How to be more ambitious: A tutorial for humans with models

An engineer who builds with Claude argues people should use models far more ambitiously than for trivial tasks, proposing four concrete modes of ambition: Speed (do the same work faster), Altitude (operate at higher abstraction to pursue outcomes rather than chores), Depth (tackle problems as top experts would), and Range (borrow techniques from other fields). Practical altitude moves include feeding the model far more raw data, asking it to identify the three most urgent decisions, and reframing a quarterly business review as a system redesign that catches problems earlier. Making altitude work often means changing workflows and giving the model programmatic access to crash reports, version mappings, and target machines so it can act autonomously.

Depth demands expert instruments and methods: require databases and statistical models instead of superficial outputs, set “unreasonably” hard targets, adversarially test assumptions, build simulations to surface new signals, and always validate results with an external judge. Range means turning nouns into verbs to reveal useful disciplines, assembling a multidisciplinary “crew” (analyst, journalist, motion designer, red team), and avoiding empty role-play. The piece ties this to self-efficacy and goal-setting research: concrete, audacious goals and examples of others succeeding reliably expand what people attempt and accomplish. The practical takeaway is to provide models with raw inputs, tool access, and explicit external validation to unlock far greater outcomes.