OpenAI's ChatGPT App Store: What It Means for Industrial Automation
OpenAI just opened the floodgates for third-party apps inside ChatGPT, and I'm watching to see which automation vendors actually show up.
By
·26 May 2026·4 min de lectura
Zero. That's how many industrial robotics companies I've spotted in OpenAI's early app announcements so far.
Look, I've been tracking this space since before most of these AI companies existed, and when OpenAI announced that developers can now submit apps directly to ChatGPT, my first thought wasn't about consumer gadgets. It was about whether anyone from our world, the people building actual machines that move actual things, would bother showing up.
OpenAI rolled out two things last week. First, developers can now submit apps for review and publication inside ChatGPT itself. Approved apps show up in a new in-product directory. Second, they released what they're calling the Apps SDK, which lets developers build what OpenAI calls "chat-native experiences" that can take real-world actions.
Real-world actions. That phrase caught my attention.
When I was at Kuka, we spent years trying to make robot programming more accessible. The running joke was that our interfaces were designed by engineers for engineers, which meant everyone else just stared at the screen. The idea that you could eventually tell a chatbot to configure a palletizing routine and have it actually do something useful? We talked about that in break rooms back in 2015. Seemed like science fiction.
Cobertura relacionada
More in Industrial
The Apple supplier priced its shares at the maximum and still had to turn away demand, which tells you something about where hardware money is flowing right now.
James Chen · 25 Jun · 5 min
Prime Day deals on Echos and Ring cameras are fine, but let's not confuse consumer gadgets with the serious robotics work happening in warehouses.
Robert "Bob" Macintosh · 25 Jun · 3 min
Amazon's CEO made his first India trip and left behind a $13 billion AI commitment and an aggressive quick-commerce expansion. The numbers are real. The execution is the hard part.
James Chen · 25 Jun · 6 min
A wave of arXiv preprints this week tackles one of manipulation's oldest problems: how do you get a robot to learn from imperfect, incomplete, or just plain missing data?
Here's the thing. OpenAI isn't building robots. They're building a platform where other people's software can live inside their chat interface. The Apps SDK, from what I can gather, lets developers create tools that respond to natural language and then execute actions, whether that's booking a flight, pulling data from a database, or (theoretically) sending commands to industrial systems.
The documentation talks about "bringing real-world actions into ChatGPT." That's vague enough to mean almost anything. Could a Fanuc or ABB build an app that lets maintenance techs ask questions about error codes in plain English? Probably. Could you eventually have a chat interface that actually programs robot paths? Maybe, though I'd want to see the safety validation on that one.
I called a former colleague who's now at a systems integrator in Michigan. He hadn't heard about the SDK yet. That tells you something about how disconnected our industry still is from the consumer AI hype cycle.
I'll be honest, I'm skeptical about how quickly this matters for industrial automation. The consumer apps will come first: shopping assistants, travel booking, that sort of thing. The industrial applications require something the SDK can't provide, which is trust.
When I worked on automotive lines, every software change went through validation processes that took weeks. Sometimes months. The idea that a maintenance tech could chat with an AI and have it modify robot parameters... that's not a technical problem. It's a cultural and regulatory one. Plant managers don't want surprises. They want predictability.
That said, there's a middle ground. Read-only applications, things that let you query robot status, pull historical data, ask about maintenance schedules, those could work. You're not changing anything, just making information more accessible. I could see Siemens or Rockwell building something like that.
OpenAI says approved apps will appear in a directory inside ChatGPT for "easy discovery." That's interesting phrasing. They're basically building an app store, and app stores create ecosystems. If enough industrial vendors build apps (and that's a big if), you could end up with ChatGPT becoming a unified interface for querying multiple automation platforms.
Or it could go nowhere. I've seen enough industry initiatives fizzle to know that "could" and "will" are different words.
The companies I'm watching are the ones already investing in natural language interfaces: Universal Robots with their programming tools, Veo Robotics before they shut down, some of the newer cobot startups. They've already done the hard work of making robot interaction more conversational. Plugging into ChatGPT's platform might actually make sense for them.
For the big players, Fanuc, ABB, Kuka, I'm less sure. They move slowly, and they've got their own proprietary ecosystems to protect. Why would they give OpenAI a piece of their customer relationship?
I don't have answers yet. Nobody does. But I'll be keeping an eye on that app directory over the next few months. The first industrial automation company to show up there will tell us something about where this is heading.
