Canadian tech marketer currently living in San Francisco, previously London and Toronto. Jack of all trades, master of none.

Fragments of wisdom collected throughout my personal and professional life.

• You're not dumb, you're distracted.
• Don't let chasing perfection hold you back from greatness. Virtues can easily become vices.
• Honesty is the best policy.
• Never dwell on mistakes; but always learn from them.
• Reduce the noise you add to a conversation by supporting others privately whose ideas fulfill your need.
• To the man with a hammer, every problem looks like a nail.
• There's no such thing as a perfect employee; there can be pretty close to perfect teams though.
• Everything is a derivative; embrace inspiration.
• When you have energy, seize it. Don't waste the chance to get something done.
• Less is usually more. It's easy to add, hard to take away.

Think of a human life as a battery with a finite charge. Right now, a small handful of companies are wired into billions of those batteries, draining them through attention. The longer they keep us plugged in, the richer they get.

We fall asleep with the phone in our hand and wake up with it still glowing. We scroll on the toilet. We watch shows at 1.5x or 2x speed because the point isn't to enjoy them, it's to get through them. Long-form is too slow, so we've collapsed our attention into fifteen-second clips, and even those we swipe past after three. We're not being entertained. We're being processed.

I've spent my career in advertising, marketing, and technology. I'm pro-business. I'm largely pro-advertising. What I'm against is the deliberate engineering of human behavior to maximize how much we consume.

The incentive is simple. More time spent in, more money out. But when you optimize a feed for time-spent, you have to dial up whatever holds attention: porn, gambling, shock content, click-bait, rage-bait. Quality, productivity, and good-faith debate are inversely correlated with the metric these businesses are paid to grow. "Social media" is a misnomer. The product is antisocial by construction.

Multiply that by billions of people across decades and it stops being a personal habit. It becomes a civilizational drain. The collective hours we hand to the feed could fund scientific breakthroughs, art, businesses, communities, the deeper relationships we keep saying we want. Instead, the charge in each of our batteries gets converted into ad revenue and dopamine spikes that evaporate by morning.

It comes down to consuming versus creating. Every hour we spend on one is an hour we don't spend on the other. And we keep finding ways to consume faster, at 2x speed, in shorter clips, with smaller pauses between them, as if the goal were to fit the most content into a life in the least time. The platforms have made consuming frictionless and creating hard. We're a species of creators being trained to be a species of consumers.

The question I keep coming back to: how do we break the cycle? What would it take, culturally, technologically, personally, to flip the ratio? To make creating as cheap and as compulsive as scrolling has become?

I don't have the answer yet.

Parkinson's Law says work expands to fill the time available. In a startup there's a sister law nobody ever names: work expands to fill the skillset available.

If you can write code, you will. If you can negotiate a contract, you will. If you can edit a demo video, you will. If you happen to be passable at design, congratulations, you're the designer now. The moment your team learns you can do a thing, that thing becomes yours, and it stays yours until somebody else proves they can do it too.

Early on this feels great. You're useful in fifteen directions at once. You ship more in a week than you did in a year at your last job. Every meeting is the most important meeting of the company, because it's about you and three other people trying to keep the whole thing alive. Your skill stack is the moat.

Then it stops feeling great.

The trouble is that capability quietly becomes obligation. The one person who can do the thing keeps doing the thing, even when the thing is no longer the right use of their time. The CTO writes the marketing copy at 1am because she's also the best writer. The CEO debugs the production deploy because he's also the most senior engineer. Nobody complains, because there's nobody else to complain to. The work has to get done, and you can do it, so you will.

The cruel part is that this dynamic punishes range. The more things you can do, the more things you have to do. The narrowest specialists end up with the cleanest weeks. The generalists who built the company drown in the surface area of the company they built.

The trap is worth naming. Capability is not the same thing as obligation, and treating them as identical is how the people most able to lead the company end up too buried in operational glue to lead anything.

The test for what to keep isn't "can I do this." It's "would the company be worse if anyone else did this." Most things fail that test. Hand them off, even when you're the only person who currently can. Especially then.

I have a problem with practicality.

Practicality is a kind of internal accountant. Before you do anything, it runs the numbers. Will this be worth the time? Is there a clearer path to the same outcome? Is this the highest-expected-value use of the next four hours? It's useful mental software when you're choosing between two job offers or deciding whether to refinance. It's poor mental software for almost everything else.

The trouble with the accountant is that it can only count things it has a unit for. Career, money, status, time. The things that actually make a life worth remembering tend to be uncountable. The friend you have because you went to that one party ten years ago. The story you tell at every dinner party.

If you ran either of those through the accountant in advance, the accountant would have killed them. The party was inefficient. The story was awkward the first time you told it. The practical mind treats the unjustifiable as risk and routes around it. What you get is a perfectly hedged life in which nothing surprising can happen, including good surprises.

I've written elsewhere about how, in a startup, if you can do something you will. Capability becomes obligation. The same trap exists in life, just pointed inward. If you can analyze a decision, you will. The practical mind is the personal-life version of the startup CTO writing marketing copy at 1am because she can. It runs because it's running, not because it should be.

There's a quiet unhappiness that comes from being too good at this. You make all the right calls. You don't waste any time. And then one day you notice that the last five years are missing the kind of texture you used to have, because every choice was filtered through a function that only rewards predictable outcomes, and the things that produce texture aren't predictable.

There's also research behind this. Novel experiences create dense memories; routine ones get compressed. It's why childhood felt endless and why your thirties feel like they vanished. Picking the practical option every time isn't just less fun. It speeds up your life. The autopilot version of you doesn't remember most of the days it lived through, because there was nothing in them to remember.

The fix isn't to throw out the accountant. You still need it. The fix is to leave a budget line for things that have no reason. Say yes to the bad idea. Take the weird coffee. Do the thing for the hell of it. The point isn't that any of these will work out. Most of them won't. The point is to keep the door open for the small percentage that will, and to remember what it feels like to be moved by something instead of optimizing for it.

Some of the best things I've ever done were things I couldn't have explained in advance, and would have skipped if I'd tried.

Brands and agencies pour billions into programmatic display and video advertising every year. They send the money to trade desks, who shotgun ads across the open web in milliseconds, supposedly placing the right ad in front of the right person at the right time. That's the pitch Google sold the industry. The reality has corrupted almost everything programmatic touches.

The first problem is attribution. Most programmatic ads don't drive new behavior. They appear in front of people who were going to buy anyway, then claim credit when those people convert. Last-touch attribution rewards whichever ad got in the way last. Brands look at the dashboard, see programmatic "working," and pour in more money. The dashboard isn't measuring lift. It's measuring proximity to inevitable sales.

I've watched this from inside. I've sat with brands as they put tens of millions into programmatic because the trade desks were really good at convincing them they had to. Programmatic isn't worthless. It's just worth a lot less than the dashboard claims, and almost always worth less than the same money put into building the brand.

The second problem is what programmatic spend actually funds. Most of the money flows toward the cheapest available impressions, because that's how trade desks hit their volume targets. The cheapest impressions live next to clickbait, scam offers, dropshipping junk, and get-rich-quick courses. Publishers who want a piece of that spend are forced to host the same garbage advertisers, which dilutes their own brands. Reputable sites running ads for crypto pump-and-dumps. News organizations running ads for plastic gimmicks they'd never review favorably. Everyone takes the money because everyone takes the money.

It's the same dynamic as a grocery store where the processed junk is cheap and abundant and the fresh food costs twice as much. The economics push every actor toward the lowest-quality option. Healthy web costs more, scales worse, and is harder to find.

Publishers respond predictably. If junk ads pay best, you produce content that performs well alongside junk ads. Clickbait articles get more impressions, those impressions are easier to sell, and the loop tightens. Publishers care more about engagement on a slideshow about celebrity weight loss than on a long investigative piece, because the slideshow funds the next month of payroll. The result is a web where the economic incentives reward gimmicks, not quality, and the people producing quality work are slowly priced out.

Meanwhile, the publishers trying to escape the rat race are running a different kind of business. Substacks, paywalls, niche memberships, sites with ten employees that get their revenue directly from readers. It's harder. It scales worse. Most of them don't make it. But they're the only ones whose business model doesn't require junkifying their brand to survive.

Almost nobody comes out ahead in the current setup. Brands pay for outcomes they would've gotten anyway. Publishers chase the floor of the impression market and dilute their reputations to do it. Readers wade through worse content with more ads and less reason to trust anything they see. The only durable winners are the trade desks themselves, and the platform that owns the auction.

What would the web look like if programmatic hadn't existed? There's a real case in its favor. Programmatic kept the web mostly free and mostly open for billions of people. Without it, the big publishers would have walled themselves off behind logins decades ago. The cost of that openness has been the optimization itself, the relentless drive to monetize every impression as efficiently as possible. We accepted a worse web than we needed because the alternative looked like less revenue per page.

The web we have isn't the web we had to build. It's the web programmatic paid for.

Most takes on AI in the workplace assume the conclusion is layoffs. The story is so consistent it's started to feel inevitable: efficiency talk, headcount cuts, AI doing some of what people used to.

It's the wrong conclusion for most companies.

If AI makes knowledge workers meaningfully more productive, the right response depends on where you sit on the size curve. Big companies should pare down. Small ones should hire faster.

In a small company, every person is doing four jobs they're not specialized for. The founder is selling and recruiting and writing copy and doing taxes. The first engineer is also QA, DevOps, and on-call. The marketer is also the designer, the publicist, and the customer support lead. None of them are great at most of what they do.

AI takes the worst-leveraged hour of every one of those people and gives them something close to an intern. The marketer can spin up working code. The engineer can produce a passable HTML mockup. The founder can draft a contract that won't embarrass them.

But generalists with AI hit a hard ceiling. The marketer can ship code but can't QA it. The engineer can prompt a design but doesn't have the taste to tune it right. The founder can draft a contract but can't spot the clause that'll matter in litigation. AI buys you the first 80% of an adjacent discipline. The last 20% is where the work actually lives, and that part still needs specialists.

This is what makes the hiring math work. AI doesn't reduce the need for specialists. It makes them more valuable, because everything around them now ships at higher volume and needs their judgment to be any good. The multiplier compounds with hiring; it doesn't substitute for it. A small company that hires AI-augmented generalists without the specialists to catch what they miss ships more, faster, and worse.

That logic flips at the other end. In a large company, where the marginal value of a new hire was already low and another product manager mostly added meetings, AI doesn't make the existing people more valuable. It just lets them absorb the work that used to justify hiring more. Big orgs are going to shrink, and they should.

There's a precondition to all this. Productivity is not the same as growth. If your bottleneck is distribution, market fit, or sales cycle, doubling your team's output doesn't double anything. It just gives you a more productive team waiting for demand. The hire-faster argument really applies where the bottleneck is throughput: agencies, dev shops, anywhere the unit economics look like "we charge for what we ship." For everyone else, AI mostly just lets the existing team wait longer.

The framing "AI replaces people" is wrong. AI redistributes them. If you're not at one of the size extremes, you should figure out which one you're betting on becoming.

Martin Gurri called it the revolt of the public. The internet didn't replace the institutional experts; it dissolved their monopoly. Journalists and academics got buried under plausible-looking content from people who'd never done the work. Trust in expertise collapsed, and nothing replaced it.

The same shape is now playing out in business. AI in the hands of generalists doesn't make experts obsolete. It dissolves their monopoly on producing plausible work.

The marketer who can't code ships features. The engineer who can't design ships interfaces. All of it is plausible, none of it is great (yet). For most companies, plausible is enough.

When the bar for shipping drops to "passable enough that nobody pushes back," the marginal value of the expert collapses. Why retain the senior designer when the AI-augmented PM produces something that doesn't get blocked in review?

The expert isn't replaced overnight. They're replaced one budget cycle at a time. Each year there are fewer of them, the work gets a little worse, and nobody notices because the baseline has moved with the cuts.

The expert also looks slow. The AI-augmented generalist ships ten variants by Thursday. The expert is still on the first one because they're actually thinking about it. To anyone running a dashboard, the generalist looks like progress and the expert looks like overhead. Plausible AI doesn't produce better work. It produces a lot of average work at extreme scale, and in most companies that's indistinguishable from progress.

What you end up with is a workforce performing expertise. Everyone has the right vocabulary, the right tools open, and output that passes the smell test of someone who also doesn't know what good looks like. The room agrees the work is fine because nobody in the room is qualified to say otherwise.

The juniors never become seniors because there are no seniors to apprentice under. By the time the work is bad enough that customers notice, the people who knew how to make it better have moved on, retired, or never existed.

This is what the revolt of the public looked like in media, just in slower motion. The credentialed class lost its monopoly. The replacement wasn't better journalism. It was less journalism, plus a thousand louder voices producing content that looked enough like journalism to satisfy people who didn't know the difference.

Be wary of anyone selling you acceleration. Craft and care have always outweighed speed in anything worth keeping.

The promise of technology is a device that does everything for you. X is trying to become WeChat. Apple wants Siri to be your interface to the world. ChatGPT and Claude are pitched as the place you'll ideate, draft, and ship. The marketing keeps insisting the future is one app, one chat, one input box.

I don't buy it.

You don't expect a hammer to also be a pencil. You don't expect a pencil to drive nails. We've had millennia to discover that the tools that work best are the ones designed for a specific job. The Swiss Army knife exists, and people own them, and most of us still reach for the real screwdriver when it matters.

The same holds for software. The pitch for an everything app is convenience: less context-switching, fewer logins, unified data. The reality is that the cognitive load of doing everything in one place is exhausting. Imagine if every Google service lived in a single app. You'd never find anything, and you'd never want to use it.

Chat and voice aren't the best UX for most work, either. Typing a sentence to perform an action is already three or four seconds slower than the click that did the same thing in 2008. For repetitive operations or anything that needs spatial precision, a sentence is the wrong tool.

What actually works is the opposite shape. A generalist intelligence layer like Claude or ChatGPT becomes the connector and the context-keeper. The specialized tools stay where they are: Figma for design, your CRM for relationships, your analytics dashboard for the funnel. The intelligence doesn't replace them. It pulls from them and hands them back when it's time to do the work.

The future isn't one app. It's an intelligent layer on top of many specialized ones, each still great at the one thing it was built for.

So when X tries to add a payment system and a marketplace and a job board, or when an AI vendor claims you'll do all your work in one chat window, I get suspicious. The companies trying to be everything are usually losing whatever they were originally great at. The ones that stay focused get to be the irreplaceable specialty inside someone else's intelligent layer.

I'd rather use six well-built tools and let the AI know how to talk to all of them than one giant blob pretending to do them all.