---
title: Physical AI and deterministic automation | OpiFlex
description: The WEF white paper on physical AI expects several kinds of robotics to coexist. Two routes to robots that handle change, and where we use AI ourselves.
image: https://blog.opiflex.se/hubfs/Images/Blog%20Images/Slide8.png
---

[Skip to content](https://blog.opiflex.se/opiflex-blog/physical-ai-and-deterministic-automation-two-routes-to-flexible-operations#main-content)

Zoeken

[![](https://blog.opiflex.se/hubfs/OpiFlex%20logo%20PNG%202017-03%20(Small).png)](https://www.opiflex.se/) 

[Back to the website](https://www.opiflex.se/)

# Physical AI and deterministic automation: two routes to flexible operations

[robot programming](https://blog.opiflex.se/opiflex-blog/tag/robot-programming) [flexible robotics](https://blog.opiflex.se/opiflex-blog/tag/flexible-robotics) [Dynamic Program Generator](https://blog.opiflex.se/opiflex-blog/tag/dynamic-program-generator) [Future of robotics in manufacturing](https://blog.opiflex.se/opiflex-blog/tag/future-of-robotics-in-manufacturing) [Physical AI](https://blog.opiflex.se/opiflex-blog/tag/physical-ai)

Updated at: 14 October 2026

There is more than one way to make robots handle change. I find it useful to look at two of them side by side, because they aim for the same thing.

In September 2025, the World Economic Forum published the white paper *Physical AI: Powering the New Age of Industrial Operations*. It describes physical AI as robotic systems capable of perception, reasoning and autonomous action. It also notes that industrial operations are shaped by rising complexity and workforce shortages, which is a good description of many of the factories we visit.

![Summary of the World Economic Forum white paper Physical AI: Powering the New Age of Industrial Operations](https://blog.opiflex.se/hs-fs/hubfs/Images/Blog%20Images/Slide8.png?width=1280&height=720&name=Slide8.png)

## Three kinds of robotics, side by side

One point in the paper stays with me. It expects three complementary robotics systems to coexist: rule-based, training-based and context-based. Not one replacing the others, but each where it fits.

That matches what we see on the factory floor. A station that runs the same part for years has different needs from one where the mix changes every day. Different tasks call for different tools, and there is room for more than one route.

## The physical AI route

Physical AI builds on the robot perceiving, reasoning and acting on its own. In the training-based kind, the robot learns its movements. It is an exciting field to follow, and many are working on it now.

## Our route

We set out to solve the same challenge, robots that handle frequent change, but in a fully deterministic way. With our patented Dynamic Program Generator, the operator describes the flow and the robot program is generated automatically, in 1–3 minutes and without robot skills. In production, the program adjusts in real time to input from ERP, quality control, smart sensors and vision.

We chose this route because of the robots we run. They are large and fast, and in production a single crash, or just a stop, can be very costly. Using AI to create the robot program would risk making it take longer to create a new program, and it would add risk. So the robot programs and movements contain no AI.

## Where we use AI

That does not mean we have left AI out. We use AI in some vision applications, for example in Goods2Robot, with our own vision system that does not need massive training. AI helps the robot see. The code that moves the robot is deterministic.

The two routes are not opposites. They are different answers to the same need: automation that is built for change rather than for a single product.

## Comparing notes

If you are looking at physical AI for your own operations, I would be glad to hear what you are considering and share what we have learned on our side. [Contact us](https://www.opiflex.se/en/contact/).

*Source: World Economic Forum, [Physical AI: Powering the New Age of Industrial Operations](https://www.weforum.org/publications/physical-ai-powering-the-new-age-of-industrial-operations/), September 2025.*

Johan Frisk

[Read more from Johan Frisk](https://blog.opiflex.se/opiflex-blog/author/johan-frisk)

## Related articles

- ### [![OpiFlex Dynamic Fenceless Safety for large industrial robots at high speed](https://blog.opiflex.se/hs-fs/hubfs/Images/Blog%20Images/p04-fenceless-safety.png?length=500&name=p04-fenceless-safety.png) flexible robotics mobile robots fenceless safety OEM & Tier 1 Fenceless safety for large robots at full speed | OpiFlex How a large industrial robot can run at full speed without a fence: the principle, the standards it meets, and what it means for floor space and manual access. Read more](https://blog.opiflex.se/opiflex-blog/fenceless-safety-for-large-robots-at-full-speed)
- ### [![OpiFlex and sustainable development: people, planet and profit](https://blog.opiflex.se/hs-fs/hubfs/Images/Blog%20Images/p22-people-planet-profit.png?length=500&name=p22-people-planet-profit.png) flexible robotics mobile robots Sustainability Sustainable automation: reuse robots | OpiFlex Three lines at 32, 24 and 26 % utilisation, one shared mobile robot at 82 %. An example of how standardised, mobile robots get more hours and are reused. Read more](https://blog.opiflex.se/opiflex-blog/robots-that-are-reused-not-replaced)
- ### [![OpiFlex Flow Editor: choose product flow instead of robot sequence](https://blog.opiflex.se/hs-fs/hubfs/image-png-Oct-13-2026-08-55-57-0386-AM.png?length=500&name=image-png-Oct-13-2026-08-55-57-0386-AM.png) Dynamic Flow Optimization robot programming Dynamic Program Generator Real-time robot adjustment from ERP and quality | OpiFlex ERP orders, quality control, sensors and vision adjust the robot's active program and flow in real time. New parts still take a 1–3 minute DPG program. Read more](https://blog.opiflex.se/opiflex-blog/a-robot-that-adjusts-in-real-time)

1. [Home](https://blog.opiflex.se)  /
2. Physical AI and deterministic automation: two routes to flexible operations

[![OpiFlex-logo-rev-G-white-3](https://blog.opiflex.se/hs-fs/hubfs/OpiFlex-logo-rev-G-white-3.png?width=250&height=75&name=OpiFlex-logo-rev-G-white-3.png "OpiFlex-logo-rev-G-white-3")](https://blog.opiflex.se/opiflex-blog/physical-ai-and-deterministic-automation-two-routes-to-flexible-operations#link)

```json
{
  "@context" : "https://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Johan Frisk",
    "url" : "https://blog.opiflex.se/opiflex-blog/author/johan-frisk"
  },
  "dateModified" : "2026-10-14T16:53:40.852Z",
  "datePublished" : "2026-10-14T16:16:48.000Z",
  "headline" : "Physical AI and deterministic automation | OpiFlex",
  "image" : [ "https://blog.opiflex.se/hubfs/Images/Blog%20Images/Slide8.png" ],
  "mainEntityOfPage" : {
    "@id" : "https://blog.opiflex.se/opiflex-blog/physical-ai-and-deterministic-automation-two-routes-to-flexible-operations",
    "@type" : "WebPage"
  },
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject",
      "url" : "https://blog.opiflex.se/hubfs/OpiFlex%20logo%20PNG%202017-03%20(Small).png"
    },
    "name" : "OpiFlex"
  }
}
```