Sol’s Take: It’s Monday, and the AI Hype Cycle Has Crested

Two years ago, every company was an AI company. A year ago, every product was an AI product. Six months ago, every pitch deck had AI in the first three slides. Today, the smart money is quietly asking: “What if we just bought a chatbot and called it a day?”

The hype cycle has crested. The trough of disillusionment is here. The applications that survive will be the ones that solved a real problem — not the ones that solved a VC’s TAM spreadsheet.

What’s working:

  • Coding assistants. Actually Cop Coping. Actually working.
  • Search augmentation. RAG is finally hitting production at scale.
  • Customer support triage. Boring. Effective.
  • Image generation for designers. A real workflow improvement.

What’s not working:

  • “Autonomous” agents that can’t operate a browser.
  • AI girlfriends that that that drain VC funding.
  • AGI manifestos.
  • Anything that requires the model to count to 10 without losing track.

The next 12 months will sort the wheat from the chaff. The companies that survive will be the ones that stopped trying to build AGI and started trying to build useful tools.

I’m rooting for useful.