This issue covers building and maintaining AI harness code, Elden Ring’s simple goal-based NPC AI, and agent-driven development practices from ex-Meta setups to collaborative coding. It also highlights security and tooling updates—from Daybreak and Mistral OCR 4 to Mythos benchmarks, the fired Google Workspace CLI creator, and running GLM-5.2 locally.

• AI harness code needs different designs for production (locked-down, safe) vs. training (flexible, exploratory), and treating either as permanent breaks down as models evolve.
• Elden Ring's NPCs achieve rich combat behavior through a simple stack-based "Goals" system mixing randomization and hierarchy, not complex behavior trees.
• A former Meta L8 engineer replaced meetings and busywork with a voice-driven planner and parallel agent workflows to focus solely on design and strategy.
• A Google engineer was fired after his self-made Workspace CLI tool went viral, illustrating tension between corporate control and grassroots innovation.

This article stitches together several Twitter threads: it starts with a plain-English analogy for what model weights are, then shows how to iteratively refine code using Perplexity AI (with a stock portfolio analyzer example), and ends with practical steps for running LLMs locally on Apple Silicon via tools like Ollama.

• Model weights are literally just a file of decimal numbers (floats) that get loaded into memory and multiplied against input tokens to produce output.
• Iterating daily on prompts (breaking big tasks into focused, checkable segments) rather than stopping at the first working result yields dramatically better outputs, as shown with a stock-portfolio analyzer.
• A single detailed "autonomous team" prompt generated a 4,500+ line full-stack fund operating system (Thesium.finance) theoretically enabling 1-2 people to replace a 10-person analyst team.
• Apple Silicon's unified memory architecture makes it the easiest, most performant path for running LLMs locally, with tools like Codex Desktop launching on Mac first.

This unrolled thread covers four topics: a plain-language explanation of model weights, strategies for refining code with Perplexity AI’s Computer, an AI-native fund system built end-to-end, and tips for running LLMs locally on Apple Silicon using Ollama. It walks through each use case with examples and practical advice.

• Perplexity's Computer platform generated 4,500+ lines of full-stack code for Thesium.finance, an AI-native fund aiming to replace a Bloomberg terminal and ten analysts with two humans overseeing automated trading agents.
• Getting usable output from AI coding tools requires iterative refining through multiple prompts, not just a single initial request.
• Apple Silicon's shared CPU/GPU memory makes it well-suited for running LLMs locally, letting users host models and keep data/inference private without needing to understand the underlying math.

This article compares running AI agents locally on a Mac Mini with Ollama and open-source models versus hosting them on a cloud server using Claude or Gemini APIs. It breaks down upfront and monthly costs—about $35/month local amortized versus roughly $73 for Gemini and $123 for Claude—and highlights performance, privacy, and usage trade-offs.

• Local Mac Mini setup costs ~$35/month amortized over 3 years vs ~$73/month for Gemini API or ~$123/month for Claude API on a cloud VPS
• Over a year, local stays around $420 while Gemini runs ~$876 and Claude ~$1,476
• Local open-source models handle summarization, document search, and private data well but lag behind Claude/Gemini on coding, complex reasoning, and long-context tasks
• Cloud pay-per-use makes sense for occasional agent use, while local hardware pays off for daily or nonstop operation

This article walks through why and how to run large language models locally, covering privacy, cost, offline access, and control. It breaks down hardware needs, quantization, PC versus Mac setups, and starter software to get models up and running.

• Local open-source models (GLM-5.1, Kimi K2.6) now match or beat top cloud models like Claude Opus 4.6 on coding benchmarks, handling 70-80% of typical daily AI tasks.
• 4-bit quantization (Q4_K_M) shrinks memory needs to ~0.6-0.7 GB per billion parameters with no noticeable quality loss.
• RTX 4090 PCs outperform M3 Max Macs on speed (100-140 vs 40-60 tokens/sec) for sub-24GB models, but Mac Studios with unified memory (up to 512GB) can run massive 671B-parameter models entirely in RAM.
• A 3090 PC build costs $1,500-2,000 and draws 450W with heavy noise, versus a quieter Mac Studio pulling only ~60W while supporting far larger models.