In an Age of Abundance, What Should We Build?

Lena Hall observes the unprecedented speed and leverage in current software development, leading to an overwhelming abundance of output. Despite these advancements, engineers face intense pressure, often working extended hours just to remain competitive. The rapid ability to build almost anything quickly has driven the value of an 'average' product to nearly zero, creating a new kind of market anxiety.

AI models generate similar answers because they operate on historical data, reflecting past occurrences and common knowledge. When tasked with developing products based on 'what users want,' AI often produces conventional solutions. This technological characteristic makes AI a convergence machine, leading to identical outputs unless guided by a distinctive human vision.

AI excels at implementing instructions but cannot autonomously decide what to build, making the human role of 'pointing' the AI critical. This involves defining the vision and objectives for AI development. Success in this new paradigm means becoming the reason why specific users choose one particular product over a sea of similar alternatives.

Lena Hall introduces the concept of a 'signal layer' as a solution to differentiate products in a crowded market. This layer comprises two main components: defining the signal and emitting it without distortion. Defining the signal involves the core aspects of the product, including its code, overall design, and strategic roadmap. Emitting the signal, on the other hand, focuses on how the product communicates its unique value through content and go-to-market engineering strategies.

Autonomous coding agents have significantly advanced in capability over the past two years, moving from solving a fraction of tasks to achieving high scores on standard software benchmarks. This progress indicates that tasks quantifiable by benchmarks are rapidly becoming automated. Consequently, any software implementation that can be verified by a compiler or test suite is effectively becoming a commodity, free to implement without human intervention.

Lena Hall advises following Paul Graham's guidance: discover genuine market needs by experiencing them personally. She argues that surveys often fail to identify new markets because these markets haven't fully formed yet. Therefore, personal need becomes the most reliable 'signal' that isn't misleading. This approach encourages building solutions for problems one directly faces or those shared by close contacts, rather than relying on aggregated data that reflects existing trends.