Priscila Andre de Oliveira analyzed 116 of her own Claude sessions from daily work at Sentry. 67% were comprehension. 2% were code generation.

Working in a codebase with 15 years of history, around 100 PRs merged per day, and 100,000 organizations depending on it, the unlock is not generation but understanding. She built a personal skill called catch me up with six exploration modes covering architecture, conventions, feature traces, syntax, testing, and history. The loop: understand what the agent found before you let it plan and implement, because a misaligned mental model is where slop comes from.

Speaker info:
- https://at.linkedin.com/in/priscila-andre-de-oliveira-ab34bb24b

AI Engineer

A senior software engineer at Sentry shares how she uses AI primarily for code comprehension rather than generation. After analyzing 116 Claude sessions, she found 67% of her AI usage was comprehension-related and only 2% was code generation. She built a custom 'catch me up' skill — a detailed prompt structured around six exploration modes (architecture, convention, feature trace, syntax, testing, history) — to quickly onboard herself to unfamiliar repositories and gather context for PR reviews. Her key takeaway: in large, complex codebases, AI's biggest value is understanding existing code, not generating new code. She advocates for aligning your mental model before prompting agents, and warns against shipping code you don't understand.

Comprehend First, Code Later: The AI Skill I Rely On Daily — Priscila Andre de Oliveira, Sentry