Reflections
1. Software was already becoming more competitive, easier to build. AI isn’t making all AI-native software homogenous, it’s elevating people’s focus on the more subjective aspects of product, like pace of feature launches, viral or irreplaceable distribution, brand, and personalization through data.
- “People asked Ben Affleck, do you think Hollywood’s going to explode because of AI and he said, absolutely not, there’s a craft to making a movie. I still think there’s a craft to building a company, a craft to any experience you’re intimate or passionate about. But I do think there’s a hybrid approach” – Mark
- “Taking the vector of “defensibility goes down because AI is produced”… it’s already pretty easy to build software with companies already in the market, time to value is already shrinking. What makes software great is such a nuanced thing. There’s so many providers out there already, people tend to gravitate to one end or the other. Even if AI is building the software, humans are still purchasing it for now, so maybe there defensibility goes down in the actual coding but goes up on purchase side, customer side, marketing side, go-to-market; there’s still ways to create outsized value outside proper hard writing of the code.” – Joe
2. At the very least, “intelligence” is becoming a commodity, but isn’t that a good thing? It forces people to think creatively and lean into what makes them authentically human. The “easy” path (e.g. copying the same software platform for new xyz vertical) is no longer an option, and people are being held accountable. Individualism and integrity in decision-making is becoming the asset.
- “When intelligence becomes a commodity, you’re all being educated for competence. our parents generation – you wanted to be the top of your class in competence. And you tolerated somebody being an ass hole if they were the best. And now I’m just noticing the tolerance for competence combined with someone being a worse human is RAPIDLY falling, because when intelligence rises and becomes so available, we’re going to want someone who cultivates the human qualities within themselves. So I was talking about to these kids about – how can you become more generous, empathetic, more spiritual, more of the person you’re capable of being as a human, until agents are capable of that.” – Joe
- “I think our stochasticity or our creativity, or our individual investing process is our greatest strength and upside. and getting there in the way we want to get there, and enabling humans to do more work, in the way they want to do it, is probably the greatest upside from AI. So I am very pro human, I think humans have to be in the loop.” – George
3. The future of AI-first software will be fought over “interface loyalty” and emotional connection to a best-in-class experience. Instead of talking about whether software is dead, I find it more interesting to think about how AI has instead been the single greatest driver of the “consumerization of software” that we’ve all been talking about for years.
- “The interface, there’s a lot of loyalty but also tremendous opportunity to innovate. ChatGPT proved that a low fidelity interface with a high fidelity model intelligence can do a lot of damage and can accomplish more than you expected with a rudimentary experience. Even the small visual metaphor of siri going from a small bubble on the bottom to your entire phone, nothing architecturally changed. What Apple is trying to do is build trust with users where, “now we can see your entire phone”. We’re looking at everything you’re looking at at this very moment, we have access to the entire dom. The contrarian take is not owning that capability but incorporating that and moving quickly and not thinking you have it figured out” – Joe
- “I’m a big believer in “interface loyalty” which is a contrarian opinion. It’s the same reason why a lot of us listen to or know the same songs. The barrier to entry to uploading music and listening to any music they like is almost zero, but with humans, you have mimetic trends that take hold in larger portions of pop. And I think really good software is like a really good song, it will be a social experience. It will have natural network effects, and that will be the key differentiator. It’s hard to differentiate between good vs great, but everyone know what great looks like” – George
4. Pricing is both functional and psychological. Inertia and predictability are a powerful tool that people are still willing to pay a premium for, especially in vertical categories. Is the world really ready, or willing, for everything to be priced based on consumption? Some will, some won’t. Above all, the promise of selling the “work” seems to be most promising.
- “The way we think about pricing is: we’ve experimented with consumption, consumption doesn’t work, too much change management, humans have too much friction to get there. We’ve experimented with renting AI employee but agentic platforms aren’t at a place where an AI agent can fully autonomously go complete a task without a human breathing down their neck. And so we’ve ended going back to seat based pricing, because it’s a premium product, and the market responded really well. If agents get as good, we’ll see people renting an employee.” – George
5. Is some of this crazy AI spend and growth experimental? It feels like the answer is yes, but it’s early. I think about trade-offs between customer experience and automation for automation’s sake, and a good rule of thumb is to prioritize customer loyalty.
- “A lot of people are implementing AI for the sake of implementing AI, irrespective of the customer experience. Have to think about why you’re doing it.” – Mark
- “There’s a lot of firms that are experimenting with AI, and you saw massive ARR growth in 2023 and the earlier part of 2024, and it was almost all experimental spend. I have, for the last year, made it our company’s mission, maybe even tagline, to say, “stop experimenting with AI, find a way to get value from it.” – George
Sometimes the best way to prepare for the future is to pay close attention to the past.