Grading My 17 Predictions for 2025: A Mid-Year Reality Check

Source: Nate’s Newsletter Date: 2025-08-22 URL: https://natesnewsletter.substack.com/p/grading-my-17-predictions-for-2025

Summary

Nate grades his own 17 AI predictions for 2025 at mid-year, prioritizing accountability over defensibility. Notable findings: AI adoption is happening in “dramatic pockets” not uniformly; his biggest blindspots were AI coding dominance, emotional attachment to specific LLM versions, and the open-source model explosion. AI companions as a real $120M market; smart mirrors as oversold. The broader value is the framework for distinguishing genuine trends from hype through falsifiable prediction-making.

Implications

• AI economics thread. The “dramatic pockets” AI adoption pattern confirms that broad enterprise-wide transformation is the exception not the rule — specific functions, teams, and use cases see 10x transformation while others see nothing. Understanding which pockets are advancing is more actionable than tracking aggregate adoption.
• AI product positioning thread. Nate’s blindspot on emotional attachment to specific LLM versions is a real product insight: users develop preferences for specific model personalities and behaviors, creating switching costs that pure capability benchmarks don’t capture. This has direct implications for AI product retention strategy.
• Watch: Nate’s full-year grade at end of 2025 as a calibration check — mid-year grades that update at year-end provide the most reliable signal quality for AI forecasting claims.