.jpg)
A few days ago I read a post by Clive Butkow.. If you don't know Clive: he founded Kalon Venture Partners, backed some of South Africa's best-known tech companies, and now runs Conducive Capital out of Johannesburg. So when he lays out how the startup game is changing, it's not a theory off a Twitter thread. It's a man who has written the cheques, sat on the boards, and watched companies live and die through more than one downturn.
His post was a list of seven shifts - old rule versus new rule. I'm crediting him fully because the spine of this is his. But it hit a nerve, because I'm living a lot of it inside my own agency right now, building AI products instead of just running campaigns. So I want to go deeper on each one and add what it looks like from the operator's chair, not the investor's.
If you're building anything in 2026, a startup, a product, a practice, a side project that's getting serious, then read this slowly. The playbook most of us learned is being torn up in real time.
Old rule: build a defensible product. New rule: you are the defensible product.
This is the one that reframes everything else.
For two decades we were told the product was the moat. Better tech, more features, a smarter algorithm, a patent. Build the better mousetrap and the world protects you.
That logic is collapsing. Anyone can spin up an AI-powered product over a weekend now. I've done it. You've probably done it. The thing that used to take a funded team six months takes a focused person a Saturday. So if the product can be cloned in a weekend, the product was never the moat.
What can't be cloned is you. Your network. Your taste. Your read on a market nobody else understands from the inside. Your ability to look at a problem and know, in your gut, from scar tissue, which version actually matters. The human judgment orchestrating the AI is the differentiator, not the AI.
This is liberating and terrifying in the same breath. Liberating because you no longer need a war chest to compete. Terrifying because you can't hide behind the product anymore. The company is an extension of your judgment. If your judgment is sharp, you win. If it's borrowed, you don't.
Old rule: build a platform, own a category. New rule: go so deep into one workflow that you become irreplaceable.
Everyone wants to build the platform. The horizontal play. The thing that does everything for everyone.
Here's the problem: the horizontal layer now belongs to the foundation model companies. OpenAI, Anthropic, Google and they own "general." You are not going to out-general them, and you shouldn't try. Every time they ship, they quietly delete a hundred horizontal startups that were really just thin wrappers.
The opportunity has moved to the opposite end: vertical depth. Pick one workflow in one industry and go so deep that you become the layer that industry can't function without. Not "AI for marketing." AI for reconciling medical aid claims for South African physiotherapy practices. Not "AI for legal." AI for one specific kind of contract that one specific kind of firm deals with every single day.
Narrow feels small. It isn't. Narrow is how you become irreplaceable, because depth is the one thing a general model can't fake. The foundation models know everything a little. You know one thing completely. That's your defensibility.
Old rule: build a better mousetrap and they'll come. New rule: technology is not a moat. Distribution is.
Let me say the quiet part loudly: intelligence is cheap and getting cheaper. The cost of a unit of "smart" is falling off a cliff. If your entire advantage is that your thing is clever, you're standing on sand.
What's actually defensible? Who you know. How you reach them. How sticky you are once you're in. Brand, distribution, and data, those are the new barriers to entry.
This is the one I feel most personally, because distribution is what an agency is. We've spent years learning how to reach people, build trust, and earn attention. In a world where the tech is commoditised, that muscle is suddenly worth more than the tech itself. The person who can build a decent product and get it in front of the right buyers will beat the genius who built a brilliant product nobody ever sees.
If you're a builder who's allergic to distribution, fix that this year. It's no longer a nice-to-have bolted on at the end. It's the moat.
Old rule: charge per seat, charge per user. New rule: charge for results.
The per-seat SaaS model was built for a world where software was a tool a human operated. You paid for access, and the human did the work.
But when an AI agent does the work, "per seat" stops making sense. Why would I pay for a seat nobody sits in? The most interesting new businesses are flipping it entirely: you get paid when the agent succeeds. Pay-per-resolved-ticket. Pay-per-qualified-lead. Pay-per-closed-claim. You're not selling access to a tool. You're selling the outcome the tool produces.
This inverts the entire SaaS playbook, and it's not for the faint-hearted. Outcome-based pricing means you're on the hook for results. You can't hide behind "well, they're paying for the license, what they do with it is on them." Your incentives and your customer's incentives finally point the same direction, which is exactly why buyers love it and why it's hard.
If you can stomach the risk, this is one of the biggest unlocks available right now. It's also a brutal honesty test: are you actually confident your thing works? Outcome pricing forces you to answer that out loud.
Old rule: raise, hire, grow, then figure out the economics later. New rule: get to real revenue with the smallest possible team, then scale on purpose.
The "growth at all costs" era is over, and 2026 investors are rewarding something completely different: efficiency.
Clive put it in a line that should be printed on every founder's wall: a $2M ARR company with four people is more interesting than a $20M ARR company with forty. Sit with that. The smaller company is the more attractive one. Not because revenue doesn't matter, but because the four-person company has proven it can turn effort into money without burning a fortune to do it. That's a machine. The forty-person company might just be a very expensive treadmill.
This is doubly true on our continent. We've never had the luxury of cheap, abundant venture capital papering over weak economics. African founders have always had to make the numbers work earlier and with less. For once, the global market is rewarding exactly the discipline we were forced to learn anyway. Lean isn't a constraint here. It's a head start.
Get to meaningful revenue first. Then scale deliberately, with intent, knowing your unit economics hold. Not the other way around.
Old rule: hire to scale. New rule: agent-first from day one.
This is the one I'd push hardest on, because it's the most actionable thing on the list and the one most people are still sleeping on.
The old reflex was simple: more work to do, hire a person. The new discipline is to pause before every hire and ask one question - can an agent do this? Not "can a human with software do this," but "can an agent own this end to end?"
Often the answer is no, and you hire the human, and you should. But a surprising amount of the time the answer is yes, or "yes for the repetitive 70% of it." The startups running remarkably lean right now aren't just being scrappy. They're being deliberate about what genuinely requires a human and what doesn't.
I'm building agent-first inside my own agency, and the shift in thinking is bigger than the cost saving. When you design the org around agents from day one, you stop asking "how many people do I need to do this work" and start asking "what's the smallest human team that can direct this work." Completely different question. Completely different company.
Build the agents first. Hire humans for judgment, relationships, and the things that genuinely can't be automated, which, conveniently, takes us right back to rule number one.
Old rule: fundraising inevitably becomes the CEO's full-time job. New rule: stay lean enough that you don't have to raise.
Every founder knows the trap. You raise, which buys time, but the raise itself consumes you. Suddenly the CEO is spending half the year in pitch meetings and data rooms instead of with customers. The fundraising becomes the job.
The founders winning in 2026 are quietly stepping off that treadmill. Not because capital is bad, it isn't, but because when you're lean and revenue-first (rules 5 and 6), you remove the need to raise. And the moment you don't need to raise, two things happen. Your terms get better, because you're negotiating from strength. And your time goes back to where it actually creates value: customers.
Spend your time on customer relationships no competitor can replicate, not on your cap table. The irony is that the founders who don't need money are exactly the ones investors fight to back. Desperation repels capital. Traction attracts it. Build the kind of company that doesn't need the cheque, and you'll have your pick of them.
Here's how I'd read all seven together, because they're not seven separate ideas. They're one idea seen from seven angles.
The product is no longer the moat. You are. Everything downstream of that follows. Because anyone can build the thing, the edge moves to judgment, depth, distribution, and discipline, the human stuff that doesn't come out of a model.
So go narrow enough to be irreplaceable. Win on distribution, not cleverness. Price for outcomes, because you're confident your thing works. Get to revenue with a tiny team. Build agents before bodies. And stay lean enough that money chases you instead of the other way around.
None of this requires Silicon Valley money or a Silicon Valley address. If anything, it rewards exactly the constraints we've always built under here. The founders who internalise this first, who stop running the old playbook out of habit, are going to look like geniuses in eighteen months. They're not. They just read the field a little earlier.
Thank you Clive Butkow for the framework that started this. The standard startup playbook is being rewritten right in front of us.
