.png){kind=logo}
A year ago, I asked a question that might seem unusual for a labor marketplace:
What happens to the Pros on our platform when the robots arrive?
Our vision is to create economic opportunities for Pros and Partners, globally. Over ten million Pros depend on us for their livelihood, and many of them were already asking themselves this same thing. We feel deeply responsible for answering it.
We also noticed something unexpected. Robotics companies were already showing up on our app to engage our Pros. They needed people who are experts in the tasks that robots are being trained to perform, and they needed access to the diverse commercial environments where robots will be deployed. They were using the exact workforce we've been building.
That's when it clicked: Instawork can provide the human workforce for the physical AI economy.
UC Berkeley Professor Ken Goldberg has identified what he calls a "100,000-year data gap" between the vast datasets used to train AI language models and the limited, specialized data available for physical robot dexterity.
That gap is why, despite billions of dollars pouring into robotics companies, we don't have humanoid robots cleaning hotel rooms or unloading trucks … yet.
We estimate the entire industry collected roughly 100,000 hours of training data in 2024. That jumped to 1 million hours in 2025. In 2026, it's projected to hit 20 million hours. That's exponential growth, and it's still only 0.04% of the way to closing the gap. A growing number of companies are racing to build humanoid and general-purpose robots: foundation model labs creating vision-language-action (VLA) models, hardware companies building the physical machines, and everyone in between. Tens of billions in capital raised. Every single one of them has the same bottleneck: data.
But here's the thing: we've seen this movie before. When ATMs arrived, everyone predicted the end of the bank teller. Instead, the number of bank tellers increased. ATMs made branches cheaper; banks opened more of them; tellers moved from counting cash to building relationships. The pattern repeats with every major technology shift: the industrial revolution, electrification, the internet. New technology doesn't eliminate work. It transforms it and creates more of it.
A new wave is coming. But this one looks more like us! With arms, legs, and eyes.
Over the past year, I've reached out to and learned from some of the smartest people in robotics learning all over the world: researchers, lab leads, founders building everything from dexterous hands to full humanoids. I've been struck by how generous people have been with their time and perspective. In a lot of ways, we have no business being in robotics -- but the more I listened, the more I started to see where Instawork could help.
One thing I heard again and again: robots learn from watching skilled humans perform precise physical tasks in real environments. This means chopping vegetables with proper knife technique, navigating a crowded warehouse floor, or making a hotel bed to brand standards. And collecting this form of data in sufficiently high fidelity is genuinely hard. You can't just strap a camera onto a random person and press record. The data needs diverse environments, diverse tasks, diverse hand movements. And it needs workers who actually know what they're doing, because a robot trained on sloppy knife work will learn sloppy knife work (not fun for anyone!).
That's a labor operations problem. Recruiting skilled workers, training them, quality-assuring their output, managing a distributed workforce across geographies and environments -- that is literally what we do. We have 10M+ Pros with verified skills across hundreds of task types, deep relationships with Partners who give us access to real commercial environments, and reliability data on which Pros show up and perform consistently. It's a combination no data collection company can replicate from scratch. Many of these labs had already been coming to us organically, and we're now working with most of the prominent teams in the space.
Here's something that doesn't get enough attention: robots need people too.
An executive at a leading robot company told me they have a key component that needs replacing every 4-6 months. Not often enough to justify a dedicated field technician, but frequently enough for the downtime to be a major pain point. With the rise of autonomous vehicles, delivery bots, and robotic deployments, more and more companies are running into this. Scaling requires field support, but it doesn't make economic sense for every robotics company to staff dedicated operators in every market. We've already launched pilot programs for battery swaps, component replacements, and robot repairs with multiple robotics companies
And we've built what we believe is the first robotics certification program for hourly workers, and over 20,000 Pros have already earned certifications in the first few weeks. On the data collection side, a certified Pro learns to operate wearable camera systems, capture clean video of skilled tasks, and label sensor data -- so when a robotics lab needs someone to record hours of bed-making technique in a real hotel suite, they're getting a trained professional, not someone figuring it out on day one. On the technician side, a certified Pro learns hardware diagnostics, safety protocols, and maintenance procedures for the specific robotic systems they'll support.
Imagine a logistics company running an autonomous fleet across a dozen facilities. A robot throws a navigation error at 2 AM in a Memphis warehouse, or a sensor module needs swapping in Phoenix. Instead of waiting days for a factory technician to fly out, a certified Instawork Pro can be on site within hours. We're also developing VR training for robot tele-operation as labs scale their data collection beyond what on-site recording alone can support.
If a billion AI devices get deployed over the coming decade, the opportunity isn't just to maintain them. It's to create entirely new categories of skilled work: robot technicians, fleet operators, tele-operation specialists, and roles we haven't even named yet.
Last year, I had lunch with the CEO of one of the largest hotel chains in the world. They're thinking hard about how to create more consistency in their housekeeping operations using automation. They've been inundated with robotics companies wanting to deploy into their hotels, but they weren't sure how to separate a great demo from a great outcome. We know those environments, those workflows, those pain points -- because we're already doing the work in those exact locations.
We’re building a robot services marketplace -- connecting robotics companies with the businesses ready to deploy automation. We already work with both sides, which means we can do more than make introductions.
The future isn't robots or humans. It's robots and humans, working together. That's what Instawork Robotics Lab is built to do. Three capabilities, one platform: training the robots, supporting them in the field, and connecting them to the businesses ready to deploy them.
With every major technology shift, the question isn't whether new jobs will be created. They always are. The question is who builds the bridge between the current world and the next one.
We believe there will be an essential role for skilled professionals in every chapter of this story -- from training the first robots to deploying fleets to designing the human-robot workflows of the future. We want to make sure our Pros are there for all of it.
For the physical AI revolution, Instawork will be that bridge. Deep expertise across the industries where physical AI will have the biggest impact. Already delivering training data to robotics labs. Already certifying workers for data collection and field operations. Already building the marketplace that connects robots with the businesses that need them.
We're excited about this next act, Instawork Robotics Labs (IRL). You should be too.
