and commercial terms are not public, and Origin takes inquiries through its website and sales@origin.tech.