Bizarrely, my 5 year plan was exactly what they did, but step 1 is release a cross-platform app and get it reliable as possible while waiting for an AI level up.
Releasing an app, especially when they're trying to get ahead of the game and are staffed and funded to do so, would beg the question "Why do just an app? It's the same as ChatGPT, which is free."
The reviews are bad, they're certainly not on the precipice of selling a million units. but they had a ton of preorders and the strategy worked, they're in the conversation, names out there, and they'll be forever associated with AI hardware. A GPT wrapper app with the same level of coverage would get the same general reaction we're seeing now. A GPT wrapper app with the coverage a wrapper app gets would get much less attention.
Now, whether or not the buyers will agree with their thinking is another story.
But there is population out there who like to go outside and show off their fancy things to other real, live people. They are the intended audience.
Yes, it very well may be that this is not fancy enough to win anyone over, but such is the nature of business. Not everything is a hit. Even successful companies like Apple and Google that you would think should know with certainty what people want have had their fair share of flops. You really don't know until you try.
It's just much easier to stand out with a beautiful hardware design than the 200th AI app released this month. And you can hardly buy anything from teenage engineering at this price point ;) Except the playdate which is much cooler.
But yeah it should have been an app.