Session

Powering KPMG Audit’s AI Future with Databricks

Overview

ExperienceIn Person
TrackData Warehousing
IndustryConsulting & Services
TechnologiesAI/BI, Databricks SQL, Lakeflow
Skill LevelBeginner
DOWNLOAD SESSION SLIDES

What does it take to turn a complex audit estate into an AI‑ready lakehouse in months, enabling auditors to move faster without losing trust? KPMG migrated nearly 400 SQL Server scripts into Databricks SQL in three months using lakehouse patterns, Lakebridge, and AI‑assisted refactoring — cutting manual effort while maintaining audit‑grade governance, lineage and performance. This session unpacks the blueprint: unified governance on Delta, SQL Serverless to handle extreme peak‑season demand and secure cross‑entity data access built for regulated environments. A focused enablement program onboarded 200+ users rapidly, embedding modern engineering practices and an AI‑first mindset. Finally, we’ll explore how this platform now powers AI/BI Genie and Lakeflow Designer, putting governed natural‑language and low‑code analytics directly into the hands of citizen analysts and front‑line auditors.

Session Speakers

Ellie Tucker

/Global Account Exec for KPMG
Databricks

Greta Nasai

/Sr. Solutions Architect
Databricks

Kevin Kolliniatis

/Partner
KPMG LLP

Ian Hudson

/Senior Manager
KPMG

Full Summary

Inside KPMG UK's audit AI transformation: rebuilding a governed foundation for trust

KPMG UK modernized its external audit analytics from a constrained on‑premise SQL estate to a governed lakehouse on the Databricks Platform. The goal was not a lift and shift, it was a redesign that unifies data, governance, analytics, and AI for a highly regulated practice where evidence, lineage, and reproducibility are non negotiable.

FAQ


They used Lakebridge for automated estate discovery and risk classification, then applied GenAI code inside the Databricks workspace to convert and refactor T‑SQL into Databricks SQL. Engineers validated every output against the original, which preserved audit quality while reducing refactoring effort by roughly 60 percent.

Unity Catalog unifies governance with lineage, access control, and audit logs across data and AI assets. Delta Lake adds time travel and versioning for reproducibility, and Delta Sharing supports secure, permissioned data exchange with audited entities. Together these controls provide the traceability regulators expect.

Audit demand spikes from late summer through spring, then drops. Databricks SQL auto‑scales to match that pattern, so teams avoid overprovisioning. The same elastic compute also runs AI tasks, which removes the need for a separate AI infrastructure stack.

About 200 users completed a tailored program covering Databricks engineering and SQL, Unity Catalog governance, and Genie, aligned to KPMG audit quality. Instructor‑led courses, certifications, agile processes, and ongoing office hours supported teams through early live audits with parallel runs and reconciliations.

The UK team delivered a Financial Report Analyzer that uses Genie and generative AI to assess disclosure completeness and accuracy. They also launched an AI‑first migration of Alteryx workloads into Lakeflow Designer and adopted a factory approach to halve time to production for each new AI use case.