AI products are not all created equal.
Many AI services struggle with user engagement and fail to retain their users. For instance, ChatGPT, the fastest consumer product to reach 100 million users, only manages to retain 56% of those users after a month. This pales in comparison to YouTube, which retains 85% of its users.
Other AI applications struggle to keep users engaged. DAU/MAU ratio for AI apps is about 14% which means that only a small percentage of users use the AI app on a daily basis. In contrast, popular consumer apps have a much higher ratio at around 60-65%. This suggests that current AI products do not provide enough value to encourage users to use them every day.
Some AI apps are seen as "thin wrappers". If an AI app only enhances the existing functionality of a foundational model without offering anything beyond a shiny user interface, it can be classified as a thin wrapper. These apps may see a spike in usage due to initial curiosity, but their popularity quickly wanes over time.
To revolutionize industries and improve productivity, companies should focus on leveraging AI to improve existing workflows rather than merely building wrappers around existing technology.
Companies like Uber and Stripe exemplify this approach. Uber transformed the transportation industry by seamlessly connecting riders and drivers, streamlining payments, and providing real-time GPS tracking. Stripe has revolutionized online payments by offering a developer-friendly platform for businesses to accept payments and manage their financial transactions. It provides tools for subscription billing, fraud prevention, and international payments, making it a comprehensive solution for businesses handling online transactions.
Neither of these are thought of as "thin wrappers" around a digital map, GPS location, or online payments.
The distinction between workflows, wrappers, and technology showcases is crucial. AI-first products that go beyond shiny but shallow solutions have the potential to revolutionize industries and transform how we work.
The key to success is building AI-first products that solve real problems and integrate seamlessly into existing workflows. These products should be intuitive, user-friendly, and significantly enhance productivity without burdening the user with prompt engineering and other technical complexities.
It's not about the technology, but it's about what the technology enables
To improve workplace productivity, we need to understand how work is done and identify which workflows can be enhanced or reinvented by AI. By focusing on workflows, AI-first products can improve usability, efficiency, and effectiveness, making them indispensable tools for users.
Moving away from chat-based AI interfaces is an important first step. Even though chat interfaces have become popular (primarily because of ChatGPT), they are not great for workplace productivity tools.
Unlike most popular productivity tools, chat interfaces lack the structured user experience that people are accustomed to and are limited in how they can integrate with existing processes.
This is particularly important when considering the typical activities of knowledge workers, such as analyzing and planning (30% of their time), creating content (40% of their time), and collaborating with others (the other 30%). A document-centric view in most of these cases is more suited to get work done.
Looking back 100 years, we can see how technological disruptions have transformed industries. A century ago, the invention of machinery like tractors and turbines revolutionized the agricultural sector, leading to increased productivity, reduced labor requirements, and overall efficiency.
How will you make the most of your newfound time?
Similarly, AI-first products have the potential to bring about significant changes. As AI automation takes over mundane tasks, users will have more time at their disposal.