404

Notes

February 22, 2026

Citrini Research publishes THE 2028 GLOBAL INTELLIGENCE CRISIS, sending SaaS stocks into freefall.

June 23, 2023

AI features that are primarily API-driven are easy to reproduce and commoditize.

Companies are scrambling to claim they’ve implemented AI-based functionality in their products. In reality, many are just making simple calls to OpenAI’s GPT-4 API. This is a ladder step ahead of 2015’s hype cycle-driven marketing pushes where every business was claiming to use ML or AI, when, in actuality, they were just running algorithms behind the scenes (versus features driven by transformers). But, from a differentiation perspective, competitors will be close on their heels.

Product managers should be looking for opportunities to apply transformer-based technologies within two distinct channels:

  1. General, commodified LLM models (like GPT and Bard) to drive features that elevate customer productivity
  2. Specific, homegrown ML models to drive features that deliver distinct insights and value to customers

The former are quick to implement and drive significant team member-to-output efficiency, while the latter are unique and augmentative to product-market fit and differentiation.