Day 001 · 2026-05-23
AI Products Fail Before the Model← All events
conference · active·Toronto
AWS Community Day 2026
Rebuilding Products with AI
89days to stage
·
Preparing since 2026-05-23
·
1 notes
A 30-minute talk on the top 10 ways to rebuild your product with AI. Drawing from real experience shipping AI-powered features, diagnosing data quality problems, building observability, and redesigning product teams.
Evolving Rules
01
Start with workflow, not the modelImproving
02
Fix data quality before adding intelligenceImproving
03
Build observability before autonomyDraft
04
Design for human review and correctionDraft
05
Treat prompts, tools, and evaluations as product assetsDraft
Journal Entries
Fragments
“Most AI product failures are not model failures. They are system readiness failures.”
2026-05-23
“The model is the easy part. The workflow is the hard part.”
2026-05-23
“AI amplifies whatever system it sits inside — including the broken parts.”
2026-05-23
“Observability must arrive before autonomy.”
2026-05-23
“Trust is the real product surface area of AI.”
2026-05-23