Mr. Market

Products are out, brains are in

Often, something that should be free is expensive. Ex: bottled water, pre-washed lettuce, filing your taxes, and a bunch more stuff.

Soon, software may become one of those things.

The question is: as the marginal cost of software production trends toward zero and the number of capable tools in any given category multiplies, does software still command the prices it does today? Or does the cost of software just freefall?

My guess is no to both. The cost probably goes down, but it doesn't death spiral.

Instead, the primary reason people buy software will change. Rather than buying it because they can't build it themselves, they may buy it because they don't want to build it or they don't have the time to spin it up or maintain it.

And when they're deciding which tool to go with, most will base their decision on one thing and one thing only: the judgement of the people running the software.

Liability in enterprise

One thing for sure is that massive enterprises in highly-regulated industries probably do not wanna build their own internal tooling, even when software is fully commoditized and democratized.

Rather than thinking purely about ROI, enterprise companies will ask the following when they're evaluating tools:

Soon, even small companies will be asking these questions. It'll be like how institutional investors evaluate fund managers. The fund's strategy can be replicated, but the alpha can't.

Investors have learned, after decades of painful experience, that past returns are a poor predictor of future returns, but decision-making process and intellectual rigor are somewhat better predictors. This is why the best funds pitch philosophy and mental models, not just track records. Software companies IMO should dedicate themselves to doing the same.

Decision quality as the #1 non-fungible asset

Classical economics identifies land, labor, and capital as the three factors of production. The twentieth century added a fourth: technology, or total factor productivity. The twenty-first century's contribution to this list, still underappreciated, is decision quality at scale.

To break it down real quick:

Decision quality is bounded by something that scales poorly: the clarity of understanding of a complex, uncertain situation. You can automate the execution of a decision but you can't automate judgement about the best decision to make in the presence of true uncertainty/novelty/imperfect information. And the future is nothing if not uncertainty/novelty/imperfect information.

While there are always adequate decisions that you can make mostly using historicals, historicals will not give you the BEST decision. That changes in little ways constantly based on dozens of small contextual differences.

If I'm a company, and I'm handing off some part of my business to somebody, I want to bet on the team I think is most likely to make the highest number of right decisions most of the time. Especially now, with things changing so quickly.

Companies that can see the future still have something valuable to sell.

Metaphor time :) Think of a tug of war

Say you have two companies with the same CRM/data warehouse/LLM APIs. They're both operating under the same market conditions and chasing the same customers. Almost everything is the same.

In this scenario, the outcome is determined by the cumulative quality of a large number of small decisions made by the teams at each firm.

These questions:

Each of these decisions has a distribution of possible answers. The best answer in each case is uncertain ex ante. A team with better models of the world, more intellectual honesty about uncertainty, and more willingness to update based on better data will crush the team with worse thinking. Not every time, but definitely OVER time.

So yeah, it's like a tug of war. Two teams, one rope. Everybody's in the mud. The winning team will be the ones who are stronger, have a better strategy for gripping the rope, AND know how to rally the team to all pull together, at precisely the right time.

The strong-and-dumb puller and the weak-and-smart puller are both suboptimal. What you want is the strong-and-smart puller.

The Warren Buffett Play

"Thought leadership" as a term has been diluted into meaninglessness and also it's sounded dumb since the beginning. It's way past time to retire it.

In its degenerate form, it means posting stupid shit on LinkedIn and having a content marketer churn out AI slop blog posts with your name on them. Nobody cares about this except for people who are totally asleep at the wheel.

What companies really need to invest in now is some version of Warren Buffett's Letters to Shareholders and Charlie Munger's Poor Charlie's Almanac. It's circulating now as founder content. Good enough.

I don't care about naming it something cute. The important thing is what it actually does.

Effective founder content publicly demonstrates:

To break it down further, if you're thinking of publishing your own thoughts publicly, here's what those thoughts need to have:

Prediction quality. Solid public forecasts. You don't have to be right, you just have to be both anomalous and rational. You have to have a new good idea that makes sense once identified. You also have to update your predictions visibly when you're wrong. These must be specific, confident predictions with very little hedging. Nobody likes a cowardly founder.

Decision transparency. Explain your reasoning in addition to your conclusions. It's often not about being right but about cogent reasoning. A buyer evaluating whether to trust you with their organization's future wants to evaluate how you think. Give them a solid window into that.

Epistemic honesty. Of course, acknowledge uncertainty. You can make bold claims while acknowledging where your knowledge may be limited such that you have to make an educated guess. Overconfidence can also be disconcerting.

Speed and economy. All else being equal, the team that reaches good decisions faster and cheaper has more capacity to make more decisions and earn small competitive edges as they go along, because they are more likely to be first movers, before markets become saturated with equivalents.

Blog posts, talks, 'handbooks', actual books, letters to shareholders, and documentary films are all valid forms of founder content. It does not matter what form it's in. It just has to hit on each of the above.

Increasingly, this is the kind of research people will do. They'll assume any product worth its salt has all the mission-critical functionality, and then they'll place their bets based on your batting average. But the most discerning buyers are gonna wanna watch your film and see your swing.

With transparent founder content, ideally, you document your thinking honestly such that everyone wins. Your prospect gets a clear view of who they're trusting with XYZ portion of their business, and you select for the types of founders and stakeholders who are likely to be happy with the service/product you provide because they resonate with your decision-making philosophy. The more honest you are and the more you put out there, the bigger your brand — but also, it's a shot at less churn, because people start to 'believe' in you.

Are you a founder? Start doing this ASAP

Here's what you should do:

Stop leading with the product. This is slightly clickbaity — you obviously will talk about the product, but as feature parity increases, it becomes less of a make or break factor for people. Maybe lead equally with the product and your philosophy. There are many ways to do this: the way your site looks (I personally like .txt for this), founder content of course, product design principles, etc. In short: brand. Brand + product should be side by side, or product second to brand.

Build a public track record of specific, falsifiable claims about your domain. Write about where you think the market is going, decisions you've made and why, times you were wrong, etc. When you're right, it's evidence of your edge, and when you're wrong, it's evidence of your epistemic honesty. Also, there are ways to be wrong in the right way. It shows where your optimisms and pessimisms lie, and what makes you bullish vs. bearish. All useful signals.

Make the team visible. These are all the other people pulling the rope. Discerning buyers care about your key people, too. Specifically, your lead eng and product people.

Compete on the quality of your predictions about your own roadmap. Show buyers a historical view of what you said you'd build and what you actually built. Buyers who understand long procurement cycles and the pace of AI development will weight this heavily. Perhaps a ledger of historical decisions.

Avoid gimmicks and use humor sparingly. This depends a lot on how sophisticated your buyer is and what market you're in, but it's a safe tack for companies selling into enterprise. Gimmicks are fun and cute for a moment but attract riff raff and lead to misalignment across the board. It is intellectually dishonest almost by definition. (YES I KNOW RAMP DOES THIS and they're successful but 1. there are exceptions to every rule, 2. I think it's tacky. I am much more moved by the fact that 70+ founders work there. I think the stuff with Kevin from The Office is a low-frequency distraction and likely a flash in the pan that'll inevitably look meh in hindsight, as these types of flash mob shenanigans often do.) Flashiness distracts from ideas. It's like how at Amazon they do documents only to reduce the likelihood fun graphics or good presentations will distract from pure information. In the same way, just give people the information they need about you to make good decisions. Be quiet but right, basically. Not loud and talkative and wrong. Think about the kinds of people in your own life you find to be influential. Be like them at scale.

Also re: humor — humor isn't bad, it's one version of authenticity, and it can be a good sign of what you're like, who you're like, the cultural crowd you fit in with, etc., which are all shorthand for how you think. So it can be useful. But don't do lowest common denominator memes, unless you're B2C or otherwise talking to a huge pool of people. It's just not worth it long term. And you want to prove you are good at long term decision making.

The history of markets is a history of value migrating as things commoditize. Always been this way, not that big a deal that it's happening now.

Re: tech industry is dying — I feel like we're not seeing the forest for the trees. We're forgetting that we're competing on value delivery efficiency: delivering the highest amount of value in the least amount of time with the least amount of effort on the part of the buyer.

Even if we switch fully to build vs. buy, that's fine. Plenty of markets are like that. People hire maids to clean their houses. No need to worry. It happened to hardware, then to operating systems, then to middleware, and now it is happening to the applications layer. It's just that now the value is migrating to the quality of thinking and judgment above the application that makes it productive. Even if AI does almost everything, as long as we are dealing with LLMs as the primary AI-driven OS, people will want a human in the loop for the near future.

IMO the #1 question businesses must focus on answering well for buyers is: how many strong and strategic people are on your team? How do they work together?

I anticipate we'll be seeing more creative, pattern-disrupting ways of answering these questions on the internet in the very near future.