~/talk

Personalizing RAG for Prompt-Driven Enterprise Analytics

links:none on file
slides:none on file

RAG-based system personalizes presentation while keeping facts constant: a concrete example shows checkout conversion falling 5% to 4.5% with mobile driving the change and a flagged payment error rate 2% to 3.5% as an unvalidated hypothesis. Three outputs are produced: reproducible SQL, metric definitions and lineage for engineers; segment breakdowns and evidence for analysts; impact, owner assignments and next steps for PMs. Architecture uses profile retrieval generation with indices for schema, docs, metric defs and curated SQL, plus eligibility, ranking, phased rollout and evaluation checks.

AI-written summary · about

  1. 0:00Introduction
  2. 0:28One Question, Three Roles
  3. 1:02What RAG Is and Why Grounding Isn't Enough
  4. 2:08Checkout Conversion Example
  5. 3:38Personalize Expression, Not Facts
  6. 4:49Engineer, Analyst, and PM Outputs
  7. 8:32Decisions Behind Personalization
  8. 12:10Retrieval Indices and Ranking
  9. 16:14Personalized Generation
  10. 17:51Rollout, Evaluation, and Takeaways
~/search
type to search 6,518 conferences, 143,918 speakers and 192,133 talks
~/help
cmdctrl k
search every conference, speaker and series
/
focus the filter search
h
home
c
conferences
s
speakers
e
series
t
light / dark — the status bar button does the same
← →
on a talk: 5 seconds back / forward
0–9
on a talk: jump to 0%–90% of the recording
?
this help
esc
close this, then clear every filter — or just the search, while you are typing in it

Ignored while you are typing. Pages without a filter rail have no search to focus.