knowledge26: 3 levers for turning servicenow into real AI ROI for BFSI.

24%. That’s the percentage of banking enterprises that have AI testing, auditing and risk processes in place, a figure ServiceNow presented¹ in a financial services spotlight at Knowledge26. For insurers, it was 21%. These numbers stayed with us after the event, leading us to a bigger question: What would it take for every enterprise in that room to move that number higher, and what’s standing in the way of real ROI from ServiceNow and AI?

For an industry built on risk management, the governance gap is striking. The average bank runs a complex ecosystem of applications, each with its own AI layer. But few institutions have fully mapped their AI tool landscape, which means some decisions and actions fall outside a governance framework. In BFSI, that’s a compounding liability and, at scale, an existential one.

The enterprises deploying AI at scale and driving results have answered this deliberately. Connected, coherent workflows across systems. Governance built in. A clear way to see, manage and audit every AI tool across the estate. That’s how the confidence calculus shifts. And ServiceNow is already the connective layer for many of these institutions. 

At Knowledge 2026, three things stood out to us as the clearest levers for turning a well-configured ServiceNow instance into real AI ROI.

1. visibility: map your enterprise’s AI landscape and connect your data.

In order to ensure AI delivers real results, you need to address two dimensions.

your enterprise AI landscape

Shadow AI isn’t a remote risk. As agents, models and automated decision layers are embedded across thousands of applications without formal sanction, it’s a real concern. The exposures this can create are specific:

• Security vulnerabilities from agents operating outside IT visibility.
• Compliance violations from models making decisions without audit trails.
• Customer experience failures that are nearly impossible to trace to their source.

In addition to uncovering areas of exposure, mapping the full AI landscape is also where institutions discover which tools are actually working, which are duplicating effort and where the next wave of automation is hiding.

connecting the data

AI performs exactly as well as the data it runs on. In most financial institutions, that data is fragmented and siloed across legacy systems, inconsistently structured, duplicated across platforms that were never designed to connect. A dispute resolution agent pulling from four disconnected sources doesn’t produce better outcomes faster. It produces inconsistent outcomes confidently.

At Knowledge 2026, a major card network gave a transparent account of how fragmented their payment dispute process had become, everything spread across disconnected systems, with no coherent foundation for AI to work from. Once they rebuilt their dispute management platform into a single system, they could finally use AI to spot fraud and process bulk tasks at scale. Recovery rates improved, and analysts were freed from the manual work that had been eating their time.

Of course not every institution can or should rebuild from scratch. ServiceNow’s Workflow Data Fabric addresses exactly this problem, connecting disparate systems into a unified, real-time data layer that AI agents can actually use, without requiring wholesale migration or consolidation.

2. build workflows AI can actually run on.

Visibility gets you to the starting line. The next step is making sure every action, human or agentic, leaves a traceable record.

With workflows built for people, some hand-offs could rely on institutional knowledge; steps could live in email threads. AI agents don’t navigate that well, and regulators don’t accept it. Tools like ServiceNow’s Now Assist, and the broader agentic execution direction ServiceNow demonstrated at Knowledge 2026, are increasingly capable of executing multi-step processes across systems, but only when the underlying workflows are structured to support them.

Workflows built for AI look different:

• Process steps are explicit and sequenced, with no ambiguity about what triggers the next action.
• Hand-off logic is defined. The agent knows when to proceed and when to escalate to a human.
• Every action generates a record: what triggered it, what data it touched, what the outcome was.

For high-volume BFSI workflows, like client onboarding across KYC, account provisioning and compliance tracking; loan processing across document intake, credit assessment and approval routing; claims intake that touches underwriting, compliance and finance, this structure makes autonomous execution both possible and defensible.

For institutions already running ServiceNow, extending its capabilities to AI-executed workflows is often incremental. The infrastructure exists. The question is whether it has been deliberately applied to the actions AI is already taking.

Those auditable workflows are also what makes governance real rather than theoretical. Without them, even the best oversight tooling has nothing concrete to monitor.

3. govern what AI does, not just what it’s allowed to do.

Knowing your estate and building auditable workflows creates the conditions for governance. Without them, governance is policy on paper, with no mechanism to enforce or verify it.

Real AI governance in financial services is a growth lever that means something specific:

• The ability to define what every agent or AI tool is permitted to do.
• Real-time monitoring that confirms it is operating within those parameters.
• A complete record available to any regulator, auditor or internal review on demand.

ServiceNow’s AI Control Tower provides that control panel: a single layer across strategy, governance, management and performance designed to discover and make your AI estate visible, explainable and operating within defined boundaries. This lets institutions answer confidently to their boards, regulators and customers that they are in control of their AI.

At Knowledge 2026, a leading US bank demonstrated this in practice. An AI agent resolved a payment failure automatically. The audit trail was complete. There was no compliance exposure because auditability was native to the workflow, not reconstructed after the fact.

The governance infrastructure that satisfies a regulator is the same infrastructure that lets you deploy AI more aggressively in customer-facing operations: faster dispute resolution, smarter onboarding, more responsive service. Operational resilience and AI scale are the same investment.

conclusion.

The institutions with visibility into their AI estate, clean and connected data, workflows that agents can actually execute and governance infrastructure that satisfies regulators, boards and customers.

As a ServiceNow partner, Randstad Digital works with BFSI organizations on maximizing their platform and AI ROI. Our SPARC framework (Skills, Platform, Automation, ROI, Cost) connects implementation to outcomes, because the technology only delivers value when it’s operating in the context of your organization, not alongside it.

reference:

1. https://www.servicenow.com/events/ondemand/knowledge/financial-services-spotlight--connected-governed-trusted-ai-.html?collectionNames=Industry+Solutions