Session

Sponsored by: Hexaware | Data Products Unleashed: How Databricks Drives Exponential Business Value

Overview

ExperienceIn Person
TrackArtificial Intelligence & Agents
IndustryEnterprise Technology, Healthcare & Life Sciences, Financial Services
TechnologiesUnity Catalog, Databricks Agents, Lakebase
Skill LevelIntermediate

The data-as-a-product mindset treats every dataset as a managed business asset-high quality, well documented, discoverable, and built for consumption. Productizing data turns it into a repeatable source of measurable value, not just an engineering byproduct. Databricks accelerates this transformation by helping organizations design, govern, and deliver trusted, AI-ready data products at scale, enabling reuse across analytics, ML, and intelligent automation. Attendees will explore how Databricks, combined with Hexaware, helps activate intelligent agents to orchestrate complex workflows, strengthen governance, lineage, trust, and accelerate delivery of reliable data products. Join us to discover how shifting beyond traditional data architectures unlocks faster, compounded value for data consumers across the organization.

Session Speakers

Bindu Birur

/Head of Data Eng & Analytics Delivery
KPMG LLP (US)

Speaker placeholderIMAGE COMING SOON

Sameer Bhagi

/Technical Associate, Data Engineer
KPMG

Full Summary

Why data products are the foundation for agent‑ready AI at KPMG

AI has revived an old discipline: treating data as governed products rather than loose datasets or reports. KPMG outlines how rebuilding its data estate around productized, well‑described assets is becoming prerequisite for AI agents that require structured, contextual, and trustworthy inputs. The conversation reframes data product thinking as the on‑ramp for agentic AI and pairs it with knowledge engineering to encode business meaning for machines.

FAQ


A data product is a governed asset with documented inputs and outputs, business and technical definitions, SLAs, contracts, and rich metadata. It is designed for repeatable, trusted consumption by multiple downstream users, including agents, and carries the governance needed to justify that trust.

Agents lack the implicit context humans apply when reading tables and columns. Without explicit ontology, relationships, and semantics, they are prone to hallucinations and incorrect joins. Knowledge engineering encodes the meaning agents need to reason over enterprise data.

KPMG structures knowledge engineering as three layers: ontology for business dictionaries and key concepts, knowledge graphs for relationships and lineage across data products, and a semantic layer that translates technical structures into business vocabulary users and agents can understand.

KPMG has built data products for resource management and pricing, and launched an Information Exchange initiative that parses contracts and engagement documents to create AI‑ready, permissioned data. Adoption required a mindset shift as users learned how to engage with productized data.

Capture business and technical metadata consistently, register it in a governed catalog, define SLAs and data contracts, and invest in ontology and relationship modeling. These steps determine whether downstream agents can retrieve contextual, trustworthy information.