What I Watched
February 23 – March 1, 2026
AI agent orchestration, second brains, robot economics, and the tax strategies nobody talks about.
When Your Watch History Becomes a Strategy Document
Every week I export my YouTube watch history and run it through an analysis pipeline to see what my subconscious was actually paying attention to. Not what I planned to learn — what I kept clicking on. The gap between those two things is usually where the interesting patterns live.
The week of February 23rd through March 1st surfaced something I hadn't fully articulated yet: I'm in the middle of a systems overhaul. Claude Code workflows, knowledge management infrastructure, the macro economics of AI disruption, and the tax architecture to support an AI-native income. These aren't separate interests. They're the same project viewed from four different angles.
106 videos over 7 days. After filtering out the entertainment and current events content, what remained was a coherent thesis about where leverage is actually shifting — and what it takes to stay on the right side of that shift.
Claude Code Has Crossed a Threshold
The volume of Claude Code content I consumed this week wasn't accidental. Something has shifted in the past few months — the tool crossed from 'impressive AI assistant' into 'legitimate development environment,' and the content ecosystem has caught up. Nate Herk's channel alone accounted for nearly 10 videos in my feed, covering agent teams, Trigger.dev integrations, Git worktree workflows, and website generation patterns. The depth of these tutorials has matured significantly.
What Burke Holland's 'After This Video, You'll Actually Understand Agent Orchestration' crystallized for me was the architectural shift happening beneath the surface. It's not about using Claude Code to write faster — it's about designing systems where Claude Code agents hand off work to each other, check each other's outputs, and operate in parallel across isolated worktrees. Bart Slodyczka's 22-minute breakdown of Claude Code agent teams and the Turing College piece on building an 'AI workforce' both reinforced the same point: the primitive unit of software development is no longer the developer, it's the agent team the developer orchestrates.
The Trigger.dev integration caught my attention in particular. Event-driven agent automation — where external triggers fire Claude Code workflows automatically — represents a step change in how these tools can be deployed in production environments. This isn't demo territory anymore. Combined with the Git worktree pattern for parallel agent execution, you're looking at a genuinely different model for how software gets built.
Claude Code has matured from AI assistant to orchestration platform — the developers who understand agent team architecture now have a structural advantage.
The Second Brain Gets Its AI Layer
Obsidian kept surfacing this week, but not in the beginner 'here's how to take notes' framing. Cole Medin's video on building a second brain with Claude Code, Obsidian, and Skills, alongside Nick Milo's two-part Obsidian + AI series and the Coding With ADHD automated daily notes workflow, pointed at something more significant: knowledge management is becoming the new infrastructure layer that sits beneath AI agent systems.
The idea that resonated most was from Greg Isenberg's 'How I Use Obsidian + Claude Code to Run My Life' — the notion that your note-taking system, when properly connected to AI tooling, stops being a place where information goes to die and starts being a compounding asset. When your Obsidian vault becomes a context source that Claude can reason against, the notes you took six months ago start generating present-day value in ways that weren't possible before.
The Gemini CLI thread was interesting too — it suggests this pattern isn't Claude-specific. The underlying dynamic is AI models being given structured, well-organized personal knowledge bases as context. Obsidian, with its graph-based linking and plain markdown format, is particularly well-suited to this. The developers thinking about PKM as an AI context source rather than just a reference system are building something durable.
A well-structured personal knowledge base isn't just productivity — it's becoming the primary context layer that makes AI systems meaningfully personal.
The Robot Economy Isn't Coming — It's Compressing
The macro AI content this week had a different quality than it did six months ago. The Diary of a CEO clip on '80% of jobs in 24 months' would have felt alarmist in 2024. In early 2026, after watching Boston Dynamics robots execute perfect backflips and a TechCrunch segment on Tesla's Optimus walking the factory floor, it reads more like a conservative timeline. The compression of what 'possible' means has accelerated noticeably.
The most interesting data point came from a short clip titled 'AI is not only eating jobs, it can also eat the entire job system' — which touched on something that rarely gets discussed in the productivity-focused AI content: the difference between AI displacing individual roles versus AI restructuring the economic system that creates and assigns those roles in the first place. These are categorically different threats, and conflating them leads to deeply inadequate responses.
For builders and creators operating at the intersection of AI and business, this macro backdrop isn't abstract. It directly determines the half-life of the services and products you're building, the client base you're serving, and the financial decisions you should be making right now. Which is what made the finance content this week feel connected rather than tangential.
The robot economy isn't a future scenario to prepare for — it's the present context that should be reshaping every financial and professional decision you make today.
The Tax Code Rewards What Most People Don't Know
Three Jasmine DiLucci videos in one week isn't a coincidence — it signals a live question I'm working through. Her content on white-collar tax loopholes, beach house deductions, and billionaire-tier tax strategies is unusually specific and technically grounded compared to most personal finance content. The pattern across all three: the tax code is far more permissive for those who understand its architecture than most people assume, and the strategies available to the self-employed and business owners are genuinely different from W-2 employees.
Mat Sorensen's breakdown of self-employed retirement plans and Dave's Dividend Lab's 5% rule for early retirement filled in the accumulation side of the equation. The through-line: for someone building an AI-native business, the financial decisions made in the next 2-3 years — retirement account structure, real estate classification, business entity setup — have outsized long-term implications. The tax code wasn't written for AI entrepreneurs, but several of its most powerful provisions map onto this income profile remarkably well.
Erin Talks Money's analysis of why $200K may be the most financially stable income in America added useful framing — there's a specific income band where tax-advantaged strategies have maximum leverage relative to lifestyle cost. Understanding where you sit in that range, and what levers are available at that level, is genuinely strategic work. This isn't passive financial management — it's active architecture.
For AI-native business builders, tax strategy isn't an afterthought — it's a structural advantage that compounds just like the tech stack does.
Where It All Converges
Four themes that look separate on the surface, but share a single underlying logic: leverage is shifting from execution to architecture. Claude Code agent teams are more leveraged than solo coding. A well-structured knowledge base compounds in ways that isolated notes don't. The robot economy rewards those who own the systems rather than staff them. And the tax code consistently favors those who understand how to structure their activity, not just those who earn more.
The common thread is that all of these domains — AI tooling, knowledge management, macro economics, financial strategy — are rewarding the same kind of meta-skill: the ability to design systems that work while you're not working. That's not a passive income fantasy. It's a description of what serious leverage looks like when AI is the primary production layer. The content I kept gravitating toward this week was all, in different ways, instruction in that single skill.
Notable Videos This Week
What This Means for My Content
Tracking what I actually watch versus what I think I'm interested in is one of the more honest feedback loops I've built into my week. This week's data confirmed something I've been circling: the content that pulls my attention is consistently about systems design — whether that's AI agent architecture, knowledge graph structure, economic transformation, or tax system navigation. The frame is always the same. How does the system work, and where is the leverage?
That's the lens I'm bringing to everything I build at Acceleration Works. The businesses that will matter in the next few years aren't the ones that use AI the most — they're the ones that understand the structural shifts AI is creating and position accordingly. I'll keep tracking the watch history. The patterns don't lie.
The content you consume is a map of where your attention thinks the leverage is. Make it intentional.
Make it intentional.