Guest appearance on Nirit Cohen’s “The Future of Less Work” podcast · June 1, 2026 · ~35-minute conversation.
Cohen opens with a framing: the lines between employee-owned and employer-owned used to be clear — everything you did at work belonged to your employer. AI second brains blur that line in both directions. The conversation covers what happens when capability shifts from individuals to human-AI combinations, who owns the brain a worker builds, and why productivity is the wrong metric for any of it. When we use AI tools, we’re building an extension of ourselves — a personal operating system for thinking — and that capability is becoming a real differentiator. The unit of work is no longer the individual; it’s the human plus the intelligence they’ve built around them and are able to orchestrate.
Where the human + AI system is already taking shape
Individual workers have been finding and using AI tools on their own since ChatGPT launched. The ones on the frontier have moved past using it as a simple answer engine and are using it to manage all the context for everything they need to do at work. At first it’s sports scores and birthday poems. Within two to four months, the AI knows more about a person’s working environment than their team does, or their boss, or their inbox.
BYOAI, and how it differs from BYOD
Fifteen years ago, iPhones came out and individual workers had better technology than their companies did, so people just used their own devices. Companies solved BYOD by offering their own equivalents — Microsoft got better, Google got better, workers got to work the way they wanted, and companies could manage it. BYOAI looks similar at first: buy your own AI subscription, use it for work. But there’s a structural difference. Consumer plans have unlimited use; corporate plans don’t. And the AI a worker builds accumulates context that feels deeply personal — workers don’t want to leave it behind.
Who owns the brain you build
Cohen pushes the IP question: not just the data, but the fact that you’re building tools that take your thinking further — who owns that when you leave? The real worry runs the other direction. Brian had been using AI this way since January of that year; within two days it changed how he worked. If he left his job tomorrow and couldn’t take it with him, it genuinely wouldn’t matter — he’d start fresh and have a new second brain within two days. The real worry is the company saying: “you’ve sufficiently populated your little personal Wikipedia. Now we can fire you and we keep your brain.” That’s always happened with documents and emails. The difference with AI is that so much of knowledge work happens inside your head — it’s invisible. Documents and emails aren’t the knowledge work itself; they’re its artifacts. Most AI systems today only target that visible layer. But an AI built as a co-thinking partner captures judgment, pattern recognition, and mullings — the level that actually matters — to the point where you could point an AI at someone’s entire working knowledge repo and have an AI-powered interface to them even when they’re not there.
Managing AI like a worker, not a technology project
Whether anyone wants AI to happen or not, it’s happening regardless. AI fails in corporations when it’s treated like a traditional technology project — evaluated, benchmarked against analyst reports, plugged together by a consulting firm. Think of it more like a worker instead: not because it’s conscious or has a soul, it isn’t and it doesn’t, but in terms of implementation. It needs its own login ID. It might read email but never send it; it might read the CRM but never write to it. And it gets trained the way humans do, with skills documents. The objections people raise — what if it hallucinates, what if it goes down — apply equally to humans, and organizations already figured out how to manage around those.
The subscribable-brain consulting model
Brian has made a public version of his own second brain — all his writing, thinking, and synthesis — online and free, so anyone can connect their AI to it. As a consulting model, imagine leaving a living, working brain behind in an organization after an engagement, and getting paid monthly for access to it. The knowledge-distribution model inverts: instead of workshops and deliverables, the deliverable becomes the knowledge system itself. For consulting, that’s a genuine game changer.
Joy as the calibration metric
The right question isn’t the macro one — consultant, engineer, CTO — it’s the task-level one: what parts of the job bring joy and satisfaction, and what parts drain energy in unhealthy ways? The things that got outsourced to AI turned out to be exactly the things that were draining; the things still done by hand are what actually bring satisfaction, like doing a podcast with an actual human. Once you know what brings joy at the task level, maximize it and minimize everything else. Individuals don’t care about productivity — companies do. Everyone’s AI system ends up different because everyone built theirs to solve their own specific problems; someone’s hard part is someone else’s fun part. The real question isn’t whether more of the same work means more productivity. It’s what the work now frees a person up to do.
Key quotes
“Your vote doesn’t count. My vote doesn’t count. It’s happening.”
“I more worry about the company saying, ‘Hey, employee, you’ve built all this… now we can fire you and we keep your brain.’”
“Individuals don’t care about productivity. Companies do.”
“Most knowledge work is previously invisible, and it was sort of off-limits to what technology could touch. AI fundamentally changes that.”
“Everything I had for 2027 and beyond I was doing myself live in production this past January. I couldn’t even go out too far.”
“What brings you joy and satisfaction — even at the task level? Once you figure that out, you can maximize it and minimize everything else.”