concepts · youtube
How To Approach Your AI Evals
Hamel Husain · Jun 21
Hamel Husain on how to actually approach AI evals — the verification half of a loop. Anchor of his 4-video eval series (Jun 2026). Evals are what make loops converge.
concepts · youtube
Don't Build More AI Agents Until You Watch This
Nate B Jones · Jun 21
Nate B. Jones argues against agent-sprawl: loops and orchestration over building more individual agents. The case for designing fewer, better-connected loops instead of proliferating agents.
concepts · youtube
Agent Literacy: Claude vs Codex Interface Philosophy
Nate B Jones · Jun 14
Claude and Codex aren't just competing coding tools - they're teaching different approaches to agent interaction. Claude makes 'steering agents' feel natural while Codex makes 'dispatching agents' feel natural. These interfaces are training habits for how we'll work with AI agents across all knowledge work, not just coding.
resources · youtube
Trust Layer for AI-Generated Office Files (Second AI Attack)
Nate B Jones · May 31
Nate B Jones's 4-stage trust-layer workflow for AI-generated office files: a hostile-reviewer prompt plus two-model QC (Codex ⇄ Opus 4.7) producing one verified output — the 'second AI attack' that catches what a single generation pass misses.
resources · youtube
Pinecone Just Demoted Vector Search. Here's the Knowledge Layer.
Nate B Jones · May 13
Nate B Jones argues the AI-agent-memory war has moved past embed-and-retrieve. Even Pinecone is repositioning vectors as one component of a broader knowledge layer that includes graph relationships, structured data, and contextual retrieval. The thesis: production agents need a layered knowledge stack, not just RAG.
resources · youtube
AI Scare Trade: Market Reflexivity and Sector Repricing
Feb 21
Analyzes how AI announcements triggered cascading stock crashes across 8 sectors in 10 days — software, private credit, insurance, wealth management, real estate, logistics, drug distribution, commercial office. Hundreds of billions in market cap. The pattern: dump first, analyze later; the drops then create the reality they feared.
resources · youtube
200-Line Prompt Killed $285B: The SaaS Apocalypse and What Survives
Feb 21
Anthropic's co-work legal plugin — ~200 lines of markdown — triggered $285B in market-cap destruction across SaaS, legal tech, and data analytics. The argument: per-seat SaaS pricing is structurally broken, but the data and accountability underneath it are not. A pricing-model crisis, not a technology crisis.
resources · youtube
AI Phase Transition: Capability Overhang and the December 2025 Convergence
Feb 21
Argues that December 2025/January 2026 is a phase transition in AI — not one breakthrough but the convergence of model releases, orchestration patterns, and proof points crossing thresholds simultaneously. The result: a massive capability overhang where what's possible has leaped ahead of what's being adopted.
resources · youtube
Code Costs Nothing, Knowing What to Build Costs Everything
Feb 21
The bottleneck in software development is shifting from implementation to specification. AI agents write code fine, but their errors are conceptual, not syntactic — they build exactly what was asked, and what was asked is wrong. The most valuable skill is now describing systems precisely enough for AI to build them.
resources · youtube
Claude Code Agent Teams
AICodeKing · Feb 7
Agent Teams is Claude Code's multi-agent orchestration feature (experimental). A team lead spawns multiple independent Claude Code instances that work as a coordinated team with peer-to-peer messaging, shared task lists, and dependency tracking — full sessions, not hub-and-spoke sub-agents.