What does it mean that speed optimization?

Making an agent faster means it can serve you more often with less friction.

Example: You have a morning review agent — a routine where you scan your task list, check your calendar, and set three priorities for the day. Currently it takes fourteen minutes. You time each component: three minutes opening and loading your task manager, four minutes scrolling through tasks to find what matters, two minutes switching to your calendar app, three minutes deciding priorities, two minutes writing them down. The bottleneck is obvious — seven of the fourteen minutes are navigation and loading, not thinking. You restructure: a single dashboard shows today's tasks and calendar side by side, pre-filtered by due date and energy level. The review drops to six minutes. Same quality of output, less than half the time. You run this review daily. Over a month, you've recovered four hours. Over a year, forty-eight hours — an entire working week. But the real gain isn't the recovered time. It's that a six-minute routine encounters less motivational resistance than a fourteen-minute one. You skip fewer mornings. The agent fires more reliably because it fires faster.

Try this: Pick one agent — a routine, habit, or recurring process — that you perform at least three times per week. Time it from trigger to completion, breaking it into discrete steps. Identify which steps are execution (actually doing the work) and which are overhead (setup, transition, context-switching, loading, searching). Calculate what percentage of total time is overhead. Now redesign to cut overhead by at least 30% — pre-stage materials, eliminate transitions, batch steps, or remove unnecessary sub-steps entirely. Run the optimized version for one week. Measure the new time. More importantly, track whether you skip or defer the agent less often. Speed optimization succeeds not when the clock is faster, but when execution rate goes up.