Shift focus from viewing AI as a tool for individual or team productivity to a force for fundamental structural change in how work is organized and executed across the entire company. This requires a mindset that moves beyond incremental improvements.
Engineering is the ideal starting point for 'self-driving' initiatives not just due to tech-savviness, but because software development has clear, verifiable outcomes (e.g., a bug fix works or it doesn't). This contrasts with subjective domains like marketing, making it a fertile ground for initial experiments.
Implementing agentic workflows doesn't eliminate challenges; it transforms them. For example, tripling code output creates a new code review bottleneck. The goal is to solve these emergent problems in service of a clearly superior way of operating, rather than expecting a problem-free state.
To spread AI use beyond engineering, use a 'pull' rather than 'push' strategy. By having engineers interact with AI agents in public forums like Slack, other departments organically see the benefits and processes, overcoming skepticism and encouraging participation without a top-down mandate.
An internal agent, deeply integrated with proprietary company data and systems, can become superior to expensive, market-leading SaaS products. Replit cancelled a seven-figure contract because their custom, integrated solution was better and more adopted by employees, fundamentally changing the 'build vs. buy' calculation.
The ultimate goal of a self-driving company is not just automating internal tasks. It's creating a continuous learning system where AI agents analyze user feedback, propose product improvements, and use A/B tests to validate them, closing the loop between the user and the product for autonomous evolution.