The author demonstrates two practical techniques for programming with System One models (fast classifiers that pick from multiple choices): tiered goal-setting for sequential decision-making, and tournament sampling for choosing among many options. He shows these approaches working in Doom gameplay and Wikipedia navigation tasks.

• Tiered goals work better than single-pass decisions: periodic prompts asking the model to choose short-term goals (e.g., "kill enemies" vs "collect armor") make it perform more intelligently than just reacting to immediate game state every 200ms.
• Tournament sampling beats confidence scoring when picking from many options: feeding 100 links at a time, then narrowing down, found the optimal path in seconds, while trying to score all 1,000+ links failed badly.
• System One models offer a practical alternative to tool calls for real-time systems where you need predictable latency and don't need full language generation flexibility.

This daily roundup covers OpenAI’s new Jalapeño inference chip for efficient data center AI, recent talent moves to Anthropic, and Google’s Gemini 3.5 Flash gaining desktop automation capabilities. It also highlights updates on open agents with GLM-5.2, Amazon’s lawsuit against Perplexity over agentic browsing, and NVIDIA’s NeMo AutoModel for faster Transformer fine-tuning.

• OpenAI and Broadcom launched Jalapeño, a gigawatt-scale LLM inference chip designed in just nine months using AI tools.
• Anthropic keeps poaching top talent from Google DeepMind, including Jonas Adler and Alexander Pritzel, following earlier departures of Noam Shazeer and John Jumper.
• Gemini 3.5 Flash now has native "computer use," letting it read screenshots and click/scroll/type across apps like a desktop assistant.
• Amazon is suing Perplexity, claiming its Comet browser disguises itself as Chrome to scrape Amazon Store data in violation of terms of service.