Inside a16z's Top 100 AI Apps Report with Olivia Moore

Only 10% of the world's population uses ChatGPT every week, yet it seems that AI is everywhere. The reality is that we are at the beginning, and the Top 100 AI Apps report by a16z with Olivia Moore proves it: the wave of new AI applications is just beginning, but already today the battle to win over the end user is heating up like never before. The central thesis is this: the real race is not between models, but between those who manage to become the platform where users "live" with AI—not only do they use it for spot tasks, but they stay there and bring their digital identity with them. The common way of thinking is that ChatGPT, Gemini, and Claude are direct competitors, stealing users from each other. Instead, the report shows that ChatGPT dominates overwhelmingly: 2.7 times more web users than Gemini, 30 times more than Claude, and an even greater mobile advantage. But the real news is that platforms like Gemini and Claude are now starting to carve out specific niches: Claude focuses on premium tools for researchers and scientists, Gemini grows thanks to creative tools like Nano Banana, and ChatGPT moves like Google, aiming to become "AI for everyone." The difference lies in the monetization strategy and the apps they enable: Claude thrives on subscriptions and professional plugins, while ChatGPT is preparing to monetize with advertising and transactions, becoming the gateway to everything from travel to personal finance. This is where the theme of "compounding advantages" comes into play—advantages that grow over time: more data, more memory, more relationships and interactions between users. Sam Altman puts it clearly: “We want ChatGPT to be AI for everyone.” And with 900 million subscribers, the strategy is cleverly designed: if you bring your identity and memory to ChatGPT, every new product can immediately adapt to you, without starting from scratch. Olivia Moore says that the idea of using ChatGPT as a universal login—bringing your memory with you—could change the rules: if all your friends are already there, changing platforms becomes almost impossible. But there is a real-life detail that revolutionizes the discussion: many companies and users do not want to mix work and private life, and segmenting the AI's memory between different personal identities will be one of the great challenges of the coming years. Looking outside the United States, huge surprises emerge: in Russia and China, Western products are almost absent, replaced by local solutions such as Giga Chat, Yandex, Dao Bow, and DeepSeek, with usage rates of ChatGPT and Gemini below 15%. In contrast, in Singapore, Hong Kong, and the United Arab Emirates, the per capita adoption of AI is very high—well above the United States, which is only twentieth in the rankings. This depends not only on technology, but on culture: in China, 80% have positive opinions about AI, while in the US, only 32% do. Cultural trust accelerates adoption; distrust slows it down. On the creative apps front, the story is turning around: in the beginning, AI was mainly used to generate images and memes, but now the main models are already doing these things well. There are only a few real stars left, and they work on more sophisticated workflows and unique styles: Midjourney, Ideogram, Suno for music, 11Labs for voice. Video is the new battleground: Chinese models like Cance 2 are ahead of the Americans, thanks to the freedom to use large amounts of training data. And then there are intelligent agents like OpenClaw and Manis: OpenClaw has surpassed React and Linux as the most starred project on GitHub, but it remains confined to technical users. Manis, on the other hand, after a more than $2 billion acquisition by Meta, was the first agent that was truly usable by the average consumer—capable of creating slides, browsing the web, and managing email without errors. A practical detail: many of the AI apps that generate the most revenue today are not accessible from a browser or mobile, but only from dedicated desktop apps such as Cursor, Granola, or Cloud Co-Work. This makes it difficult to measure success with just web and mobile data, and shows that the future of AI will be within increasingly invisible tools and environments. But the real game-changer is the way personal memory is becoming the heart of the AI experience. Olivia Moore says: "In two years, if a new app doesn't already know everything about you as soon as you open it, it will seem broken to you." The onboarding phase as we know it today will no longer exist: total personalization will be the standard, and those who don't offer it will be left behind. If this idea concerns you, on Lara Notes you can press I'm In — it's not a like, it's your way of saying: from today, this perspective is part of how you think. And if tomorrow you find yourself telling someone that in Russia and China they use AI completely differently from us, on Lara Notes you can mark the conversation with Shared Offline: because discoveries that make you change your mind deserve to be remembered. This Note comes from the a16z report and podcast: it just saved you 35 minutes.