We're running faster on the treadmill, but aren't moving forward. In fact, we might be moving backwards as we create more systems no one fully understands and therefore can't diagnose or improve.
AI models can now write entire software projects with minimal intervention from the user. I myself spent December "building" and iterating on a retirement planner tool. A related phenomenon is the use of OpenClaw, a free and open source tool that allows users to orchestrate multiple agentic AI instances using messaging platforms like Slack. The podcast "How I AI" includes a number of interesting examples that seem to point to these agents being capable of autonomously doing economically valuable work. One person details how she uses 5 agents to run her homeschool, finances and coding projects, while a senior developer shares how these agents have allowed him get oriented more quickly when tackling unfamiliar codebases.
That extrapolation might be a mistake though. As Cal Newport points out: code is different from most other knowledge work. There are millions and millions of lines of code online, including code snippets surrounded by explanations on Stack Overflow. So, yes, software developers can "write" code faster, but that was never the real bottleneck. For any company, the real bottleneck has always been and continues to be building products and offering services that enough people are willing to pay for. Figuring out what problems are worth solving, positioning effective solutions, and prioritizing iterative improvements continues to be the biggest challenge facing companies.
Unique insights, judgement, and taste are still the enablers of successful businesses. A firehose of code with new features deployed daily has never driven market success: it's never positioned a product or built a sustainable business. We're creating more and more noise, and the signal is getting fainter and fainter.
Effective leaders who create clarity are needed more than ever to light a path through the cacophony. The leaders who figure this out will have a significant advantage. That's what the next post is about.