CBA Live 2026: Banks don’t have a technology problem. They have an operating discipline problem.

Having just returned from CBA Live in San Diego, I spent my time moving between general sessions, product deep dives, and conversations with peers across the industry.

Like most conferences, there was no shortage of energy around what’s next: AI, data, modernization. All the right topics were on the agenda. But as I listened more closely, not just to the presentations, but to how people described their day-to-day challenges, a consistent pattern started to emerge: an execution gap between strategy and frontline behavior.

The gap isn’t innovation. It’s execution.

We’re not lacking innovation. AI, better data, new platforms and modern tooling are everywhere. If anything, there’s an abundance. But across sessions, one reality kept surfacing: most banks still can’t operationalize what they already know works.

It’s the system that breaks first, not the technology. 

During CBA LIVE, I moderated a conversation between leaders from U.S. Bank and Truist. It didn’t take long for the discussion to move past tools and training. The focus quickly shifted from “what are we deploying?” to “why isn’t this sticking?” 

The issue wasn’t capability. It was operating discipline. A few things quickly became clear:

• Execution doesn’t break in the branch. It breaks earlier, when too many initiatives are pushed without clear prioritization or defined workflows.
• Adoption doesn’t happen after rollout. It happens during live client interactions, where teams either attempt to use/find the right tool or default back to existing habits. 
• Tools don’t change behavior. Habits do. 

Both leaders reinforced this from slightly different angles, but with the same conclusion: enablement doesn’t fail because people don’t care; it fails because it’s competing with too many priorities and isn’t embedded in how work actually gets done.

What separates the banks making progress is how they manage change at the operating level:

• Banks win when prioritization is enforced at the leadership level
• Manager routines reinforce expected behaviors consistently
• Enablement is embedded into workflows, not delivered as standalone training

In other words, this is an upstream problem: competing priorities, fragmented ownership and too many initiatives hit the frontline before enablement even begins.

Frontline teams are already balancing service, sales, support and digital guidance. Layering in “one more tool” or “one more initiative” doesn’t solve that. It only adds noise. The banks making progress are doing something different: fewer, clearer priorities, enablement embedded in routines, coaching tied to real client conversations and reinforcement that shows up consistently. One-time launches are easy. Sustained behavior change is not.

The same pattern showed up in underwriting.

I expected the SMB underwriting session to focus heavily on data and models. And it did. But the more interesting part was what sat underneath that conversation. It wasn’t really about better predictions, but whether organizations can actually run what they build. 

On the surface, the conversation focused on faster decisions, smarter data and next-gen credit. And the data story is real. Cash-flow and transaction-level data are improving predictive accuracy, particularly for newer businesses that don’t fit traditional models. Speed in underwriting isn’t a modeling problem. It’s an operating model problem. The conversation quickly moved from “can we predict this better?” to more practical questions. Can we run this consistently at scale? Can we explain it? Can we defend it? 

The banks making progress aren’t focused on marginal model improvements. They’re redesigning end-to-end decision workflows, standardizing paths and SLAs, structuring exception handling, and embedding governance directly into the process. The question isn’t “how do we get better models?” It’s whether those models can be run consistently, at scale and under scrutiny.

Everyone wants AI…until governance enters the conversation.

It would’ve been easy to assume AI would dominate every discussion. It did in volume, but not always in substance. The more credible conversations had a different tone: less hype, more realism.

Across sessions, the same constraints came up:

• Explainability still matters
• Adverse action still matters
• Model risk management still matters

These are not edge cases but rather design constraints that slow many initiatives down.

What I heard repeatedly was a familiar pattern: prove the model works, build internal momentum, then, because governance wasn’t incorporated from the start, hit friction when trying to move into production. Governance shouldn’t be a constraint. It should be part of the architecture.

We don’t have a data problem anymore.

Across sessions, there was a consistent acknowledgment that banks are sitting on more data than they know what to do with. Between transaction-level visibility, embedded platform activity, and internal operational signals, banks have more data than they can realistically use. 

Banks don’t lack data; they lack the ability to translate data into decisions. Data flows remain fragmented, features aren’t standardized, policies don’t align, and workflows don’t fully leverage what’s available. The data exists, but it doesn’t translate to impact.

This isn’t a series of disconnected problems. It’s the same breakdown showing up everywhere.

What became clearer as the week went on is that these aren’t separate issues. They’re the same issue showing up across different parts of the bank: branch enablement, underwriting, AI and governance. Different conversations. Same root problem:

There’s no system connecting priorities → behavior → outcomes.

• In the branch: priorities don’t translate into consistent behavior 
• In underwriting: data doesn’t translate into consistent decisions 
• In AI: models don’t translate into consistent impact 

These are different domains, but with the same breakdown.

AI isn’t the fix. It’s the multiplier.

In one of the more grounded discussions featuring leaders from U.S. Bank, Atlantic Union Bank and PNC, a consistent theme emerged: AI is often positioned as the solution, but in reality, it amplifies whatever system it sits atop. If your current state includes fragmented workflows, inconsistent data and/or unclear ownership, AI will scale that inconsistency. If your operating model is tight and includes structured workflows, clean data pipelines, and defined governance, AI becomes a multiplier.

That’s where the gap is showing up right now. Many organizations are investing heavily in AI capabilities, but the underlying processes they rely on remain inconsistent, manual or fragmented. The result isn’t transformation. It’s faster inconsistency.

The institutions that actually see impact are those that treat AI as something that sits on top of a well-defined system, not as something that replaces the need for one. That’s the difference between experimentation and transformation.

What actually matters coming out of this

As I take a few moments to think back on the few days at CBA LIVE, here are the key points that resonated with me:

1. Lead with operating discipline, not technology: Technology follows structure. Not the other way around.
2. Design governance into the system: Explainability, auditability and control need to be built in, not layered on.
3. Fix exception handling: That’s where both speed and consistency break down.
4. Embed enablement into the flow of work: If it sits outside the workflow, it won’t stick.
5. Treat these as systems, not tools: Underwriting, enablement, and decisioning are not isolated functions.

Final thought

There’s no shortage of ideas in banking right now. However, the institutions that will advance are the ones willing to accept this premise: This is not a technology transformation.

It’s an operating discipline transformation. Everything else, including AI, follows from that.

If you were at CBA LIVE, I’d be interested in what stood out or resonated with you. If you are working through these challenges, especially around prioritization, governance, or enablement, I would be interested in how you are approaching them. These are solvable problems, but they require a different approach than most organizations are taking today. Connect with me to continue the conversation.