Enough with models and chatbots. Where is my AI App Store?
A shift of focus from models to applications is a necessary condition for AI Adoption. We are a long way from “there is an AI App for that” — but it's coming.
A few days ago, I hit prompter’s block. I went through my mental and actual notes and tried to formulate my tasks into questions. Nothing. With shock, I realised that prompting is not always a natural interface. Sometimes, prompting can be a real chore.
Lacking specialist competence and context, AI can get on your nerves. Why can’t I have a Finance AI to crunch my quarterly results and a Marketing AI to check what bots say about us? Must I really chat about everything with an eager assistant and not an expert? I want hyper-specialised, context-aware, plug-and-play, multi-modal professional apps.
In short, I want an AI App Store.
Where is my AI App Store?
To be sure, the market is bubbling and clever applications are emerging in all areas. At the time of writing, Rundown lists around 960 Apps. However, this is only 0.05% of the apps available on mobile, and for most people, chatbots are synonymous with AI. We are a long way from the “there is an App for that” motto of the mobile explosion.
AI Labs hoard the attention and margins by design, as the Economist reports. To finance their relentless pursuit of AGI, Labs keep raising their prices, squeezing the App makers’ growth potential. Everyone remains on friendly terms, but competitive pressure is building.
In response, App makers are on a mission to add specialist value on top of LLMs, threatening the latter with commoditisation. To a lawyer, Gemini and Claude offer nondescript, generic capabilities. The real choice is between Harvey and Genie, and how their legal AI features match the company’s specific needs.
Whisper it: LLM commoditisation is a precondition of widespread AI adoption.
I don’t need a 5mm drill. I need a 5mm hole.
We cannot have widespread AI adoption without Apps, and the industry is already adjusting to close that gap. As Sequoia comments, there is no one left to compete on infrastructure or models — the CapEx requirements are too high. But as corporate IT and systems integrators transition from Build to Buy strategies, the field is wide open for App makers.
A wave of innovation and eye-watering valuations in the App layer is on the way. As attention shifts to the App makers, they will gain the upper hand on pricing, putting the Labs in a tight spot: either abandon their bone-headed obsession with AGI, or be absorbed by the Cloud providers. No matter how that plays out, a shift of focus from models to applications is a necessary condition for AI Adoption.
My prompt fatigue reminded me that as the fascination with the technology fades, I am growing impatient with one-size-fits-all bots. As the adage goes, no one needs a 5mm drill — they need a 5mm hole. I look forward to my box of AI Apps for all the holes I need to drill.