Agentic AI at scale: AWS, Deloitte and Toyota – News

As organizations push to translate AI investments into tangible operational value, attention is increasingly shifting from experimentation to execution. Agentic AI has gained momentum because it extends beyond isolated models or content generation tools, enabling systems that can plan, reason and take action across interconnected business processes. Unlike earlier AI approaches, agentic AI operates across multiple data sources and systems, supporting continuous decision-making rather than one-off predictions.

At AWS re:Invent 2025, Rima Olinger, director of Amazon Quick Suite at AWS; Chris Jangareddy, managing director and partner at Deloitte; and Jason Ballard, vice president of digital innovations at Toyota Motor North America, discussed how this new AI paradigm is reshaping supply chain operations. Their conversation focused on reducing manual effort, improving planning accuracy and building adaptive processes capable of responding to disruption. The discussion underscored why agentic AI is becoming foundational to enterprise transformation — and what it takes to deploy it responsibly and at scale. (* Disclosure below.)

Understanding the state of agentic AI and its challenges

Despite growing interest, many enterprises struggle to clearly define agentic AI and distinguish it from traditional automation or generative AI deployments. AWS described agentic AI as systems that can interpret business objectives, reason across constraints, interact with multiple data environments and execute actions autonomously. While this expanded capability unlocks significant value, it also introduces new layers of complexity.

At TMNA, supply and demand planning highlighted these challenges. Core planning processes depended on more than seventy interconnected spreadsheets, manually assembled each month by dozens of planners. This reliance on fragmented data and legacy workflows is common across enterprises and creates friction that limits the effectiveness of agentic systems without broader modernization.

Organizational readiness presents an equally significant hurdle. Agentic AI influences decision-making, customer engagement and employee workflows, requiring trust, transparency and clearly defined governance. AWS emphasized that technology alone drives only part of the transformation. Enterprises must also prepare their workforce by clarifying how AI decisions are made, when human oversight is required and how accountability is maintained. Without this alignment, adoption often slows due to uncertainty or resistance.

Another persistent issue is the gap between pilots and production. Many organizations demonstrate early success with agentic AI prototypes but struggle to scale them across the enterprise. Deloitte noted that the absence of standardized architectures, risk controls and repeatable operating models often prevents promising use cases from advancing beyond experimentation.

How AWS and Deloitte addressed these barriers

AWS and Deloitte have developed a joint approach designed to help enterprises move agentic AI from concept to production. AWS provides the secure infrastructure, foundational AI services and agentic capabilities needed to support multi-step reasoning and orchestration. Deloitte complements this with domain expertise, transformation frameworks and operational execution that align AI initiatives with business outcomes.

Deloitte outlined a structured, three-phase methodology. The first phase prioritizes identifying high-impact use cases capable of delivering measurable returns within six to twelve weeks. This ensures agentic AI efforts are grounded in business value rather than exploratory experimentation.

The second phase introduces a multi-agent system accelerator, co-developed with AWS and integrated with Amazon Bedrock and AgentCore. This environment incorporates governance and compliance from the outset. Deloitte collaborated with AWS to establish one hundred controls using AWS Audit Manager, enabling enterprises to safely test, evaluate and operationalize agentic AI workloads.

The third phase focuses on value realization. Use cases are measured against defined performance indicators such as efficiency gains, forecast improvements and operational resilience. Only initiatives that meet established thresholds are expanded, ensuring disciplined scaling.

Together, these elements provide enterprises with a repeatable foundation for deploying agentic AI securely and at scale. AWS emphasized that the partnership also includes joint investment and specialized AI resources, allowing customers to move from concept to impact in weeks rather than years.

Applying the methodology at Toyota Motor North America

TMNA set out to rethink its operations across the supply chain, with goals centered on responsiveness, planning accuracy and improved employee and customer experience. Deloitte’s approach aligned closely with these objectives by embedding agentic AI directly into end-to-end workflows rather than layering it onto existing processes.

The engagement began with a detailed assessment of TMNA’s challenges, including lengthy planning cycles, manual data aggregation, demand volatility and limited flexibility in responding to disruption. AWS and Deloitte worked alongside TMNA to establish an agentic architecture consisting of a standardized platform layer, an intelligence layer for AI models, an agentic foundation for managing agents and an experience layer designed for planners and operations teams. This structure allowed TMNA’s business units to operate within a consistent, secure framework rather than developing isolated solutions.

Agentic workflows introduced an AI-enabled “companion” for planners, capable of generating recommendations, simulating scenarios and continuously learning from outcomes. TMNA emphasized that this approach was designed to augment, not replace, human expertise. By eliminating spreadsheet-driven coordination, planners gained broader operational visibility and could focus on complex decision-making and strategic analysis. Both Deloitte and AWS stressed the importance of maintaining human oversight, particularly in situations requiring contextual judgment.

Workforce enablement was a core component of the initiative. AWS highlighted its enterprise AI training programs, while the alliance with Deloitte ensured TMNA teams developed the skills and confidence needed to adopt agentic workflows responsibly.

Results and broader lessons for enterprises

TMNA’s adoption of agentic AI has delivered measurable operational gains. In supply chain planning, the transition from a spreadsheet-heavy process involving forty to fifty planners to an agent-assisted model represents a substantial simplification. Over time, TMNA expects planning activities to be managed by a smaller group of planners who oversee broader scopes of responsibility, reflecting role elevation rather than workforce reduction.

Forecast accuracy improved by approximately 20%, while planner productivity increased by 18%. Agent-driven simulations also introduced proactive, self-healing capabilities. Rather than responding reactively to disruptions, TMNA can anticipate issues and receive automated recommendations to maintain operational continuity.

Customer-facing processes also benefited. Vehicle delivery ETA workflows evolved from legacy green-screen interfaces and manual system switching to a modern agent-driven experience. Agents now identify delays, resolve routine issues independently and escalate exceptions when human judgment is needed, improving both employee experience and customer satisfaction.

TMNA noted that progress would have been slower without the combined strengths of AWS’s AI platform and Deloitte’s transformation expertise. Shared ownership, shared risk and shared outcomes were central to accelerating results.

Our perspective

The collaboration between TMNA, AWS and Deloitte highlights several principles enterprises should consider when pursuing agentic AI. Successful initiatives begin with clearly defined business problems, not technology-first experimentation. Agentic AI delivers the greatest value when applied to operational bottlenecks, data-intensive workflows and repetitive coordination tasks.

Equally important is early and continuous workforce involvement. Positioning AI as a trusted collaborator — rather than a replacement — helps build confidence and adoption. Enterprises must also establish standardized architectures with embedded governance, security and compliance to ensure scalability and trust.

Starting with narrowly scoped use cases enables organizations to validate value, refine controls and build momentum before expanding to more complex multi-agent systems. Continuous measurement and iteration are essential, as is close collaboration across technology, operations and leadership.

When these elements come together, agentic AI becomes more than an innovation initiative. It evolves into a durable engine for operational resilience, efficiency and long-term enterprise modernization.

Feel free to reach out and stay connected through robs@siliconangle.com, read @realstrech on x.com and comment on our LinkedIn posts.

(* Disclosure: AWS and Deloitte sponsored this segment of theCUBE. Neither AWS nor Deloitte have editorial control over content on theCUBE Research, theCUBE or News.)

Photo: News/AI