"Great features, not products," leveraging found productivity, and the era where we can no longer believe our eyes.
We must continue to sharpen our perspective on the viability of AI-first companies, and we must be creative about how the excess capacity of humanity will be used in the modern age.
Edition #4 of Implications.
If you’re new, here’s the download on what to expect. This ~monthly analysis is written for founders + investors I work with, and a small group of subscribers.
If you missed the big annual analysis (shared more broadly), or recent private editions of Implications, check out archives here. Last month’s deep dive on the changing mechanics of brand creation generated a lot of feedback...
This edition explores forecasts and implications around: (1) the extent of workflows transformed by AI and the viability of AI-first products as stand-alone companies (what’s investable, what’s not), (2) how all of our newfound time and productivity unlocked by AI will be allocated, and (3) to get away from the AI topic, check out the “Ideas, Missives, & Mentions” at the end.
For companies and for careers, the rise of AI will float fewer boats. But our everyday experiences will be transformed for the better.
I was talking with a well known venture capitalist a few weeks ago who remarked that “80% of the work in 80% of jobs will be completed by AI within the next ten years.” This wild projection has prompted a number of thoughts and questions. Today, so much of every workflow is still “work” beyond the idea or intention of the workflow. Humanity has always just accepted that ideas are ubiquitous and it’s all about the work that follows. As Thomas Edison once quipped, “genius is 1% inspiration, 99% perspiration.” Work is hard, expensive, and risky. Now, as AI changes the equation, will ideas become more important and the work required less of a moat? I believe this is true for every process in an enterprise, and every production process (not coming up with ideas, but the EXECUTION OF IDEAS…people often confuse the two) that needs to be performed, reviewed, and approved, and everything in between. Consider the cost of customer support and success for every company (which can surpass 10% in some cases); what happens when these logic-based workflows and decision trees are all automated with AI? What happens when ideas become the differentiator? What are the implications for society? While companies become more profitable, how do these deflationary pressures impact the economy? And what parts of the economy will remain inefficient and labor-intensive by design? What will people do with all their newfound free time? No doubt, this will be an ongoing exploration but let’s dive into a few thoughts and implications.
The viability of AI-first products as stand-alone companies will rely on data moats, privacy preferences for consumers and enterprises, developer ecosystems, and GTM advantages.
We’re seeing a lot of “great features, not products” emerge in the start-up ecosystem. Certainly with Large Language Models and “text to media”, it feels like everyone is creating interfaces on these APIs with some degree of prompt augmentation underneath and launching them as products. As soon as I shared the “Historical Figures” app in January on Twitter, dozens of other entrepreneurs starting sharing their own version of chats with Gandhi, MLK, or other historical figures…and many of these were standalone apps already submitted to app stores. Most of them monetized with the same “credits” mechanic, where free users get 100 credits and then need to buy more to continue their inquiries. Of course, all of these are built the same way (at least initially), and thrive or die at the whim of their underlying API calls. Same goes for image generation, and so many other features. The ultimate red flags for me are (1) when prompt augmentation is the core differentiator, (2) lack of a network effect (whether developing a self-improving core and proprietary dataset, or in how product grows and retains users) (3) lack of a step-function re-imagination of a core workflow, and (4) failing to apply the traditional values of businesses I believe in - the importance of design and product sensibility, alongside a series of other factors (aka the tech won’t cut it alone)
The apps that do become companies will have a proprietary growing data source, unlock network effects, reimagine a workflow for a particular market of underserved customers, and/or will enable consumers or enterprises to leverage their own data so the model becomes infinitely more personalized and indispensable as you use it. Consider an AI-chat health app that knows everything about you, your family history, your Whoop or Apple Watch data, you daily food intake, and your health goals. This would make for an incredible resource when coupled with the power of leading mega AI models. But you’d never want to expose or turn this data over to a third party. The same goes for enterprise use cases. Companies that nail this have legs as general AI capabilities become commoditized. There are some startups emerging that help private stores of data work with mainstream models at scale and without compromising privacy. Network effects may include style transfer marketplaces, proprietary models or shortcut recommendations or prompt suggestions trained from user preferences. Finally, startups (or new products in big companies) have an opportunity to use AI tech to reimagine a workflow that is both 10x better than current solutions AND requires a different object model for which incumbents cannot be retrofitted. Enterprise VC Tomasz Tunguz summed up the AI startup vs. incumbent ecosystem by suggesting “The most successful startups leveraging generative models will innovate both in the application of the technology & also in its distribution: finding new applications, new customer segments, novel sales & marketing strategies - out of the focus & acumen of the incumbent giants.” The “out of focus and acumen” part is key.
Where will the newfound productivity go (and where not)? More of it will go to entertainment, philosophy, and creative projects than we might think.
Assuming equal access to powerful AI tools, the work we do will become higher order work. Edison's famous quip, "“genius is 1% inspiration, 99% perspiration" was true in a world where the work to pursue ideas was extremely burdensome and expensive. As AI changes the equation, will ideas become more important and the work required less of a moat? The “workflow” to execute ideas and coordinate actions has been the underpinning of organizational design and the core principle of enterprise software for decades. Now, as workflow itself is reduced and optimized by a step-function, we have an opportunity to reimagine what humans do and who we hire to do it. What types of human jobs will become MORE important in the modern organization of tomorrow?
The parts of the economy (and the jobs) that will be least impacted by AI are those in governments, regulated by governments, tradition-driven, or are (shocker) anti-tech. The articulation of this argument by Marc Andreessen really struck me, and he started with the history of price increases and decreases by industry (graph below). His point is that industries like hardware, entertainment, and fashion are transformed by technology because they can be (no restrictions), and such products become more affordable as a result. But when it comes to medical care, housing, and education, regulatory hurdles, traditions (college accreditation), and hesitations resist technological advances. The adoption of AI will be no different, Marc argues. And as a result, we can expect jobs to shift from non-regulated industries to those that are regulated. Is this where we want human capital to be allocated? These are obstacles, but also opportunities given the need to make such services more affordable for all.
As the quantitative is automated, the qualitative is unleashed. As AI does the “thinking,” we can focus more on “feeling”: But lets go back to the more free and less regulated industries that are likely to free up our productivity. What might we be able to focus on more when our typical workflow becomes automated? I have always been fascinated by consumer psychology and product cycles devoted to how an experience makes customers feel (after all, customers are more likely to rave about a product doing what they DIDN’T expect than talk and tweet about what they did expect). However, in classic companies, we lack the time and there are so many low-hanging quantitative opportunities (A/B tests, incremental improvements in copy and pricing and and and) that we seldom have cycles for exploring the philosophy and heart of a product. I foresee a resurgence of intrigue into the softer and more distinguishing elements of brands and products - new ways of telling a brand story, more “small things that make a big difference” impacting how customers feel about a product, and entirely new practices driven by consumer psychology. In such a world, qualitative skills, psychology, hospitality, emotional intelligence (EQ), and other non-scalable talents will not only become more in demand, but a more central part of a company’s secret sauce.
Non-scalable work that builds relationships: At the top of my list are non-scalable relationship-driven actions that move the needle for our products or customers. Consider the impact of an especially thoughtful salesperson on your retail experience. Imaging having a live onboarding session for new services you are trialing. Consider the sensation of getting a handwritten note in a box of goods you purchased from a new direct to consumer brand. Perhaps we’ll spend less time hiding behind screens and orchestrating scalable campaigns (which will be done by AI!), and more time driving word-of-mouth marketing?
Perhaps Einstein was right, “imagination is more important than knowledge.” As knowledge is instantly accessible, tools for imagination and creative skills will become essential. Pardon my bias, but CREATIVE LITERACY is probably 10x more important than knowing calculus or memorizing state capitols in a world where all information - and how the dots connect - is instantly accessible to us. How we educate students - and allocate our human brain power - will fundamentally change. What if our extra capacity of energy and brain power went to creativity - both the creation and consumption sides of the equation? Imagine if everyone knew the art of video editing, took up the art of photography, or felt creatively confident enough to paint or conceive wild prompts for generative creativity tools. I can imagine a world with 1000x more movies, music, live artist exhibitions, and new genres of creativity we can only imagine. Perhaps we are entering “the era of storytelling” - where every brand and person’s story is told in extraordinary ways for the rest of us to listen and watch? We will all have more creative confidence and be outfitted to move people emotionally.