The dashboard is dead. What replaced it is the real AI story in financial services

What it actually looks like when the transition works

At Randolph-Brooks Federal Credit Union, the largest credit union in Texas, a VP in a compliance function who'd never written a SQL query started building their own analytics in Databricks. They built over 200 dashboards.

That's a cultural shift, not an IT project. And it happened not because the data team evangelized the tools, but because one business user started explaining the value to thier peers. Peer credibility travels differently than internal advocacy. It reaches people who would never listen to the data team.

Alicia Estrada, Data Intelligence Director at Randolph-Brooks, didn't try to control that momentum. She channeled it. Rather than restricting who could build, she built the certification path, established shared definitions, and let the sprawl teach itself. When two teams reported on the same metric and the numbers didn't match, those teams became motivated to align. They went through the same governance reckoning the data team had already been through, just faster, and when they came out the other side they owned the outcome in a way no mandate could produce.

That's the pattern. Democratize first. Govern as the business demands it, not before. And let the governance conversation be driven by the people who now feel the pain of inconsistency rather than the people who always knew it was coming.

The "what is a member" problem

This sounds simple until you try to answer it at scale. At a credit union, does "member" include joint account holders? Inactive accounts? Youth savings accounts opened by parents? The answer determines everything downstream, from marketing segmentation to risk reporting to regulatory filings. When IT owned analytics, that definition lived in a data dictionary nobody outside the data team had ever read. When the business owns analytics, the definition becomes a live, daily, operational question. And if the definition isn't consistent and governed, every agent, every dashboard, and every AI response gives a slightly different answer.

This is why the data foundation matters, not as an abstract principle, but as the literal prerequisite for everything else working. You can't democratize what isn't governed. You can't govern what isn't defined. And you can't define what only lives in one team's head.

Mark Angler, Senior Vice President at TowneBank, built that foundation in-house with a small team. Naming conventions, data dictionaries, reconciliation frameworks, quality checks. His line was sharp: speed without structure is just faster debt accumulation. TowneBank codified their engineering standards and best practices into AI skills, then used those to build the capabilities the institution actually needs. The result is architectural independence, the ability to make deliberate choices about what to buy, build, and integrate without being cornered by any single vendor.

Adrian Glace, CTO of Amalgamated Bank, thinks in four layers: experience, intelligence, data, and integration, with governance running across all of them. When those layers are clearly defined and owned, every platform decision gets easier because the principles are already in place. You evaluate how well something serves the whole architecture, not just what it claims to do in isolation.

The vendor noise is a symptom, not the disease

Every financial services leader in that room was getting pitched by every hyperscaler, every SaaS platform, and every frontier model company simultaneously. Mark called the LinkedIn AI landscape "AI slop" and said nobody has a silver bullet. The room nodded.

But the vendor noise isn't really the problem. The problem is that when the business still depends on IT for every analytical question, every new vendor looks like a potential shortcut. The pitch cycle exploits the bottleneck. Remove the bottleneck by putting governed data products in the hands of the business, and the vendor noise gets quieter on its own. You evaluate tools against an architecture you already own, rather than hoping each new one will finally make the business self-sufficient.

The data company that already solved data preparation

Jin Oh, Senior Director of Innovation and GenAI at Moody's, brought a different lens. Moody's covers 594 million entities with firmographic data, 210 million companies with financials, 2 billion ownership links, and sources over a million articles daily from 28,000 news sources. Clients working with that data through Databricks are saving 60-80% of the time they used to spend cleaning and verifying before any analysis could begin.

That's the job changing. And it connects directly to the democratization thesis: when the data preparation burden drops by 60-80%, the barrier to putting decision-grade intelligence in the hands of bankers, credit officers, and compliance teams drops with it.

Jin's advice was practical. Begin with the decision you need to make, then work backwards to the data that decision requires. If your credit memo process takes days, that's the problem. If your early warning signals only run quarterly, they're not early.

The time back story runs deeper than you think

A bank CEO recently said something that reframed the entire AI conversation: the real prize isn't headcount savings. It's giving people their time back to do work that actually matters.

That framing showed up in every corner of this discussion. Jin described clients reclaiming 60-80% of the time previously spent cleaning data. Alicia described engineers freed from pipeline troubleshooting to spend their expertise on problems that actually require it. Mark described a team that used to maintain ETL pipelines now building the capabilities that make the bank competitively distinct.

When you democratize governed data, time comes back at every level of the organization simultaneously. That's the compounding effect most financial institutions haven't yet experienced, and it's why the institutions that have made this shift are pulling ahead of the ones still running the old model.

Databricks is building the platform for exactly this shift

As these practitioners were describing this on stage at DAIS, Databricks was announcing capabilities that map directly to what they've been building by hand.

Genie Ontology is an automatic context layer that extracts business knowledge from connected sources, determines authority based on usage, and serves it to Genie while respecting source permissions. For institutions wrestling with the "what is a member" problem, or any regulated environment where consistent definitions and auditability are non-negotiable, this is the single-source-of-truth answer built into the platform rather than bolted on afterward. In Databricks' own testing, Genie answered 84.5% of questions correctly on the first attempt, compared to 52.4% for the strongest general-purpose coding agent, the difference between an AI tool you can trust in a credit review and one you have to double-check every time.

Unity Catalog Metrics lets teams define KPIs once as governed, reusable objects, then query them consistently from SQL, BI tools, APIs, and agents. Define the metric once, share it everywhere, and when two teams report different numbers they're wrong about the query, not the definition. That's the scalable answer to the governance reckoning that follows democratization.

Genie ZeroOps, currently in private preview, is a background agent that monitors data and AI assets, detects pipeline failures, performs root cause analysis using Unity Catalog lineage, and proposes a tested fix in a sandbox for human review. The engineer approves. The fix deploys. The pipeline keeps moving. For financial institutions where pipeline errors carry regulatory implications, the human-in-the-loop design is the right call.

Unity AI Gateway extends Unity Catalog governance to the runtime interactions between models, agents, and tools, hard spend caps, PII guardrails, content filtering, contextual service policies, and unified audit trails across every agent interaction. As agentic workflows scale, this is the governance layer that makes democratization safe, not just fast.

More than one practitioner in that room said it felt like Databricks had been reading their mind.