
Proactive quality: bad vs. sad

What happens after coding is solved? | Fiona Fung (Claude Code & Cowork)
Topic 10 of 20
Published Jun 21, 2026 · Clip 44:26–48:24 (3:57)
Fiona says the team wants to keep getting better at proactive quality: tracking key experiences with metrics over time and detecting quality issues earlier. She introduced a 'bad vs. sad' framework - bad is a very bad, irrecoverable error; sad is a recoverable pain point that can stack into bad - and gives each team high agency to define what constitutes bad and sad for their own surfaces, like CLI crash rates versus flickering. Lenny's takeaway is that monitoring and tests beat more human review at this velocity, closing the loop so agents know what success looks like. They close with the team's swear-word dashboard, born when an engineer suggested tracking swear words after a frustrating period, as a fun signal of user frustration.