The app paradigm inversion

There's a paradigmatic shift going on in terms of how people are now using AI to get things done.

Some examples from the last week:

• We ended up with slightly messy online whiteboard after running a pre-mortem activity with a client. I used Gemini 3 Pro to convert it into the start of a spreadsheet-based risk register.
• A friend told me his family is using ChatGPT to figure out what they can make from their regular wonky veg box delivery. They also suggested I could do likewise to create a custom nut roast for Christmas based on how I like it.
• My son used Perplexity to find a new clothes shop that he didn't know existed to find specific items he was looking to buy.

Our actions start with intentions. Now, instead of having to either choose from a bewildering array of apps – or try and figure out the correct term to enter into a search engine – we can just ask AI.

This, for better or worse, inverts the app model. An 'app' becomes something that the AI uses to perform it's job of satisfying the user.

The AI already has the user's context from previous interactions and revealed preferences. What the app is doing is providing additional context. That might be the availability of hotel rooms or price of flights if it's a travel app, for example.

Some recent guidance for developers provided by OpenAI shows how different this approach is in practice:

[B]uilding apps for ChatGPT is a different environment. Users aren’t “opening” your app and starting on the home page. They’re having a conversation about something and the model can decide when to bring an app into that conversation. They’re entering at a point in time. In that world, the best apps look surprisingly small from the outside. They don’t try to recreate the entire product. Instead, they allow users to access a few specific powers while using the app in ChatGPT: the concrete things your product does best that the model can reuse in any conversation.

Outside of ChatGPT, your app is often the destination. Users:

• Tap your icon
• Enter your environment
• Learn your navigation and UI patterns

Most product decisions flow from that assumption: “We own the screen.” You can invest heavily in layout, onboarding, and information architecture because users are committing to your space.

Inside ChatGPT, your app plays a different role:

• It’s a capability the model can call - for both context and visual engagement.
• It shows up inside an ongoing conversation.
• It’s one of several tools the model may orchestrate.

That means the “unit of value” is less your overall experience and more the specific things you can help the model and user accomplish at the right moment

While there are many benefits to individual users with this approach, there remain systemic downsides and drawbacks. These include:

• The amount of information being gathered by AI companies on individual users increases the surveillance panopticon. This is particularly problematic given that these companies are based in a country that is becoming more authoritarian.
• Given that these companies have a lot of money to return to investors, there will be a temptation to influence the choice of their users through partnerships, sponsorship deals, or something more nefarious.
• The process of 'removing friction' from online interactions diminishes digital skills and prevents users from creating mental models of how the digital ecosystem operates.

Whether or not you or I think that this paradigmatic change is good, bad, or indifferent, the fact is that it's happening. That's not to embrace some kind of techno-determinism, but rather just to observe which way the wind is blowing.

It's going to make for interesting decisions to be made by those serving users, as they've essentially now got at least three ways of accessing your digital services: the web, native apps, and apps within AI tools like ChatGPT.

My hope is that regulators, who seem to cotton-on to these kinds of changes years after they've taken place, also see which way the wind is blowing and help to ensure that we don't end up locked inside monolithic, proprietary digital ecosystems.