Enterprise AI has entered a more demanding phase. Early programs were largely designed to test what GenAI could do. As those experiments have matured, attention has shifted toward how AI can change business performance across core operations.

Prabhakar Appana, Senior Vice President and Global Head, AWS Business Unit, HCLTech, has seen that change emerge rapidly in client conversations over the past 18 months. Enterprises are increasingly looking beyond isolated pilots and considering how AI can operate across customer service, claims processing, software engineering, manufacturing, supply chains, financial reconciliation and IT service management.

“The conversation has moved from ‘Can AI work?’ to ‘How quickly can AI transform my business?’” he said. 

This shift significantly expands the scale of the challenge. Enterprises have already demonstrated that AI can generate content, support employees and automate individual tasks. The next stage involves connecting AI to business processes where agents can reason, collaborate and act across systems and functions.

Appana believes this requires greater discipline in how organizations prioritize AI investment.

“Enterprises don’t need hundreds of AI models and AI use cases. They need a handful of high-value, scalable use cases that fundamentally change how work gets done.”

The emphasis on a smaller number of valuable use cases reflects a broader shift toward measurable outcomes. Productivity, cycle-time reduction, cost optimization and customer experience are becoming central to how enterprises assess whether AI programs deserve to scale. 

Scaling AI changes the operating model

Enterprise-wide Agentic AI introduces a much larger operational challenge as agents begin interacting with business applications, data, employees and other agents.

“Deploying an AI agent is easy. Running thousands of trusted AI decisions across an enterprise is an entirely different challenge,” said Appana.

At enterprise scale, model performance is only one consideration. Organizations need governance, security, observability, identity management, compliance and human oversight around AI. They also need lifecycle management as agents change, learn and take on new responsibilities.

This is an important part of our work with AWS. AWS brings AI services, infrastructure, orchestration frameworks and Responsible AI capabilities, while we contribute engineering expertise, industry knowledge, enterprise integration and global delivery. Together, these capabilities can help clients build AI environments that are trusted, resilient, auditable and able to operate at enterprise scale. 

Redesigning enterprise processes for Agentic AI

The move toward Agentic AI also gives enterprises an opportunity to reconsider processes that were designed before intelligent agents existed.

Many workflows still reflect human constraints: manual approvals, repeated handoffs, fragmented systems and long decision chains. Simply inserting AI into each stage can make an existing process faster while leaving its underlying inefficiencies intact.

“The biggest mistake enterprises make is automating yesterday’s processes instead of redesigning tomorrow’s business,” said Appana. 

A more effective approach begins with the desired business outcome and examines how workflows can be redesigned as agents interact with people, applications and data. Some steps can be automated, while others may no longer be needed.

Appana identifies opportunities for this type of redesign across customer service, claims management, finance, procurement, supply chains, software engineering, HR, IT operations, security and compliance. Recent work in sectors including banking and manufacturing is also providing lessons on how enterprises can remove manual effort and reduce process friction while maintaining appropriate controls.

This is the basis of what Appana describes as Agentic Process Transformation: using agents' capabilities to reconsider an end-to-end process rather than treating AI as another layer of automation.

Co-innovation creates a path to repeatability

As enterprises move toward larger deployments, the way technology partners work together is changing as well.

Our relationship with AWS increasingly involves working together earlier in the innovation lifecycle. Teams can identify business problems, validate use cases, test emerging technologies and develop accelerators before taking solutions into wider client environments.

The objective is to create approaches that can be applied repeatedly across client environments. Once an approach has been tested with clients, the resulting experience and assets can shorten future implementations and support adoption across industries.

Appana describes the relationship in terms of co-building, co-innovation, co-investment, co-marketing and co-selling, backed by a decade of collaboration between HCLTech and AWS. 

That work spans Agentic business process transformation, Amazon Bedrock AgentCore, Amazon Q, cloud modernization and developer productivity, alongside emerging areas such as Physical AI.

Defining business value from the outset

As AI becomes embedded in core business functions, leaders are placing greater emphasis on its economic impact.

“The ROI of Agentic AI isn’t measured by the number of models deployed. It’s measured by how businesses perform differently,” said Appana.

Business KPIs should be defined at the outset of an initiative. Depending on the use case, enterprises may measure productivity, revenue acceleration, customer satisfaction, operational resilience, employee experience, development velocity, quality, compliance, security or cost optimization.

Defining those outcomes before implementation gives teams a baseline against which they can measure progress. It also fosters clearer communication between technology teams and the business about what an AI investment is expected to achieve.

Ahead of AWS re:Invent, Appana is seeing those conversations move further into the boardroom. CEOs, CFOs, COOs and business leaders are increasingly involved in decisions about how AI can transform business functions, modernize legacy operations and improve competitiveness. Clients are also looking across industries for lessons they can apply within their own organizations. 

The next phase of enterprise AI will test an organization's ability to translate rapidly evolving technology into sustained business performance. For Appana, progress begins with clean, reliable data that provides accurate business context. It also requires trusted platforms and processes designed around the capabilities of AI, with business outcomes defined from the outset.