Exceptional people in flawed systems will produce subpar results. Before focusing on individual performance, leaders must ensure the underlying systems are reliable and resilient. As shown by the Southwest Airlines software meltdown, blaming employees for systemic failures masks the root cause and prevents meaningful improvement.

Unlike private sector products that target specific demographics, government digital services must cater to an extremely diverse user base, including people with low income, no permanent address, and vast age differences. This necessitates a rigorous, non-assumptive approach to user research and accessibility from the outset.

Treat government programs as experiments. Define success metrics upfront and set a firm deadline. If the program fails to achieve its stated goals by that date, it should be automatically disbanded rather than being given more funding. This enforces accountability.

Unlike private companies seeking product-market fit within a specific segment, designing digital public infrastructure (DPI) requires a different mindset. The goal is creating a level playing field that enables *everyone* to participate and allows markets to innovate on top.

In siloed government environments, pushing for change fails. The effective strategy is to involve agency leaders directly in the process. By presenting data, establishing a common goal (serving the citizen), and giving them a voice in what gets built, they transition from roadblocks to champions.

When executing a large-scale government technology project, physically separating the technology team (e.g., in Bangalore) from the political and bureaucratic hub (e.g., in Delhi) is a crucial tactic. This creates a shield, allowing engineers to focus on deep work without constant interference.

To navigate the high stakes of public sector AI, classify initiatives into low, medium, and high risk. Begin with 'low-hanging fruit' like automating internal backend processes that don't directly face the public. This builds momentum and internal trust before tackling high-risk, citizen-facing applications.

In environments with highly interconnected and fragile systems, simple prioritization frameworks like RICE are inadequate. A feature's priority must be assessed by its ripple effect across the entire value chain, where a seemingly minor internal fix can be the highest leverage point for the end user.

A project's success equals its technical quality multiplied by team acceptance. Technologists often fail by engineering perfect solutions that nobody buys into or owns. An 80%-correct solution fiercely defended by the team will always outperform a "perfect" one that is ignored.