Contact us: https://bit.ly/4fAFrVx
Jack walks through a practical pattern for letting anyone query a live database in plain English—without needing SQL skills or being locked into a single ecosystem. He compares Microsoft Copilot Analyst to a custom-built approach, then breaks down an end-to-end pipeline: retrieve the right context (schema + business definitions + examples), generate SQL deterministically, validate/secure execution with proper DB permissions, and return results as tables or charts (including adaptive cards for chat apps like Teams). He also covers “auto-learning” via storing successful Q&A pairs back into a vector DB, plus key risks like hallucinations, governance, over-permissioning, and how to mitigate them with explainability, telemetry, and safeguards.
00:00 | Introduction: Copilot Analyst & why AI analytics keeps changing fast
05:05 | The goal: query a live database in plain English (not just spreadsheets)
08:18 | The 3-part grounding recipe: schema + domain definitions + example Q&As
11:44 | Prompting for deterministic SQL (and why context engineering matters)
13:12 | Extracting + validating SQL (structured outputs vs parsing)
15:08 | “Auto-learning” the system: saving successful queries + updated definitions
18:05 | Governance: telemetry, curation, and improving quality over time
20:03 | Security reality check: prompts won’t protect your DB—permissions will
25:17 | Making it useful: scalars, tables, charts + adaptive cards in chat
48:23 | Audience Q&A: complexity, explainability, recovery, and role-based access
About SSW: https://linktr.ee/SSWTV
Links:
LinkedIn: / ssw
Twitter: / ssw_tv
Facebook: / ssw.page
Instagram: / ssw_tv
Tiktok: / ssw_tv
Get in touch: Jack Watts – AI Strategy & Implementation Leader
LinkedIn: / jackawatts
#ai#dataanalytics#sql#naturallanguageprocessing#rag#vectordatabase#copilot#microsoftteams#datagovernance#cybersecurity#promptengineering#azure#plotly#adaptivecards#businessintelligence
V1 – Bahjat
About the speakers
Jack Watts
I'm a seasoned technology consultant with over 20 years of experience spanning software development, AI & machine learning, operations research, and automation across Australia and international markets.
Currently, I lead AI initiatives at an R&D studio, where I focus on both the technical implementation and strategic direction of AI projects. My role encompasses supporting AI engagements from concept through delivery, while maintaining awareness of the broader social and economic implications of these technologies.
Related videos
