Now Go Faster

Chapter 8: When Clarity Becomes Cheaper Than Pretence

AI is categorically different from previous technology waves because it collapses the cost of synthesis, making clarity cheap and pretence expensive for the first time. The gap between organisations that adapt and those that resist compounds each quarter.

AI reveals one set of truths and amplifies one set of risks. The question that determines which force prevails is economic.

The split between software's growing centrality and unchanged power structures survived three waves of technological change, each deepening the centrality without altering the structures. The internet was a channel you outsourced: software became customer-facing at scale, but you procured it from an agency rather than building it, so no structural change was required. Mobile and cloud differed only in degree. Companies hired CTOs, built engineering organisations, and launched digital transformation programmes that consumed hundreds of billions of dollars globally, but grafted these onto existing governance, where annual budgets still ran on capital-project cycles and “the business” still decided what to build while engineering “delivered.” Most of these efforts disappointed, and the documented failures share a signature: the structures governing the technology had not changed. Chapter 10 examines the pattern at documented scale.

AI's reader capability is categorically different. It cannot be fully absorbed as another workflow layer, because its purpose is to test the layers against one another. By collapsing the cost of synthesis, it makes clarity cheap and pretence expensive.

Chapter 7 established the mechanism: bounded reconciliation that once required weeks can now be repeated in hours. The economic consequence is not merely cheaper analysis, but a change in what organisations can afford to leave unreconciled.

Cheap synthesis only has structural force when someone is accountable for the result; where ownership remains diffuse, AI produces insight without consequence.

The industry's own measurement programme has arrived at the same conditionality from survey data. DORA's 2024 report measured delivery stability falling where AI had been adopted; its 2025 report, across nearly five thousand respondents, found throughput improving while instability persisted, and framed the overall result as amplification: AI made strong delivery systems faster and weak ones less stable, each at higher volume. The variable that decides which way the amplification runs is the quality of the structure underneath the tooling, the same condition cheap synthesis depends on, now measured across thousands of teams instead of argued from one.

When clarity was expensive, organisations optimised for impression management: rewarding ambiguity and valuing the appearance of understanding over understanding itself. When clarity is cheap, these behaviours become liabilities. Four consequences follow.

Empty documents become visible. A strategy with no falsifiable commitment and no anchor in the system used to be safe, because the only thing that could expose it was a quarter-long review nobody would fund. Now the review costs an afternoon, and the forty-page deck that survived a decade of offsites turns out to commit to nothing, which everyone had agreed not to mention.

...

See also: Full contents · Preview chapters · Illusions in the Boardroom