Representative End User Clients
Representative Automation Clients
Representative Software Clients
As supply chains become increasingly complex and vulnerable to disruption, artificial intelligence (AI) is emerging as a key enabler of smarter, more resilient operations. During his session, “Steering Supply Operations in the AI Era,” at the Manufacturing Innovation Conclave during ARC Advisory Group’s 24th India Forum in Bangalore on July 9–10, 2026, Naveen Krishna Yamarti, Global Supply Chain Capability Lead, Digital at AstraZeneca, shared how the company is using AI to advance autonomous manufacturing, strengthen regulatory compliance, and build more self-healing supply chains.
The central message was clear: AI cannot transform supply chains on its own. Its success depends on connected data, integrated systems, strong governance, and an operating model capable of turning intelligence into action.
Yamarti’s presentation, “Steering Supply Operations in the AI Era,” is available on YouTube or can be watched here:
A New Reality for Supply Chains
Supply chains have always operated under uncertainty. Demand fluctuations, production constraints, regulatory requirements, logistics disruptions, and changing market conditions require organizations to make decisions with imperfect information. The objective has never been perfect prediction, but continuously improving visibility, responsiveness, and resilience.
AI is changing how organizations approach these challenges. More capable AI models and agentic technologies can automate repetitive activities, improve forecasting, identify emerging risks, and support faster decision-making. At the same time, they introduce new requirements for transparency and control, particularly in highly regulated sectors such as pharmaceuticals, healthcare, finance, and energy.
Three forces are reshaping supply operations simultaneously: AI acceleration, increasing regulatory pressure, and geopolitical disruption. Trade restrictions, regional conflicts, tariff changes, transportation bottlenecks, and shipping disruptions can quickly affect global supply networks. Organizations therefore need operating models that can detect changes early and respond before disruptions become business-critical.
Building the Digital Foundation
For AstraZeneca, this transformation supports its broader 2030 ambitions, including launching at least 20 new medicines, accelerating innovation across the drug development lifecycle, advancing sustainability, and achieving $80 billion in Total Revenue.
The challenge is significant. Developing a medicine can take 10 to 12 years, from molecule discovery through development and ultimately delivery to patients. Accelerating this journey without compromising safety, quality, or compliance requires innovation not only across research and development, but also across manufacturing and supply operations.
Yet some of the biggest barriers remain fundamental.
Fragmented data continues to constrain AI adoption. Enterprise resource planning systems, laboratory platforms, quality applications, procurement tools, manufacturing systems, and operational technology often operate in silos. Without an integrated data foundation, organizations struggle to create a consistent view of operations or generate reliable insights.
Reactive decision-making is another challenge. Traditional dashboards and reports primarily explain what has already happened. The next generation of supply chains must move toward predictive and proactive decision-making by identifying potential risks and recommending actions before they affect operations.
Finally, spreadsheet-based planning remains common across supply functions. Although spreadsheets offer flexibility, they limit scalability, consistency, and automation. Replacing manual processes with intelligent, connected systems is an important step toward more autonomous operations.
Building a Self-Healing Supply Chain
AstraZeneca’s response is reflected in its strategic vision for an AI-enabled ecosystem designed to connect development, manufacturing, and commercial supply operations.
The architecture begins with existing enterprise systems, including SAP, laboratory information management systems, procurement platforms, manufacturing applications, and operational technology environments. An integration layer connects these previously isolated platforms, enabling information to move more effectively across the organization.
Above this foundation sits an intelligence layer incorporating AI agents and orchestration capabilities. These technologies can coordinate processes, generate recommendations, and execute defined actions within established governance frameworks.
The longer-term ambition is an end-to-end digital thread connecting the journey from molecule discovery to patient delivery. Rather than treating AI as a collection of individual pilots, this approach aims to create enterprise-wide intelligence that can support decisions across the supply chain.
AI Use Cases Delivering Results
AstraZeneca is already applying AI to several operational and quality challenges.
Predictive deviation management uses AI to analyze manufacturing and quality data and identify potential deviations before they occur. In one implementation, the technology helped prevent 14 batch-related issues within a year, protecting production output and reducing operational risk.
AI-powered batch record reviews are addressing another time-intensive activity. Review times have been reduced from approximately six hours to about 40 minutes per batch while maintaining compliance requirements.
AI is also being applied to Corrective and Preventive Action (CAPA) management, helping accelerate investigations, identify recurring patterns, and support faster corrective actions.
In regulatory reporting, AI-enabled capabilities can help teams quickly retrieve and organize documentation and supporting evidence, reducing preparation effort and improving responsiveness during inspections and audits.
The next wave of initiatives extends AI into everyday operations. Shift-handover copilots can generate structured summaries for operators, while SOP knowledge assistants can provide rapid access to standard operating procedures across manufacturing sites. Anomaly-detection engines can identify abnormal operating conditions before they develop into production or quality issues. Meanwhile, supply-disruption intelligence can combine logistics visibility with AI to model scenarios involving shipping delays, transportation bottlenecks, or geopolitical events.
Governance Is the Foundation
In regulated environments, AI adoption must be accompanied by strong governance.
Four principles are particularly important: model validation, comprehensive audit trails, explainability, and disciplined change control. AI systems must be validated for performance and reliability. Recommendations and actions need to be traceable. Regulators and business stakeholders must be able to understand how critical conclusions are reached. At the same time, organizations need change-control processes that balance the rapid evolution of AI with operational stability and compliance.
Rather than viewing regulation solely as a constraint, organizations can use it as a framework for building more trustworthy and scalable AI systems.
From Reactive to Autonomous Supply Chains
AstraZeneca’s transformation highlights a broader lesson for supply chain leaders: the competitive advantage from AI will not come simply from having more AI tools. It will come from building an operating model in which data, intelligence, people, processes, and governance work together.
The emerging self-healing supply chain is not about removing humans from decision-making. It is about giving teams earlier signals, better insights, and greater capacity to focus on decisions that require human judgment.
As supply networks face increasing volatility, regulatory scrutiny, and geopolitical disruption, organizations that combine AI with strong digital foundations and operational discipline will be better positioned to move from reactive supply chain management toward predictive, adaptive, and increasingly autonomous operations.