Taskgentic Agentic AI Strategy and Value Realisation Framework for Communications Service Providers

Turning Agentic AI ambition into measurable, sustainable business value

Executive perspective: From AI experimentation to sustainable business advantage

Agentic AI is creating new opportunities for Communications Service Providers (CSPs) to improve customer experience, simplify operations, increase productivity and develop new sources of revenue. But deploying AI, even at scale, does not automatically translate into business success.

The executive challenge is broader than identifying valuable use cases or demonstrating early benefits. CSPs must establish where agentic AI can make a meaningful difference, invest in the right capabilities, turn that investment into measurable returns, and ensure the cost of operating these systems remains commercially sustainable over time.

This last point is particularly important. Unlike traditional software, agentic AI can incur ongoing consumption costs each time agents reason, plan, call tools, exchange information or complete tasks. Token consumption, model usage, repeated reasoning and unnecessary interactions can all influence the cost of delivering a business outcome. As adoption grows, these costs can become material and, if left unmanaged, erode the very benefits the technology was intended to deliver.

A successful strategy must therefore address three connected priorities:

• Create business value: Focus on business outcomes that matter, from better customer experiences and faster service resolution to improved network operations and greater workforce productivity.
• Realise and prove ROI: Establish clear investment cases, baselines, measurable benefits and accountability for translating improvements into financial and operational returns.
• Manage the ongoing cost of AI: Control the cost of running agents through appropriate AI gateways, deliberate model selection, efficient token generation, usage policies, task design and continuous cost-to-value monitoring.

These priorities cannot be managed independently. A use case that delivers operational benefits may still underperform financially if it consumes too many tokens, relies on unnecessarily expensive models or generates repeated, low-value agent interactions. Equally, minimising token costs at all times can undermine quality, reliability or customer outcomes. The objective is not simply to use fewer tokens or choose the cheapest model; it is to achieve the right balance of performance, risk, cost and business value for each task.

For CSP leaders, this calls for a joined-up approach across strategy, finance, technology, network operations, customer experience, security and workforce transformation. Investment decisions need to consider not only the cost of building and deploying agents, but also their ongoing operating costs, the controls needed to manage consumption and the evidence required to demonstrate sustainable returns.

The Agentic AI Strategy and Value Realisation Framework provides a structured way to make these decisions throughout the lifecycle. It connects strategic ambition and business priorities with use-case selection, investment, organisational readiness, design, deployment, operational management and value measurement. It also treats AI consumption economics as an ongoing management discipline, rather than a one-off technical optimisation exercise.

The central question for CSP leaders is therefore not simply, “Where can we use agentic AI?” It is:

“Where can agentic AI strengthen our competitive position, how will we demonstrate an attractive return on investment, and how will we sustain that return as usage and operating costs grow?”

1. Why CSPs need an Agentic AI Strategy and Value Realisation Framework

CSPs are under pressure to improve margins, manage increasingly complex networks and IT environments, strengthen customer relationships and find new avenues for growth. Agentic AI offers a potential route to these outcomes, but the opportunity comes with investment choices, operational dependencies and new cost dynamics that need to be managed deliberately.

Without a clear strategy, organisations can accumulate disconnected pilots, duplicate capabilities and agents that perform useful tasks without producing a compelling business case. They may also underestimate the cost of operating those agents at scale. Consumption can vary according to task complexity, model choice, context length, the number of reasoning steps and tool calls, and the extent to which agents repeat or expand work unnecessarily.

A robust framework helps CSPs avoid these pitfalls by connecting four questions:

1. Are we solving the right business problems? Prioritise opportunities against strategic objectives, customer needs, operational challenges and realistic delivery constraints.
2. Will the benefits justify the full cost? Assess expected benefits against implementation, integration, governance, infrastructure and ongoing AI consumption costs.
3. Can we operate the solution efficiently and responsibly? Establish the architecture, capabilities, controls and operating practices needed to manage quality, security, risk, model usage and token consumption.
4. Are we achieving and sustaining the expected returns? Measure outcomes after deployment, compare actual performance and costs with the business case, and adjust, scale, redesign or retire solutions as the evidence dictates.

This requires a lifecycle perspective. ROI is not secured simply by approving a business case or launching an agent. It must be tested and managed as real usage reveals how the system performs, what it costs and whether it continues to deliver the intended outcome.

For CSPs, strategic value realisation therefore means balancing business impact, service quality, operational resilience, risk and the full cost of ownership. That includes making deliberate choices about when to use a more capable model, when a smaller or less expensive model is sufficient, how gateways enforce routing and usage policies, and how agent workflows can avoid unnecessary context, repeated reasoning and excessive token generation.

The aim is not to constrain innovation through cost controls. It is to create the conditions for agentic AI to scale with confidence—delivering measurable benefits while keeping consumption, risk and operating costs aligned with the value being created.