By Rashmi Kotipalli and Ratna Rao

With plans for agentic AI squarely in scope, many organizations still aren’t finding ROI.

While the Kyndryl Readiness Report 2025 found that 68% of organizations are investing heavily in some form of AI (including agentic), only 54% can currently report ROI on their AI investments.

In response, many business leaders have started to look inward. Their thinking: Maybe if we can prove agentic AI use cases within our own systems and workforce, we will be better positioned to deliver credible, scalable solutions across products, services and customer-facing operations.

As a result, many IT leaders in charge of digital workplace services now find themselves in the novel position of being tasked by their CTOs to deliver fast, clear, measurable results on early agentic use cases.

It’s a sound strategy. And when IT leaders approach it as an opportunity rather than an obligation, they can emerge as the unlikely heroes of enterprise AI – provided they have the right roadmap to guide them.

Digital workplace services can drive fast, measurable results from early agentic use cases.

Why is DWS ripe for agentic AI pilots?

It’s a safer place to fail and learn: DWS includes high-volume but bounded processes — such as IT support, endpoint management and digital employee experience — that are well-suited for iterative pilots, A/B testing and human-in-the-loop oversight. This makes it possible to experiment and learn quickly before advancing towards higher levels of autonomy.
There’s plenty of need and it’s easy to demonstrate clear outcomes: Employee digital experience, service desk efficiency, device and software optimization, and workflow automation remain persistent pain points with measurable metrics — such as MTTR, ticket volume, employee satisfaction (eSAT) and asset utilization — making early impact easier to prove.
It’s a telemetry-rich environment: DWS generates a continuous stream of endpoint, application, collaboration and sentiment data. These signals enable rapid feedback loops, more effective model tuning and clearer tracking of performance and ROI.
It helps accelerate trust and readiness: Starting with DWS allows organizations to build confidence in agentic AI, refine governance and oversight models and harden integration patterns. That foundation is critical before teams can extend AI into higher-stakes business workflows.

What does “high value” look like for DWS — and which use cases deliver it?

IT leaders should start their initiatives by focusing on 3 or 4 use cases with low complexity and low-to-moderate risk that deliver high-impact, tangible value early on. By demonstrating positive ROI for agentic AI use cases with quick wins, organizations can build early momentum and trust from senior leadership to progressively adopt more advanced capabilities as they mature in their AI journey.

Our customers have so far found success in the following:

How it works: Agent orchestrates intake, classification, and resolution and uses knowledge bases, runbooks, and endpoint telemetry to run safe actions (patch, policy update) and escalate with context when needed.
End user benefits: Faster fixes, fewer handoffs, clear status; reduced wait times; proactive prevention of repeat issues.
Sample metrics: MTTR, first-contact resolution, deflection rate, ticket volume reduction, agent action success rate.

How it works: LLM-powered assistants grounded in enterprise data and policies; contextual answers via Teams/Outlook/web; form-filling, request routing, and guidance through compliant workflows.
End user benefits: Quicker, accurate answers; fewer forms and clicks; consistent policy adherence, 24/7 assistance.
Sample metrics: CSAT/eSAT, assistant adoption, average handle time, policy compliance rate.

How it works: Correlates endpoint health and usage telemetry with sentiment surveys, recommends right-size refresh, removes unused licenses, and triggers automated remediations (driver updates, config changes).
End user benefits: Fewer slowdowns and crashes, software that matches needs, longer device lifecycles without performance pain.
Sample metrics: DEX score, crash/slowdown rate, license utilization, refresh deferrals, cost avoided.

How it works: Agentic AI automates routine workflows across IT, HR, and business operations.
End user benefits: Increased productivity, reduced manual effort.
Sample metrics: Workflow completion rate, automation adoption.

How to define success — and what it takes to get there

Progress is iterative.

Based on what we’ve seen work in practice, the following timelines reflect a realistic path from pilot to enterprise impact:

Pilot and foundations (0-90 days): Focus on a small number of agentic AI use cases that balance low implementation complexity with impactful benefits and measurable KPIs. Stand up core data pipelines and RAG-based grounding, establish initial AI governance guardrails and launch limited-scope assistants and automations.

Prove value and harden (3-6 months): Expand coverage and deepen integration. Connect agents with UEM and ITSM systems, add human-in-the-loop controls and begin quantifying improvements across MTTR, ticket deflection and DEX. At this stage, agentic orchestration can be introduced – but only within constrained workflows.

Scale and transform (6-18 months): Broaden agentic AI across multiple functions. Embed agents as “digital teammates” with clear XLAs. Mature telemetry, governance and training programs, and continuously optimize both cost efficiency and employee experience.

How to anticipate and navigate common pitfalls

Challenges, and occasional failures, are inevitable along this journey. Take steps to anticipate friction and put clear mitigation plans in place before scaling, such as:

Mitigation: Adopt unified telemetry and a shared data platform, standardize schemas and use RAG to ensure trustworthy grounding.

Mitigation: Establish a centralized AI governance board, to take charge of auditing models, enforcing policy-based access and establishing human oversight by design.

Mitigation: Invest in role-based enablement: prompt starter kits, access to “low-code” and “no code” development tools, sandbox learning environments and clear communication around benefits and safeguards.

Mitigation: Lead with process-first design, modular architectures and measurable ROI; iterate within guardrails before expanding agent autonomy.

How Kyndryl is uniquely positioned to help

To move beyond fragmented pilots and achieve enterprise-scale AI adoption, many teams will want to consider partnering with a company that has proven expertise, ecosystem depth and operational scale.

Kyndryl exemplifies this role across five critical touch points:

Recognized leadership: Kyndryl is positioned as a Leader in the 2025 Gartner® Magic Quadrant™ for Outsourced Digital Workplace Services, validating its ability to deliver secure, innovative, and outcome-driven solutions globally.
Proven Impact at scale: The Kyndryl Bridge platform powers transformation with nearly 100 million automations per month and over 3 million actionable insights monthly, enabling 1,200+ customers to realize nearly $2 billion in annualized savings through proactive incident avoidance and optimized operations.
Deep ecosystem partnerships: Kyndryl accelerates AI adoption through strategic alliances with:
- Microsoft – Azure AI, Copilot integration and the Microsoft Acceleration Hub for AI innovation.
- ServiceNow – End-to-end workflow orchestration and ITSM/ITOM automation.
- Nexthink/Lakeside – Advanced Digital Employee Experience (DEX) telemetry and analytics.
Successful real-world transformation: For WPP, for example, Kyndryl modernized a global digital workplace by decommissioning roughly 4,700 servers, migrating 1,000 workloads to cloud, upgrading networks across 47 campuses, and automating resolution via Bridge—delivering agility and measurable cost savings.
Kyndryl Agentic AI Framework: This enables enterprises to progress from pilots to AI-native operations, orchestrating secure, self-learning agents with human-in-the-loop governance for responsible scaling.