2 months in. Robots live in production, $1M contracted for year one. Proud of what this team has built. 🦾
Looking forward to @ycombinator demo day on dec 2!
I quit my job 2 months ago. Today, @deployrobo has robots live in production and $1M in contracted revenue for year one, $2.4M for year two.
The last couple of weeks have been crazy: preparing, commissioning, and finally watching our robots do real work on a customer’s floor.
Generalist robots can’t do everything yet, and that makes selling them genuinely hard.
Customers don’t buy potential. They buy work getting done.
As a deployment company, our job is to close that gap:
→ Find the use cases generalist robots can actually handle today
→ Stay right at the edge of what they’re capable of
→ Build a solution that adds real value to the business we’re serving
→ Go after customers with scale and use cases that fit our technology
That’s exactly what we did.
Through YC I talked to a lot of potential customers, with lots of flights and site tours.
One of the biggest lessons: knowing when not to take on a customer matters as much as closing one.
There’s a lot of work ahead: a lot of engineering, and a lot of making our customers happy.
Thank you to everyone who’s been on this journey with us.
We have the greatest team, one that’s been too busy to even take a group picture, haha.
It’s been a great 2 months, and hopefully a fulfilling lifetime.
See you at @ycombinator Demo Day. 🦾
4
3
477
2 months in. Robots live in production, $1M contracted for year one. Proud of what this team has built. 🦾
Looking forward to @ycombinator demo day on dec 2!
I quit my job 2 months ago. Today, @deployrobo has robots live in production and $1M in contracted revenue for year one, $2.4M for year two.
The last couple of weeks have been crazy: preparing, commissioning, and finally watching our robots do real work on a customer’s floor.
Generalist robots can’t do everything yet, and that makes selling them genuinely hard.
Customers don’t buy potential. They buy work getting done.
As a deployment company, our job is to close that gap:
→ Find the use cases generalist robots can actually handle today
→ Stay right at the edge of what they’re capable of
→ Build a solution that adds real value to the business we’re serving
→ Go after customers with scale and use cases that fit our technology
That’s exactly what we did.
Through YC I talked to a lot of potential customers, with lots of flights and site tours.
One of the biggest lessons: knowing when not to take on a customer matters as much as closing one.
There’s a lot of work ahead: a lot of engineering, and a lot of making our customers happy.
Thank you to everyone who’s been on this journey with us.
We have the greatest team, one that’s been too busy to even take a group picture, haha.
It’s been a great 2 months, and hopefully a fulfilling lifetime.
See you at @ycombinator Demo Day. 🦾
218
Deployment is the part after the demo. Here's how we made our arms' motor connection bulletproof, open-sourced so anyone running @i2rt YAMs can use it.
Great work by @avvasai 🚀🚀
A robot that works in a demo isn't deployed. A robot that keeps working on shift 400 is.
Our @i2rt_robotics YAM arms kept dropping their connection: USB adapters falling off the bus, ports wearing out, an arm stopping mid-motion and falling. Every patch we added made the system more complicated.
So we removed the root cause. We decoded the arm's CAN protocol and wired it straight into the Jetson Orin. A few dollars of hardware, no USB in the motor path, and 81,000+ messages in a cable-yanking stress test with zero dropped.
This is the stuff nobody posts demo videos of, and it's the whole job. Building an autonomous system is one thing. Making sure it keeps working is deployment.
We wrote it all up and open-sourced it, wiring guide included: deployrobotics.co/blog/cutti…
Huge shoutout to Sai Avva Pranav on our team for leading this one.
@NVIDIARobotics @nvidia compute all the way.
1
178
RoboJudge tracks what a robot is doing, so it knows when to call for help!
Humans are very good at intuitively knowing when they have done a task badly thanks to years of living on earth.
Teaching a robot exactly that is what we need.
Come watch us on yc demo day!
How does a robot know when it’s wrong?
Introducing RoboJudge from @deployrobo
A big part of deploying robots is building autonomous stations that know when something’s off: when to call for help, when to reset, when to just try again.
This matters more than it sounds. Until a robot can tell you it’s stuck, someone has to watch it all the time, which is a huge bottleneck in robotics deployments.
Once it knows to call for help, one remote operator can monitor a whole fleet instead of babysitting a single station.
So we built RoboJudge. It uses the same data we use to train and finetune the robot on a task, and learns not just the task but how it’s progressing over time. So the robot knows if it’s on track, or if it needs help.
Super excited to showcase RoboJudge at @ycombinator demo day!
2
197
Deploy (YC F26) retweeted
How does a robot know when it’s wrong?
Introducing RoboJudge from @deployrobo
A big part of deploying robots is building autonomous stations that know when something’s off: when to call for help, when to reset, when to just try again.
This matters more than it sounds. Until a robot can tell you it’s stuck, someone has to watch it all the time, which is a huge bottleneck in robotics deployments.
Once it knows to call for help, one remote operator can monitor a whole fleet instead of babysitting a single station.
So we built RoboJudge. It uses the same data we use to train and finetune the robot on a task, and learns not just the task but how it’s progressing over time. So the robot knows if it’s on track, or if it needs help.
Super excited to showcase RoboJudge at @ycombinator demo day!
800+ pens packed and a lot of attendees trying their best to trip up our robot. 😄
Thanks to everyone who stopped by our booth at DELIVER America! 🦾
Our robots packed 800+ pens for attendees over the course of the show at DELIVER America
What still blows my mind: we literally drove down from SF to Vegas the day before and collected about 30 minutes of fine-tuning data to make all this possible!
12
17
924





