Session

Blueprint to Breakthrough: Activating Banking Analytics, AI, and Agentic Use Cases

Overview

Experience	In Person
Track	Artificial Intelligence & Agents
Industry	Financial Services
Technologies	Genie
Skill Level	Intermediate

This session continues the story of how a Regional Bank is modernizing its AI and analytics journey, evolving from building the foundation to activating value through measurable business outcomes. Powered by Databricks and Deloitte’s industry experience, the Enterprise Data and Analytics Hub is a one-stop shop for line-of-business and enterprise consumption, anchored by a governed Data Marketplace that simplifies discovery and reuse of trusted data products. As part of this evolution, Databricks serves as the knowledge base for enterprise agentic and AI capabilities, and Databricks Genie is being leveraged to accelerate development and delivery of business-ready solutions through conversational and developer-enabled experiences.

The session will highlight how the bank is operationalizing these capabilities through a prioritized pipeline of high-impact use cases by accelerating speed to insight, improving consistency across reporting and analytics, and realizing value at scale. Discover how the bank is embedding advanced analytics and AI, including agentic and multi-agent systems, into decisioning and workflows with strong governance, security, and controls.

Session Speakers

IMAGE COMING SOON

Mohan Sankararaman

/Chief Information Officer
First Horizon Bank

Nikhil Kaushik

/Senior Manager - AI & Engineering
Deloitte

Full Summary

Inside First Horizon Bank's data rebuild: from silos to a governed marketplace and agentic AI

First Horizon Bank, a mid-sized regulated institution based in Memphis, has spent the past two years replacing fragmented warehouses with a unified data foundation. The conversation details how a governed hub, a live data marketplace, and early agentic capabilities are coming together on the Databricks Platform. For a deeper look at how governance and intelligence combine in financial services, the Databricks Platform for Financial Services outlines the architectural principles that make this kind of rebuild possible. The effort matters because it shows how a bank can move from "data as a service" to data as a product, then use that discipline to make conversational analytics reliable today and agent-built pipelines plausible tomorrow.

Governance by design on a medallion foundation

The rebuild began in 2025 with an Enterprise Data Hub structured on bronze, silver, and gold layers, plus a new platinum tier for locked-down regulatory reporting. More than 35 systems of record, such as deposits and loans, now feed curated products into the hub so analysts spend less time sourcing and reconciling and more time on analysis.

Governance is built in, not bolted on. Unity Catalog provides lineage down to attributes, data quality is enforced from ingestion, and every published dataset carries a data contract that defines ownership, refresh cadence, and expected quality. In a regulated setting, that trust layer enables safe scale.

A productized marketplace for consumption

The centerpiece is a live marketplace that functions like a data app store. Built on Databricks Apps and integrated with Unity Catalog, it lets executives, managers, and reporting analysts browse gold-layer products, review business and technical metadata, inspect end-to-end lineage, see quality controls, and request access through workflows tied to the bank's security framework.

Treating each dataset as a product with an owner and lifecycle prevents uncontrolled proliferation, keeps governance explicit, and sustains trust as usage grows.

Conversational access and the path to agentic-first delivery

Early agentic capabilities are already in use. A talk-to-your-data interface answers prompts such as "summarize deposit accounts and their status," returning charts for active, inactive, closed, and dormant accounts, principal balances by status, and distributions by account type. Rich context from Unity Catalog, including glossaries, contracts, lineage, and quality signals, improves accuracy and relevance.

The bank has stood up a multi-agent framework and is adding an ontology layer so agents can build pipelines and visualizations, not just answer questions. The target for 2027 is agentic-first across the data development lifecycle, with human-in-the-loop oversight, and interest in agent ops to reduce manual pipeline maintenance.

Data as a product, not a service

The organizing principle is a shift to data as a product. Products have owners, lifecycles, and investment, which makes access decisions clear, quality expectations explicit, and changes manageable. That discipline underpins both the marketplace's credibility and the defensibility of agent-generated outputs.

What a product mindset unlocks for AI in a regulated bank

A governed product model, paired with visible lineage, quality, and contracts, turns conversational analytics from a demo into a durable capability. It also lays the groundwork for agent-built pipelines and dashboards that a bank can audit, support, and scale.

FAQ

Gold hosts curated datasets for broad analytics and dashboards. Platinum sits alongside gold as a more locked-down tier for regulatory reporting, where stricter controls and stability are required.

It runs on Databricks Apps on top of Delta Lake and integrates with Unity Catalog for lineage, metadata, quality, and access control. Executives, operating managers, and reporting analysts browse governed products, review context, and request access through workflows tied to the bank's security framework.

Contracts and lineage define source, ownership, refresh cadence, transformations, and quality upfront. In a bank, making governance explicit at publication time prevents uncontrolled proliferation and avoids the regulatory exposure that follows ad hoc data sprawl.

Yes. Users can prompt summaries such as deposit account status, principal balances, and account type distributions. Responses are grounded in governed gold-layer products and informed by metadata, lineage, and quality signals from Unity Catalog.

The team is enabling agents to build pipelines and visualizations using an ontology layer and the platform's context. A multi-agent framework is in place, and the bank is exploring agent ops to automate maintenance so limited analytics staff can focus on ideas and outcomes.