Intelligence Track
4 July 2026
The Brief
Two accounts of the same disappointment arrive together: a wave of large firms discovering they blew annual AI budgets in months while fewer than one in a hundred can show real returns, and a studio that rebuilt its whole design process around AI only to find its sprints still take exactly five days. The promised dividend — speed, throughput, cheaper output — is the part that keeps failing to show up. What does show up in both is that the payoff moves to whoever spends the reclaimed capacity on judgment: disciplined routing and outcome metrics on one side, taste and getting the hard decisions right on the other. Set against that, the day's one concrete competitive move — a global leisure platform pushing again into Indian SME corporate travel — wins on onboarding friction and local compliance, not on any capability claim. The question the category faces: when the speed story collapses, does budget flow toward governance and taste, or toward one more round of output nobody asked for?
Quartz · 4 Jul 2026
Runaway enterprise AI bills — one company exhausting its full annual budget in four months under per-token pricing — are pushing large firms toward per-project caps, spend dashboards, and governance gates, against a backdrop where fewer than 1% of companies report returns above 20% and most land at 1–5%.
A studio owner reports rebuilding his entire design process around AI across four real client products and finding sprint duration unchanged — five days before, five days after — arguing the gain is not speed but reclaimed time that has to be redirected into thinking, taste, and resolving the hard decisions while they are still cheap to change.
If cycle time doesn't compress, any AI-tooling business case built on velocity or headcount reduction is claiming a benefit that isn't there — the defensible case becomes quality-per-cycle, which changes what a design org buys tools for and how it justifies that spend to finance.