The connected system wins the decade

Michael Rouleau

Chief Executive Officer

Nearly every conversation I have with a financial services leader lands in the same place. They know AI works, they've seen it work in their own shop. They want to know why it's so hard to run at scale.

It's the right question, and after three years of watching institutions try, the pattern is hard to miss. The ones pulling ahead run customer experience, data, and AI as one connected system. The ones stuck run three separate programs that compete for budget, talent, and attention, and the seams between those projects are where pilots go to die.

The gap already shows up in the numbers. McKinsey projects that AI pioneers in banking will open a lead of roughly four points of return on tangible equity over slow movers. A gap that size doesn't close on its own.

Watch what the platforms built

Don't take my word for it. Look at what four of the world's leading technology companies built in 2026. Each spent the year hardening a different layer of the same system.

Data: At Databricks' annual summit, CEO Ali Ghodsi named the industry's real problem. Models are smart enough already. The missing piece is governed, connected data to reason over, and that's the gap Databricks spent the year building toward.

Intelligence. Anthropic's biggest financial services release this year wasn't pitched on raw model capability. It was a library of ready-to-run agents for credit memos, KYC, underwriting, and month-end close, each wrapped in the audit trails and human checkpoints that regulated deployments demand. Anthropic knows where production stalls, and built for it.

Experience: The experience layer moved the same way. Salesforce went from single-purpose assistants to orchestrated teams of agents running inside the trust boundary regulated institutions already rely on. Its Agentforce Operations product now handles loan underwriting end to end, pulling data from tax returns, chasing signatures, and checking every detail against policy while a loan officer focuses on the customer instead of the paperwork. Twilio built persistent customer memory that carries context across every channel, so a customer who starts in a chat and ends on a call has one continuous conversation.

Four companies, one architecture. Data agents can trust, intelligence regulators can audit, and experience customers feel. None of these layers is worth much on its own. The value shows up where they connect, and the institutions treating them as one system are the ones getting agents into production while everyone else runs pilots.

Building the connected system

Here's how I'd put it across a table. Your experience layer is how you serve the customer. Your data is how you know the customer. AI is how you do both at scale without adding headcount.

The data comes first. Most institutions get this backwards. They buy the AI tool, point it at a mess, and get a faster mess. You can't automate a mess. Once the data is solid, experience and AI draw from one foundation instead of running as two more initiatives with owners of their own. Put an agent on fragmented data and you get confident, fast, wrong answers.

From there, three disciplines keep the system from splitting back into three projects.

Start narrow. Prove it on one workflow before you scale it anywhere. KYC screening, claims intake, whatever makes the value clearest for you. Run it end to end, so the same data powers the customer's experience and the agent acting on their behalf.
Bake governance into the platform. Policy should be enforced before an agent acts, with every decision logged and audit-ready by default. That's different from reviewing logs after something's gone wrong, and done right, it's what lets you move fast safely.
Measure what the business measures. Time saved and user-reported efficiency describe a tool, not a system. Tie your numbers to something the business already tracks, loan funding speed, claim resolution time, application intake volume. If you can't measure the before, you can't prove the after, and programs that can't prove the after rarely survive their second budget cycle.

Where Zennify fits

Most partners in this market run one layer. They do customer experience, or data, or AI. Zennify runs all three, in regulated industries, and that intersection is the whole point.

We're deep in the Salesforce, Databricks, and Twilio stack, and we're putting Claude to work inside regulated financial services every day. That seat means we can tell you when headless Claude fits a workflow and when Agentforce does, and what your data has to look like before either one earns its keep.

If you're deciding where to begin, start with an honest read of where your three layers actually stand. That's exactly what our Digital Maturity Assessment gives you, an evidence-based baseline in days, and it costs you nothing. The roadmap follows from there.

We know this works because we built the system inside our own business first, before we asked any client to follow us. More on that soon.