The increasing dependence on the middleman (ft. DevDay 2025)
This is written as a reaction of DevDay 2025, OpenAI’s annual developer conference.
With technology, you never interacted directly with a product, there always were middleman. Suppose it’s a website, so you have
- the browser
- the operating system
- the computer manufacturer
- the internet company
And more. The vision of OpenAI is it to become such a middleman to all the services that you use on your computer. And a middleman of this size hasn’t existed so far. Let me explain.
Apps in ChatGPT
OpenAI announced the new Apps SDK, a ways for SaaS companies to integrate their services into ChatGPT. The first thought to that might be “whoa why would a SaaS like Canva want or allow to have such a middleman”. And the answer, especially for smaller companies, is distribution. If they don’t integrate, they’ll just get completely replaced by the competition, because when my Mom tells ChatGPT to design a photo album, then it’ll use the service that has integrated with ChatGPT, and my Mom won’t question that decision. When creating a presentation, people already spend a lot of their time in ChatGPT to assemble the contents. They used to copy-paste them out into a presentation tool, but when they can generate, edit, and export a presentation deck right on ChatGPT using the Canva app in ChatGPT, there’s no need to ever visit the Canva app, aka no reason to subscribe to Canva.
It’s an absolutely scary, but genius move. Sooner or later, a “normal person” (non-technical) will be able to spend 100% of their time on their computer in the ChatGPT interface. I am sure that sooner than later they’ll integrate a music player (as a Spotify integration), video player (as a Netflix integration), email interface (like a Gmail integration) or maps interface (like a Google Maps app) and invent a new kind of messaging service (I have some ideas that I’ll share later).
While for long people thought that companies such as Spotify, Canva and Booking would pay OAI for their LLM API and implement their own AI assistants and integrate those into their apps. Seems like OpenAI wants to turn the tables. The problem for the SaaS companies is though, that for example only running the backend of a slides company isn’t profitable and holds no way of upselling whatsoever. And slowly, OpenAI can replace Canva in ChatGPT with their own presentation tools and get the fastest adoption rate ever.
Adoption rates
Speaking of adoption rates, something that OpenAI really has figured out is how to incrementally make steps in a way that makes sure everybody adopts them. Every person who has no clue of technology whatsoever. They keep the chat interface that knows everyone front and center and slowly add features around that.
And the biggest advantage is, people trust their ChatGPT, so if it recommends a product, they try it. I can’t think of a service that people trust more than ChatGPT (as in, if there is an ad in Google or Amazon, people dislike that, but if ChatGPT had ads in the text, people would buy that).
From Model to Product
ChatGPT was the first consumer LLM wrapper and thus got all the news. “Have you tried ChatGPT” was a standard phrase amongst people outside the tech bubble by the start of 2023, so competitors such as Claude or Gemini simply didn’t get the media attention that ChatGPT got. That means that “ordinary” people like my Mom only know ChatGPT, and OAI knows that too.
They know that continuously fighting against the competition for the best model isn’t sustainable nor fun. They use their adoption advantage now to push from model to product. That allows them to focus on building the best integrations such as the Apps SDK now while just making sure their model is “on par” with the competition.
All I’m saying, respect OpenAI. You really are winning.
And if you missed it, here’s the introduction video to Apps in ChatGPT: