Posting my latest YouTube video here! 🤖
It starts as a review of the reBot B601-RS, but dives into my vision for the future of robotics.
Imagine telling a robot what to do, pointing, or demonstrating a task, and having an AI agent figure out the rest.
Rather than one giant neural net controlling every joint, agents write code and orchestrate traditional control, vision, navigation, and learned policies as needed.
As AI gets smarter, the bottleneck shifts from intelligence to real-world context and communicating human intent.
0:00 Intro
1:30 Unboxing
3:55 Performance
5:20 AI + voice programming
7:30 Teleoperation
8:25 Learned physical AI policies
9:15 Agentic robotics
11:57 Verdict
I might be commuting a lot more soon.
Strongly considering a Tesla.
I've wanted one for a really long time and with the miles I might soon be putting on, it might actually be the smarter financial choice (or whatever I need to say to justify it, right? 😆)
Last week I experimented with GPT-6 Astra controlling my robot arm.
The results were pretty incredible and reinforced a thesis I've been developing about the future of AI robotics.
Instead of giving an AI direct control of the robot's motors, I gave Astra a set of basic capabilities:
- A RealSense depth camera to observe its environment
- Linear movements within a bounded workspace
- A gripper with a fixed downward orientation
From there, I had Astra write its own robotics software, using YOLO for object detection and OpenCV for localization.
The result?
I can now give the robot a prompt like "arrange these blocks in a 2×2 grid," and it can identify their positions, plan the movements, and execute the task.
Here's what I think is interesting:
We're seeing increasingly capable multimodal LLMs that can reason about physical environments and write the code to interact with them.
Rather than relying exclusively on end-to-end "physical intelligence" models to directly control robot joints, we can use coding agents to generate deterministic robot programs that can be inspected, tested, and validated before execution.
There are still limitations.
Astra's coding and reasoning can be slow, sometimes taking 5-10 minutes to implement a new capability. But once that code works, the robot can execute it repeatedly without waiting for the model to regenerate it.
And modular software architectures are already helping reduce that iteration time too.
I believe this approach could be one of the fastest paths to bringing AI into real industrial robotics.
Imagine a factory where an operator describes what needs to be automated, and an AI agent observes the environment, writes the robot program, and iterates until the task works.
We're not there yet. But the building blocks are coming together.
I'm more convinced than ever the Midwest is the best place to build a robotics startup
- Deep manufacturing talent
- Lower land & development costs
- Dense high-value manufacturing network
SF is giving us infinite intelligence
Now we use it to reindustrialize the American core
I am happy to announce the camera-to-robot coordinate transform is complete!
Super tedious though…
I am considering building an open source automated tool for this with the RealSense, but doing it manually is a good baseline comparison.
The year is 2026.
Almost all robotics development is done with AI.
Yet I’m sitting here with graph paper and a marker gathering data to generate a correction function for Cartesian movements. 🤓