Debanjan Banerjee, Global Service Management Director at Pernod Ricard India, outlines the shift from AI pilots to “trustworthy autonomy.” Explore how the spirits giant is leveraging embedded AI agents in HR, IT, and Supply Chain to automate sub-processes while building the governance frameworks
As enterprises move beyond AI pilots, the question is no longer whether AI can act, but how much autonomy to grant, and where human oversight remains essential. AI agents are increasingly embedded across business processes, from HR and IT operations to supply chain and marketing, delivering measurable efficiency gains and business impact.
In this conversation, Debanjan Banerjee, Global Service Management Director at Pernod Ricard India, shares the company’s pragmatic approach to AI agents, the tangible results achieved so far, and the frameworks being built to ensure trust, accountability, and sustainable scale.
CIO&Leader: Where are you today on AI agents? Are you still experimenting, running pilots, or operating AI agents in live production? Which business functions are using them today?
Debanjan Banerjee: Over the past several years, we have successfully adopted AI at scale across multiple geographies, particularly in Sales and Marketing. This journey has given us deep insights into the key drivers of success and critical considerations for enterprise-wide adoption.
With the rapid proliferation of Generative AI and the growing discourse around AI agents versus Agentic AI, we are leveraging our lessons to establish a robust framework. This includes platform and infrastructure readiness, regulatory compliance, security and governance, business case validation, tech resource development, and workforce skill development, ensuring the right environment for sustainable scale and success.
Specifically on AI agents, we primarily leverage AI solutions embedded within enterprise-grade platforms like Workday and ServiceNow. This allows us to benefit from continuous product evolution and quality outcomes without heavy internal overhead. The result is measurable productivity and efficiency gains at the task and sub-process level.
Given the pace of innovation in underlying models and the broader AI ecosystem, we believe it is more strategic to capitalize on investments made by leading software providers rather than building proprietary AI agents from scratch. Current use cases span IT Service Management (ITSM), HR processes, transversal knowledge search and retrieval, and cybersecurity, delivering value across the organization.”
CIO&Leader: What’s the most advanced AI agent you’ve deployed? What tangible business impact has it delivered so far?
Debanjan Banerjee: Some of the most impactful AI agents are deployed in HR and Operations. These function as intelligent knowledge bots, supporting employee onboarding and offboarding, automating routine tasks, and orchestrating workflows for standard requests.
In parallel, AI agents in ITSM streamline ticket resolution and automate multi-party actions based on predefined rules, effectively completing sub-processes without human intervention. While it’s difficult to single out one agent as ‘the most advanced,’ the collective impact has been significant: cycle-time reduction, improved employee experience, cost optimization, and more consistent service delivery. These agents also indirectly reduce operational risk.”
CIO&Leader: Where do you draw the line on autonomy? Which decisions can AI agents make today, and which are off-limits?
Debanjan Banerjee: Today’s AI agents handle workflow routing, knowledge retrieval, task automation, and some sub-process orchestration. We are shaping conditions for Agentic AI gradually, starting with low-risk actions to build confidence while refining governance frameworks.
“Today’s AI agents augment human decision-making rather than replace it, starting with low-risk actions while we refine governance frameworks.”
Key considerations include interpretability, traceability, disruption management, ethical AI, and continuous learning. As trust grows, autonomy can extend to more complex tasks, but human quality control remains essential. At present, AI primarily augments human decision-making rather than replacing it, a strategy reminiscent of how RPA evolved, where governance and iterative scaling were crucial.
CIO&Leader: What level of risk have you delegated to AI?
Debanjan Banerjee: Our deployments in Sales and Marketing have delivered measurable business impact, top-line growth, optimized marketing spend, and operational efficiencies. In operations, AI has improved inventory management and supply chain planning, directly influencing P&L and balance sheet through better demand fulfillment and resource utilization.
However, we have not yet delegated high-stakes financial or reputational decisions to autonomous AI agents. We are preparing by building governance frameworks, controls, and risk thresholds to ensure AI agents can extend traditional AI frameworks responsibly.
CIO&Leader: Can you trace and explain autonomous decisions? Could you explain AI decisions to a regulator tomorrow?
Debanjan Banerjee: We are preparing for explainability and accountability in autonomous decision-making. Lessons from past AI deployments, including bias detection, model drift management, and interpretability, have been operationalized by our data science teams.
“Our goal is ‘trustworthy autonomy’—internal stakeholders and regulators alike can verify how decisions were made, who approved the rules, and how risks are mitigated.”
Autonomous AI introduces new governance requirements. Human-in-the-loop safeguards are essential, and the right stakeholders, data scientists, tech developers, compliance officers, risk managers, and business leaders ensure decisions are traceable, rules are explainable, and accountability is clear.
Our goal is ‘trustworthy autonomy’ internal stakeholders and regulators alike can verify how decisions were made, who approved the rules, and how risks are mitigated.
CIO&Leader: Which decisions do you see machines taking over, and which will always need humans?
Debanjan Banerjee: Currently, AI agents focus on low-risk, efficiency-driven tasks where productivity gains are clear and compliance is maintained. Enterprise deployment requires careful evaluation of general-purpose versus domain-specific models, ROI, and sustainability.
“In three years, AI agents will likely manage sub-processes autonomously in HR, IT operations, and supply chain planning, while humans retain oversight for strategic, ethical, and high-stakes decisions.”
Looking ahead, AI agents will increasingly manage workflow orchestration, routine approvals, operational optimization, ticket routing, expense approvals within thresholds, and personalized onboarding and coaching. Strategic, ethical, and high-stakes decisions will remain human-led. In three years, AI agents will likely manage sub-processes autonomously with frameworks for explainability, auditability, and risk calibration, while humans retain oversight on strategic decisions.”
CIO&Leader: Will AI agents become a core execution layer, or will scale stall? What’s your five-year view?
Debanjan Banerjee: AI agents will scale beyond current use cases and drive enterprise efficiency. They will redefine user experience, both internally and externally. Five years out, we envision touchless, humanless execution of processes, where employees operate systems through natural language.
Foundations like infrastructure, integration, and data must connect islands of solutions—from complex ERPs to specific applications. Seamless navigation and portability will enable better insight-driven decisions, more human connections, and ultimately deliver improved products and services at the right time for every individual.