GFO

Sequoia, 2023

From Generative AI's Act Two

Generative AI applications are expected to progressively move from level 0 to level 5 autonomy, becoming increasingly agentic and capable of solving problems end-to-end.

View original source ↗

Credibility?Composite credibility score, weighted blend of Specificity, Accuracy, and Calibration. Higher means more credible.

61/ 100

Evaluated

Specificity?Was the claim falsifiable? 100 means a precise, dated, quantitative prediction. 0 means an unfalsifiable platitude.

35

Accuracy?Did the predicted thing happen by today? 100 means clearly yes, 0 means clearly no, 50 means mixed or partial.

72

Calibration?Was the magnitude and timing right? 100 means right number and date. 0 means off by an order of magnitude or many years.

55

Reasoning

The Sequoia 'Act Two' prediction was directional rather than quantitative — it described a progressive movement from level 0 to level 5 autonomy without specifying exact timelines, percentages, or measurable milestones, which limits its falsifiability (hence a moderate specificity score). As of mid-2026, the trajectory is broadly confirmed: the industry has demonstrably shifted from purely generative, prompt-response AI toward increasingly agentic systems. Multiple credible sources confirm that 2025 marked a decisive inflection point, with Level 3 autonomous agents becoming the dominant paradigm, enterprise adoption of agentic pilots reaching ~25% of GenAI users in 2025 (Deloitte), and Gartner projecting 40% of enterprise applications will embed AI agents by end of 2026. Platforms like Anthropic's Claude Code, OpenAI's ChatGPT Agent, and others are demonstrating end-to-end task completion. However, the prediction implied a smooth progression all the way to Level 5 (full autonomy), and the evidence shows the industry is currently clustered around Level 2–3, with Level 4–5 still largely aspirational. The calibration is moderate: the directional trend is correct and the pace is roughly on track, but full end-to-end autonomous problem-solving at scale (Level 5) has not been achieved by mid-2026.

Sources

Last evaluated 6/1/2026, 7:18:37 PM, claude-sonnet-4-6