Latest / The AI Practitioner Brief
Anthropic's RSI Risk, Grok's Caps, and OpenAI's Pando — November 3, 2026
Anthropic just told its own investors it might be close to AI that improves itself. xAI capped Grok access because serving it costs more than it earns. Same day, opposite problems. We go through the self-improvement threshold buried in Anthropic's S-1 filing, the math behind why Grok's free users outgrew its paying ones, and how OpenAI rebuilt its own version control to keep up with its coding agent.
Get full access to The AI Practitioner at aipractitioner.substack.com/subscribe
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.