Agentic AI: The art of the possible

Our insight ‘From discovery to deployment: unlocking value through an agentic AI strategy’ spotlighted the gap between deploying and scaling AI and the creation of tangible organisational value with upward traction in professionalising and positioning organisations for accessing, scaling, managing and measuring AI capabilities.

KPMG’s Global AI Quarterly Pulse Survey Q2 2026 found that only 7 percent of respondents report having an established return on investment with respondents citing data security/privacy/risk concerns (33%) and pressures to demonstrate value to investors or the Board (24%) as key factors influencing AI Strategy as well as responsible AI and governance being a key AI priority for organisations (28%).

This evolution represents a sizeable shift across introducing agentic AI, embedding agentic AI across workflows and launching agentic AI as a commercial, measurable offering that will transform every layer of an organisation. As organisations expand their use of AI, their operating models, governance and workforce capability are not built for this purpose resulting in expanded scope but limited impact and increased risk.

Bridging strategy and reality

The question that organisations are looking to answer: how to structure and approach bridging the gap between an agentic AI strategy and reality in the most complete, efficient, measurable and compliant way without compromising on quality?

Agentic AI introduces autonomous entities, evolving decision flows, and real-time risk exposure that traditional static models were never designed to govern. As AI agents scale, the challenge shifts from deployment to orchestration.

This is a mammoth question for organisations to tackle. There are an abundance of challenges during transformational programs and AI brings with it additional layers of risk, particularly in relation to envisioning the future, what your organisation can achieve and substantiating that the vision can be brought to life for your organisation.

Having an agentic AI strategy is not enough. Failing to operationalise your agentic AI strategy effectively in an era where market, regulatory and technological conditions are continuously in flux can result in your organisation missing the strategic window of opportunity.

The target operating model

The inception of any shift in how your organisation operates is a crucial stage for structuring and shaping a successful future state vision. The KPMG Target Operating Model (“TOM”) can support organisations in identifying how agentic AI can impact every layer of an organisation, provides a proven, ready to-use solution set, and prepares organisations for the future supporting your organisation in translating your strategic ambition into tangible outcomes.

Failing to adapt TOM’s in an agentic AI context can lead to process ownership breaking down across agent chains. The Target Operating Model is structured across six design layers (Service Delivery Model, People, Functional Process, Technology, Performance Insights and Data and Governance) and can guide your organisation’s agentic AI journey in envisioning what the future of an agentic AI could look like gaining the maximum benefit from your agentic AI activities, simplifying the challenges faced by your organisation and choosing the most efficient approach to turn your long term ambition into measurable outcomes through tailored, practical delivery.

Failing to create a TOM for agentic AI means organisations control and performance models are falling behind real-time agent behaviour.

Success depends not only on having a clearly defined, long term agentic AI strategy but also on the design, development and operationalisation of the agentic AI strategy.

Below outlines each of the six design layers encompassing KPMG’s Target Operating Model (“TOM”) as the execution bridge for your agentic AI strategy and some of the implications that organisations face where an agentic AI strategy is not designed and/or implemented effectively and/or a Target Operating Model (“TOM”) is not considered or designed and/or implemented appropriately.

6 design layers: Encompassing KPMG’s Target Operating Model

Service delivery model

What is delivered, how it is delivered, where and by whom the processes should be performed by, across agentic AI activities.
People

Roles and responsibilities, organisational relationships, and skillsets to execute all components of agentic AI activities.
Functional process

End-to-end process execution for the design, development, deployment and ongoing running of agentic AI activities.
Technology

Technology that is used to support each part of the processes and how technology is integrated across agentic AI activities.
Performance insights and data

Reporting requirements and the way they will be reported to drive key business insight and enhance decision making.
Governance

Oversight of agentic AI activities including identifying the specific risks, controls and security segregation requirements related to processes for continued compliance, accountability and transparency.

The cost of strategic redesign

Limited or no returns

If the intent, documentation and vision of the agentic AI strategy all exist but have not been translated into actions, processes, accountability or tangible outcomes this could result in strategy investment yielding limited or no returns.
Inconsistent and fragmented execution

Different areas of the organisation can interpret the agentic AI strategy differently with inconsistent and fragmented execution proliferating.
Strategy fatigue

If the agentic AI strategy consumes time and budget without any tangible outcomes or return on investment, this can result in loss of credibility and confidence in future strategic initiatives.
Elevated risk exposure

When day to day operations are misaligned to strategic intent, operational risk increases across controls, escalation paths, decision making and regulatory requirements.

Strategy

Below outlines some of the key considerations in exploring how KPMG’s Target Operating Model (“TOM”) can support your organisation with a structured approach to explore the art of the possible and bring your agentic AI vision to life.

Service delivery model

Have you documented and defined the Service Delivery Model for your agentic AI activities?
People
- Have you documented and defined all roles, and responsibilities, organisational relationships, and skillsets needed to execute all process areas for your agentic AI strategy?
- How will your organisation’s structure need to change for your agentic AI strategy to be designed, developed and implemented?
Functional process
- Do you have a process taxonomy documenting the end-to-end processes for your agentic AI activities?
- Have you documented and defined the process flows?
- How will your processes need to change and will new processes be required as a result of your agentic AI strategy?
Technology

Have you considered what are the key technology requirements needed to design, develop and implement your agentic AI strategy?
Performance insights and data

Have you considered what will be reported and how to obtain insights and predictive analytics for the performance of your agentic AI strategy?
Governance
- Do you have oversight of how agentic AI is governed and deployed in your organisation?
- Do you understand the key risks and have you got the right controls and security segregation requirements for continued compliance, accountability and transparency?

How KPMG can help

As organisations accelerate their adoption of agentic AI and explore designing, developing and implementing agentic AI strategy, they face a common set of challenges, from defining strategic ambition to navigating regulatory complexity, designing scalable, measurable, operating models, and maintaining global consistency.

KPMG supports leadership teams to translate strategic ambition into practical, scalable operating realities that can support your organisation in envisaging a sustainable, trusted and competitive agentic AI strategy.

Strategy

Agentic AI regulatory readiness assessment

KPMG’s Consulting practice advises organisations in designing, developing and implementing strategy supporting you in defining a clear long‑term ambition for your agentic AI strategy.
Agentic AI Target Operating Model (“TOM”)

KPMG can support your organisation in designing a Target Operating Model (“TOM”) translating your strategic ambition into a scalable operating model.

Execution

Framework design and implementation

Our team supports organisations in designing and refreshing governance, risk and internal control frameworks to establish robust procedures for agentic AI.