You didn't need to predict the App Store

Remember “mobile first” movement?

A lot of the debates about AI and how to prioritize remind me of conversations I was having then.

(aside: beware of Vonnegut references)

Mobile blind-sided many as the shift was quick and the pivot required structural redesign for many applications.

Let’s level set

By Oct 2016. Phones + tablets crossed 51 % of web traffic. Desktop spilled its coffee. [1]
Facebook jumped first. By mid-2014, 62 % of its ad cash came from phones. [2]
Yahoo shrugged. Mobile felt like extra chores. Verizon later hauled off the scraps for $4.8 B. [3]
Nokia once owned 40 % of handsets. It snubbed touch screens. Microsoft bought the phone unit in 2013. So it goes. [4]
BlackBerry had half the US smartphone share in 2009. By 2016 it was a rounding error. [5]

When Users expectations switch, chasing them hurts when you think they’re where you left them when you found PMF.

AI is going to flip things

65 % of firms now run generative AI in at least one function. [6]
Early Copilot testers finish routine tasks 29 % faster. [7]
Yet only one org in ten says it’s “flying” with AI. [8]

Choosing to wait gifts momentum to someone else. Cats laugh.

Three suggestions for this cycle

Be AI-ready, not AI-perfect Hide models behind thin wrappers and feature flags. Swap GPT-4o, Claude 3, or Llama 3 like lightbulbs. GitHub’s new “Models” workspace shows the trick. [9]
Keep humans in the loop Stripe’s Radar flags odd payments, then pings real people. The model learns. The guardrails tighten. Everyone sleeps better. [10]
Today is the worst AI will ever be Bigger models + more data follow a tidy power curve. [12] Upgrade or watch others sprint past. So it goes.

What now?

Ship a tiny pilot now. Measure real lift. Design for hot-swaps every quarter. Mobile toppled giants in five years; AI levels up each release note. Pick your side of the case study.