Physical AI is launching: co-bot welders

Investment Analysis Clubs Macro Economic Trends and Risks
1

Yes, indeed.

AI-driven robot cars are expanding, providing commercial taxi service on public roads today (all the while collecting more and more data).

AI-driven robot trucks are coming.

AI-driven robotic package- and item-handling in e-commerce and logistics is here (all the while collecting more and more data).

AI co-bot welders are arriving.

(And unions are taking this kind of technology into account in their contract negotiations.)

It starts with humans guiding/assisting the robot welders.

How do we think it progresses from there?

I wonder how much of the data from every-day production will be gathered as training data to train the future AI robots?

Probably all of it.

So humans are effectively training the robots while at the same time getting production work done: humans and the robots are accumulating the data that will support future automation with AI-driven robots.

In AI co-bot welding, the human handles physical material preparation and software inputs to configure the task and then the AI co-bot executes physical welding.

  1. Cohesive Robotics (Argus OS + Universal Robots)
    Cohesive Robotics builds an AI-driven vision system specifically designed to turn standard 6-axis cobots (primarily Universal Robots) into zero-programming welders.
    • AI Component: Their Argus OS uses an arm-mounted 3D camera to scan parts instantly. It maps millions of data points to recognize common weld joints, gap variations, and tack welds right out of the box.
    • Production Deployment: Used extensively in high-mix job shops for aerospace sub-assemblies and architectural metals. Human workers simply clamp irregular or changing part profiles onto a shared table, and the AI instantly generates the exact weld trajectory without a single line of manual code.
  1. Hirebotics (Beacon Recommendations System)
    Hirebotics pioneered smartphone-controlled cobot welding using Universal Robots hardware, but integrated cloud-based AI to handle process parameters.
    • AI Component: Their Beacon Recommendations platform uses a machine learning engine trained across thousands of real-world production runs.
    • Production Deployment: Employed by high-mix manufacturers like Athena Manufacturing. When an operator configures a new part, the AI analyzes the joint type and material specs to recommend precise voltage, wire-feed speeds, and cutting parameters. This allows shops to train inexperienced operators to monitor automated stations, freeing highly skilled welders for specialized tasks.
2 Likes
2

Ten use cases for physical AI at various degrees of development.

This development will only increase and accelerate as humans get better at iterating software development pipelines for robots that collect data in an automated way while on the job in production settings. This is automated reinforcement learning.

No. Use Case Est. Market Penetration Production Use
1 Autonomous Warehouse Transport (AMRs) 18% – 22% Early Mainstream: Standard infrastructure for modern e-commerce fulfillment centers.
2 AI Inspection & Visual Quality Control 12% – 15% Expanding: Rapidly replacing traditional pixel-matching vision systems. Highly integrated into semiconductor and pharmaceutical lines.
3 Vision-Guided 3D Bin Picking 8% – 12% Growth Phase: Fast adoption in automotive and electronics assembly.
4 Intelligent Sorting & Recycling Automation 7% – 10% Legislative Push: Driven by labor shortages and strict municipal sorting mandates. Deployed heavily in high-volume, modern material recovery facilities.
5 Autonomous Commercial Floor Cleaning 5% – 8% Niche Mainstream: Highly visible in large commercial transport hubs such as airports and major retail stores.
6 Smart Mixed-Case De-Palletizing 4% – 6% Emerging Standard: Found primarily at high-throughput distribution hubs for major retail and grocery giants.
7 Vision-Infused Collaborative Assembly (Cobots) 3% – 5% Early Adoption: Growing in small-scale electronics and medical device assembly where humans and robots share workspace.
8 High-Value Agricultural Crop Harvesting 2% – 4% Specialized Entry: Highly successful in indoor vertical farming and specific orchard crops (like apples/berries), but raw outdoor fields are still largely manual.
9 Autonomous Drone Security & Infrastructure Inspection 2% – 3% Regulated Growth: Deployed as the standard path for high-risk environments (oil & gas, electrical grids), but restricted by commercial airspace regulations.
10 Last-Mile Commercial Delivery (Sidewalk Robots) < 1% Localized Testing: Highly concentrated around university campuses and closed urban pilot zones, facing regulatory headwinds on public city sidewalks.
2 Likes