I had an opportunity to speak at SaaStr in the Bootstrap track. Here is an overview of the talk I gave.
Thinks 986
Bloomberg on China: “Rising joblessness means less income for young people and reduced spending on goods like mobile phones or entertainment and travel, which curbs economic output. While it’s difficult to quantify, research suggests young people in China are an important driver of overall consumption in the economy, and are big luxury spenders too. That said, the 16-24 age group makes up about 7% of China’s urban labor force, according to Citigroup Inc., so is not as big of a driver of consumer spending as older groups. High unemployment affects confidence in the economy and could weaken productivity if it’s prolonged. In China, it’s also driving social disaffection among young people, prompting many to drop out of the rat race completely in a phenomenon known as “lying flat.” And it risks stoking social instability if young people become more angry and frustrated about their lack of opportunities.”
Rauno Freiberg: “Design can feel like there’s no science to it—only feel and intuition. Even researchers have trouble grounding interaction design practices in science, inherently treating them as a mysterious black box. While from my own experience that’s partly true, I have been trying to deconstruct and excavate the why behind great displays of interaction design. The essence of the word “interaction” implies a relationship between a human and an environment. In my experience, great revelations surface from making something—filling your headspace with a problem—and then going for a synthesizing daydreaming walk to stir the pot.
WSJ: “The world is undertaking two great technological transitions—toward greener energy production and toward more AI-enhanced decision-making. The two are intimately linked. Machine learning has long helped to analyze and improve engineering, such as finding more-efficient battery designs, as well as orchestrating and improving energy delivery. Now cutting-edge AI is poised to take that collaboration a step further. AI can analyze data, find patterns in it and then concoct solutions to problems faster than any human researcher could. But it isn’t a straight path to sustainability. Ever more-complex AI models consume ever more energy to run, for instance—eliminating some of the climate-friendly gains AI produces…Building better batteries. Improving policy-making. Matching supply and demand more accurately. Artificial intelligence may make it easier to arrive at a greener future.”
Interview with Rethink Retail
Kirat Anand of RETHINK Retail interviewed me at eTail Boston.
Thinks 985
NYTimes: “By 2050, people age 65 and older will make up nearly 40 percent of the population in some parts of East Asia and Europe. That’s almost twice the share of older adults in Florida, America’s retirement capital. Extraordinary numbers of retirees will be dependent on a shrinking number of working-age people to support them. In all of recorded history, no country has ever been as old as these nations are expected to get. As a result, experts predict, things many wealthier countries take for granted — like pensions, retirement ages and strict immigration policies — will need overhauls to be sustainable. And today’s wealthier countries will almost inevitably make up a smaller share of global G.D.P., economists say. This is a sea change for Europe, the United States, China and other top economies, which have had some of the most working-age people in the world, adjusted for their populations. Their large work forces have helped to drive their economic growth. Those countries are already aging off the list. Soon, the best-balanced work forces will mostly be in South and Southeast Asia, Africa and the Middle East, according to U.N. projections. The shift could reshape economic growth and geopolitical power balances, experts say.”
FT: “Prepare for a Napoleonic world, then. The most important governmental trend today is the rise of protectionism. In the US, Europe, China and India, the state is turning from open trade to the cultivation of domestic industries. One justification is strategic: don’t count on frail or hostile regimes for essential goods. Another is progressive: give skilled manual labour a break for once. Both trace back to the election-winning arguments of Donald Trump in 2016. And so we have something of an irony to chew on. Populism, which sets itself against the elite, against the “deep state”, is going to leave it more powerful, not less. The technocrat, vilified so recently, will be the string-pulling figure of our age, dispensing subsidies, guiding this economic sector, shunning that one. Corporate leaders will have an ever tighter and more collusive relationship with government, not as a corrupt byproduct of the system but as a central feature of it. Populism was meant to take the governing class down a peg or two. Its main legacy will be something close to the opposite.”
The Generalist: “Stablecoins are more than just an asset – they’re the foundation for a modernized financial system…These digital currencies maintain a fixed price relative to a reference asset – typically the US dollar. Unlike much of the crypto landscape, stablecoins offer no dramatic run-ups and speculative gains. They are designed to maintain their value, to act like a dollar in digital form. …For people around the world, from Buenos Aires to Lagos, stablecoins offer an easy way to access a dollar equivalent. When your domestic currency rapidly fluctuates in value, diminishing your spending power, having access to an asset that retains its value is a game-changer. It also has the potential to be a faster, easier, and cheaper way of sending cross-border payments like remittances. Stablecoins also have utility for businesses. Though we’re earlier into exploring these use cases, companies like JPMorgan, PayPal, and Stripe are experimenting with the technology. In the coming years, more big-name players may enter the fold to take advantage of the technology’s native benefits.”
Andy Kessler: “From Musk and Zuckerberg to Gates and Jobs, rivalry builds the future…Why so many pairs of competitors? Mostly because we only remember the winners and runners-up…Let’s face it, we all love a good race: Red Bull’s Max Verstappen and Mercedes’ Lewis Hamilton in Formula One. Jonas Vingegaard and Tadej Pogacar and their teams in the Tour de France. It’s the same with new technology. Stasis equals death. Alexa battles Siri. Facebook Reels drafts TikTok. OpenAI brought out its generative AI tool ChatGPT before it was ready for prime time because Google was close to launching Bard. Meta’s LLaMA is behind but gaining. A high-stakes race for sentient computing and artificial general intelligence is on. I’m skeptical we’ll fully get there, but we’ll all enjoy and benefit from the many cage matches along the way.”
Interview with The Signal
I spoke to Dinesh Narayanan, co-founder and editor of The Signal. An excerpt:
Dinesh: Rajesh, what made you write Startup To Proficorn?
Rajesh: So, Dinesh, conventional wisdom is that the only way to really build a business is to go out and raise venture capital. Venture capital, angel, money, whatever it is. The reality today is that 99.9% of businesses are self-funded. You know, you look at the kirana store, you look at very small startups, they’re not raising capital.
The challenge for a lot of these bootstrapped startups, which are private, is essentially: how do they scale? And that’s the problem I try to address in the book, how can you build what I call a proficorn. So it’s private, it is bootstrapped, it is profitable, and most importantly, it’s highly valuable. And I define “highly valuable” as a valuation in excess of $100 million, because if you look at a unicorn, which is typically $1 billion, founders end up owning 10% of the company.
And if you do that with a proficon, which is valued at $100 million, it’s pretty much the same outcome for the founder. That’s the core message, that you can actually have an alternate approach to building the business where the founder is in greater control, there’s a lot of freedom. And you can build valuable large businesses. I’ve done it twice.
**
The full interview is available on Spotify, Apple Podcasts, Amazon Music, Google Podcasts.
Thinks 984
Economist: “The most significant benefits from new forms of ai will come when firms entirely reorganise themselves around the new technology; by adapting ai models for in-house data, for example. That will take time, money and, crucially, a competitive drive. Gathering data is tiresome and running the best models expensive—a single complex query on the latest version of Chatgpt can cost $1-2. Run 20 in an hour and you have passed the median hourly American wage. These costs will fall, but it could take years for the technology to become sufficiently cheap for mass deployment. Bosses, worried about privacy and security, regularly tell The Economist that they are unwilling to send their data to modify models that live elsewhere. Surveys of small businesses are not encouraging. One, by GoDaddy, a web-hosting company, suggests that around 40% of those in America are uninterested in ai tools. The technology is undoubtedly revolutionary. But are businesses ready for a revolution?”
FT: “Once upon a time, people joined social media networks so they could connect with one another. I signed up to Facebook in 2007 to see what my friends were up to online. It’s hard to remember why it was so interesting to look at lots of blurry photos of a night out, but I spent a lot of time doing it. That has now been superseded by content from strangers. I still have all my social media accounts but I rarely post anything. For many of us, the point of TikTok, Instagram, YouTube and Twitter is not to upload our own posts or look at what our friends are doing but to watch a small number of popular creators. Instead of talking to one another, we have become mostly silent onlookers. This is the result of the TikTok-ification of social media. On TikTok, videos are not designed to connect existing contacts. They are content consumed by the biggest crowd possible…What this all means is that social media companies are no longer reliant on the network effect of real-world relationships that made Facebook so compelling in the first place. Who cares if your friends haven’t joined a particular social network? They are not the ones whose content you’re interested in anyway.” NYTimes: “Think of the current large social networks as various European nations at the dawn of the 20th century. Often ruled by monarchs and autocrats (C.E.O.s), they exist in an uneasy balance with their own users and with one another. Users and social networks have an unspoken agreement: In return for entertainment, utility and an audience, users hand over control. If the network chooses to kick you out, you’re out in the cold. Choosing to leave one platform means losing your audience forever. You can’t take it with you.”
Thomas Sowell: “No matter how disastrously some policy has turned out, anyone who criticizes it can expect to hear: “But what would you replace it with?” When you put out a fire, what do you replace it with?” [via CafeHayek]
Two interesting reports: OpenAI and GPT-4.
Heather Heying (in 2001): “Education should not serve primarily to make the student feel good about herself in the moment. Education is about enriching the lives of students so that they may live informed, enlightened lives in which they have the curiosity to ask “why?”, the knowledge to ask “are you sure?”, and the courage to ask “is this right and good?”.” [via Arnold Kling]
iDarpan: How Mirror Worlds and Digital Twins will Revolutionise eCommerce (Part 13)
Martech 2.0
iDarpan is the latest in a series of innovative marketing concepts that I’ve discussed over recent years. The overarching theme, which I refer to as Martech 2.0, challenges the current imbalance in ad budgets, which are predominantly allocated towards new customer acquisition. This practice often results in AdWaste and subsequent strain on brand profits. Martech 2.0 seeks to answer a constant query among marketers: “How can I reduce the Customer Acquisition Cost (CAC)?” The solution isn’t merely to optimise the ROAS (Return On Ad Spend) on prominent adtech platforms. Instead, it advocates a different approach – —concentrating on harnessing the spending power of existing customers.
Martech 2.0 revolves around increasing sales while reducing marketing costs, thereby enhancing profitability. It is about bringing back existing customers for more, and ensuring they get their friends. It is about laying the groundwork for exponential forever profitable growth, and eventually, a profipoly, by building remarkable products that function as profit generators, enabling brands to establish a competitive edge and ultimately, a profit monopoly (or ‘profipoly’).
Here is a summary of these concepts.
ProfitXL (PxL): The ultimate goal is to revolutionise the Profit & Loss statement by substantially increasing profits. This transformation utilises the SHUVAM framework, comprising Storytelling, Hotlines, Unistack, Velvet Rope Marketing, Acquisitions, and Metrics. It includes advancements like Email 2.0 (utilising AMP for interactive emails) and Loyalty 2.0 (providing Atomic Rewards – micro-incentives – for customer attention and data). Velvet Rope Marketing prioritises the top 20% of customers, who contribute to 60% of the revenue and can yield 200% of profits, acknowledging the long tail phenomenon’s inefficiency.
Inbox Commerce: Minimising the ‘funnel friction’ can be achieved by bringing conversion actions closer to the customer—directly in their inbox. Tactics such as Email Shops, Reactivation Sequences, and Engaging Footers can eradicate three profit pitfalls: a vast majority of email recipients not clicking through to the website, a large fraction of clickers not making a purchase, and a significant portion of email IDs in the database remaining disengaged.
Adtech-Style Martech: By adopting a performance pricing model, martech companies can rectify their past mistakes and benefit from unlimited budgets. This adjustment necessitates a ‘progency’ mindset, merging the product’s power with the creative approach of an agency. This strategy could significantly mitigate the $200 billion AdWaste faced by B2C/D2C companies, providing a sustainable profitable growth model for brands.
iDarpan: The focal point of this series, iDarpan leverages the concepts of mirror worlds, digital twins, and the technologies driving the Metaverse and Generative AI. It helps brands beautify every profit killing customer experience—delighting both the customer and the eCommerce manager. Central to iDarpan is the Large Customer Model (LCM), which predicts and navigates customers on their journey to satisfaction.
Martech 2.0 is about bringing these new ideas to fruition. As Peter Drucker once said, “There are only two things in a business that make money – innovation and marketing, everything else is cost.” Here is my adaptation: “There is only one thing in a business that makes a money machine – innovation in marketing.”
Thinks 983
FT: “The book is one of the most treasured and defining products of civilisation. From priceless tomes of learning in the great libraries of the world to battered, much-loved paperbacks in beach bags or on our shelves, books are a familiar, vital and reassuring constant in a world abounding with distractions. But what actually is a book? Since the Epic of Gilgamesh was first written down on clay tablets in Mesopotamia 4,000 years ago, it has been a mobile and ever-evolving technology. Papyrus scrolls gave way to the parchment codex; then, after Johannes Gutenberg’s invention of moveable type in the 15th century, transformed into the printed volume all of us are familiar with today. That evolution continues. For decades now, books have also been found as electronic texts on e-readers and in multiple forms online. And yet technological expansion brings threats — most recently with the emergence of artificial intelligence, which leads some to sense that the age of the book, epitomised in the west by Gutenberg’s great advance in printing, is now in its death throes.”
Eric Newcomer: “There are a number of big challenges facing the startup and venture capital ecosystem. Few startups are going public and when they do they’re at smaller valuations. The median tech IPO has fallen to $69.5 million this year. The IPO slowdown has cut off the most important source of exits for startups. And remember that we had a backlog of IPO candidates. There are more than 1,000 unicorns. Investors had hoped that many would go public one day. Stripe, Instacart, Reddit, and many more remain private. Valuations are down and deal volume is down. Late-stage funding rounds are rare and when they’re happening (outside of generative AI) they’re often shoring up troubled late-stage startups. New venture capital firms are going to have trouble raising more money. Private equity firms and crossover firms seem to be flaking on growth stage rounds. The ultimate problem here is that many startups are just not as valuable as investors once thought they were.”
Christopher Penn on AI: “The bigger and more capable models get, the more tasks they can handle. Every time we have a big leap forward in model capabilities, that opens the door for us to hand off more tasks to AI. Does your book draft need a sensitivity reader or a first-pass editor? Feed it to a model with a suitably large context window and have it do the initial work. Do you want to rewrite a work of fiction you wrote in one universe to another universe? The largest models can handle that task. Do you want to write thousands of lines of code? Also doable. In fact, GPT-4’s Code Interpreter…is absolutely mind-melting in how good it is. What we – and by we, I mean most AI practitioners – have been saying for quite some time now is that AI isn’t going to take your job, but a person skilled with AI will take the job of a person who isn’t skilled with AI. That’s… sort of true. Again, there’s nuance. There are some jobs, some content creation jobs, where AI will absolutely take that job if it’s valuable enough to do so.”
Anticipating the Unintended: “Having a fab in India will not make India self-sufficient in chips. A fab in India will perhaps produce 40,000 wafer starts per month (WSPM). The top foundries produce more than a million-and-half WSPM. India’s semiconductor market is far more diverse and huge for two-three fabs to handle. Nevertheless, a fab is important as it will begin a process of learning chip manufacturing. It will require a few “knowledge decades” before Indian firms can master this supply chain segment. It’s like planting a seed of a Banyan tree—it will reach full maturity only decades from now.”
iDarpan: How Mirror Worlds and Digital Twins will Revolutionise eCommerce (Part 12)
Large Customer Models
One concept that needs some elaboration is that of the relationship between Large Language Models (LLMs) and Large Customer Models (LCMs). The key to this analogy lies in the predictive capacity of these models, where LLMs anticipate the next words in a text sequence and LCMs predict the next actions in a customer journey.
The transformative potential of Large Language Models (LLMs) and Large Customer Models (LCMs) pivots around their predictive capacity. LLMs like GPT-4 excel in generating human-like text based on prior sequences of words. By analysing extensive text data, these models learn the probability of particular words following a given string of text, effectively predicting what comes next. Companies like OpenAI leverage LLMs for various purposes, including drafting emails, writing code, creating written content, and more. The power of these models stems from their ability to capture context, recognise patterns, and make educated predictions about subsequent textual sequences.
LCMs, in parallel, function on similar principles but operate in a different ‘language’—the language of customer behaviour. They predict the next action in a customer’s journey by analysing extensive datasets of customer interactions and behaviours. The aim is to recognise patterns and sequences leading to specific outcomes. A prime example would be a good recommendation engine, which predicts what product a customer is likely to be interested in next, based on their browsing history and past purchases.
However, LCMs surpass LLMs in their ‘vocabulary’ breadth. They not only consider individual actions, like browsing a product or making a purchase, but also can consider a wide array of external factors. For instance, factors like the time of day, weather conditions, or significant events like holidays and festivals can influence customer behaviours.
What makes LCMs truly powerful is their ability to fuse learnings from vast datasets with real-time behavioural information. They capture the ‘customer language’ at an individual level, learning from each interaction to refine the predictive models. They anticipate behavioural changes, offering marketers the opportunity to adjust strategies proactively.
LCMs are inherently designed for continuous learning and improvement. They allow eCommerce businesses to dynamically optimise customer interactions based on real-time behavioural changes. As the LCM evolves and adapts over time, it refines its predictions, making every subsequent customer interaction more personalised and relevant.
In the fiercely competitive eCommerce arena, where customer loyalty is a prize to be won, LCMs can be game changers. When integrated within systems like iDarpan, they facilitate true one-to-one personalisation. By accurately predicting the next steps in a customer’s journey, brands can create highly customised experiences that resonate with customers, fostering loyalty and driving repeat purchases. In the long run, LCMs can enable brands to carve out a ‘profits monopoly’ (profipoly). As they mature, brands can maximise profitability by effectively nurturing their customer base.
With iDarpan, LCMs can power an eCommerce revolution, driving customer engagement at an individual level and shaping marketing strategies through predictive customer behaviour. This transformative approach can transform the future of eCommerce, mirroring the customer to reflect their needs, preferences, and behaviours, crafting a shopping experience that feels uniquely their own.
Thinks 982
Law Liberty: “Markets work, so why don’t people like them? That’s the basic question Rainer Zitelmann tries to answer in his book In Defense of Capitalism. The book is really two books stapled together. First is a systematic demolition of many common arguments against capitalism. This portion is delightfully fun to read. Zitelmann’s writing style advances his arguments quickly and concisely, with plentiful references to other sources backing up his claims (the book has over 800 footnotes). Each of the first ten chapters presents a common anti-capitalist argument followed by an empirical record of whether it is true. Zitelmann destroys claims such as “capitalism promotes selfishness and greed” and “capitalism entices people to buy products they don’t need.” He does not do so from a theoretical economic point of view. Trained as a historian, Zitelmann relies on the record of the past to adjudicate the truthfulness of anti-capitalist claims…The second half of the book is completely different from the first. It’s about survey research that Zitelmann commissioned for the book about people’s attitudes toward capitalism.”
DeepMind CEO Demis Hassabis: “I think, in chess, the problem-solving and strategizing, I find it a very useful framework for many things and decision-making. Chess is basically decision-making under pressure with an opponent, and it’s very complex, and I think it’s a great thing. I advocate it being taught at school, part of the school curriculum, because I think it’s a really fantastic training ground for problem-solving and decision-making. But then, I think actually the overarching approach is more of the scientific method. So I think all my training is doing my PhDs and postdocs and so on, obviously I did it in neuroscience, so I was learning about the brain, but it also taught me how to do rigorous hypothesis testing and hypothesis generation and then update based on empirical evidence. The whole scientific method as well as the chess planning, both can be translated into the business domain. You have to be smart about how to translate that, you can’t be academic about these things. And often, in the real world, in business, there’s a lot of uncertainty and hidden information that you don’t know. So, in chess, obviously all the information’s there for you on the board. You can’t just directly translate those skills, but I think, in the background, they can be very helpful if applied in the right way.”
Aatish Taseer: “If nothing is done to make air-conditioning more energy-efficient, India alone is projected to use 30 times more electricity in 2030 than it did in 2010. Globally, air conditioning is projected to account for 40% of the growth in energy consumption in buildings by 2050—the equivalent of all the electricity used today in the U.S. and Germany combined. It’s enough to send a chill down the spine of the most ardent of AC evangelists. The irony of a world made hotter by our need to be cool strikes some as proof of our rapacity. To me, having grown up in the place where so much of the new demand is coming from, I see it as part of a necessary realignment. As the global south gets richer, it will act as a frontier and laboratory. My hope is that it will achieve a miraculous breakthrough in energy efficiency, even as it asserts an inalienable right to sit in AC.”
Andy Matuschak on memorization: “We could use that word to refer to the practice of learning more trivia. For instance, a thing that I and some people I know have done is, we’ve gone through a book called Cell Biology by the Numbers, which says all of these things like, how big exactly is a nucleotide? Like how much volume does it take up? It’s kind of helpful occasionally to know that it’s about a nanoliter. And that can help you model things. So you can just commit all of those things to memory, right? That’s one kind of memorization. And we could talk about how LLMs affect that. But I just want to make the case that so much of what you do and experience day to day is memory bound, or is memory influenced in important ways. For instance, your ability to understand a difficult argument, even in the course of a text, is memory bound. Some of that’s working memory. But your ability to understand an argument that has many steps in it, more steps than you can keep in your working memory, depends on your ability to think of some of those steps in terms of some stuff that you already know, so that you can kind of reduce it or abstract it.”