A note from Julia

AI testing and AI quality requires unlearning some comfortable habits, then learning a new set of tools that are rarely handed to you in one clean list. copilots, agents, LLM-powered workflows, model-backed features, adaptive interfaces, RAG apps, generated code, and software whose behaviour depends on prompts, tools and non-deterministic outputs. when systems are non-deterministic, quality is not proven by one good run…quality is measured across samples, slices, traces, risks, failures, behaviour, user-experience, root-cause visibility and time.

Julia Testing AI, page 1 · August 12, 2026 Save to shelf
Notes like this take four seconds in Booksense. Scan the paragraph, it lands on the right page, and comes back for review later. Take your first note

More from readers of Testing AI

More from Julia

See their profile
Testing AI p. 1

Stop losing the best parts of what you read.

Booksense pushes your highlights back to you for review, so a note from page 174 is still yours a year later. Free on iOS.

Julia on Booksense 127 notes 18 hours of reading