Emergence of multiagentic AI in S&P

• Agentic AI is redefining sourcing and procurement by moving beyond content generation to execute governed, end-to-end procurement workflows.
• Procurement orchestration is becoming the new value driver, coordinating sourcing, contracting, compliance and supplier interactions across the business.
•  Multiagentic AI systems are helping transform procurement from siloed automation into connected, adaptive enterprise ecosystems.
• A strong enterprise data core is foundational to scaling agentic AI, enabling faster, more context-aware decisions across the source-to-pay life cycle.
• Successful procurement transformation with agentic AI depends on governance, human oversight, security, interoperability, data readiness and change management.

The evolution of agentic AI

The rise of large language models (LLMs) in 2022 redefined S&P, bringing intelligence and flexibility far beyond traditional automation. Elaborate tasks like evaluation of supplier-submitted RFX are now completed in minutes, with greater accuracy and relevance.

Multiagentic AI has quickly advanced from basic content generation to managing a complex, multistep procurement workflow.

In 2023, retrieval-augmented generation (RAG) allowed AI to instantly access contracts, requisitions, and invoices summarizing key terms and flagging purchase order issues using live enterprise data.

By 2024, agentic AI advanced further: Autonomous agents now execute multistep procurement tasks, from reviewing contracts and benchmarking to onboarding suppliers and monitoring compliance, automating entire workflows and alerting teams to risks or gaps within autonomous procurement workflows.

Configurable do-it-yourself (DIY) agents are also emerging, tailored to run source-to-pay (S2P) plans, analyze spend, flag compliance risks and suggest sourcing strategies aligned to company policies.

Agentic AI is not just automating tasks; it enables connected, intelligent ecosystems. The next section explores how agentic AI drives tighter integration across enterprise data and business functions, unlocking strategic value at scale.

Agentic AI impact on enterprise ecosystems

As agentic AI capabilities mature, especially with the rise of multiagentic AI systems, enterprise platforms will evolve from merely interconnected systems to individual and ultimately unified ecosystems. Initially, AI agents bridge functional silos by connecting disparate systems, such as sourcing, finance and HR. Over time, individual business functions develop their own orchestrated agent layers, enabling more intelligent and cohesive operations and interfunctional collaboration. Ultimately, these domains converge into a unified enterprise ecosystem, where a central procurement orchestration layer governs groups of specialized agents that operate effectively across sourcing, planning, risk and other areas. This shift replaces static workflows with adaptive, continuously learning systems that enhance decision-making and align with organizational goals, setting the stage for S&P to transform from transactional processes into strategic, digitally enabled value drivers. 

Source-to-pay ecosystem enabled by agentic AI

As the enterprise ecosystem is evolving, the S2P ecosystem within represents a holistic, integrated framework that encompasses the entire S2P life cycle—from identifying sourcing needs to completing supplier payments. The S2P ecosystem brings together a suite of interconnected processes, technologies, and stakeholders, designed to drive efficiency, transparency and value across every stage of S&P.

Within this ecosystem, each component plays a critical role in enabling collaboration, data-driven decision-making and end-to-end process automation.

Enterprise data core
The data core is becoming increasingly important as it grows more connected and synchronized across the ecosystem, serving as the foundation layer of the framework. The role of traditional enterprise resource planning (ERP) systems is expected to evolve, with their functions and data increasingly integrated into the enterprise data core. The data core may increasingly serve as the primary hub for enterprise data, potentially reducing reliance on traditional ERP systems and capabilities. Enterprise systems feature agents that can interact with other agents, signifying interconnectedness and collaboration across the ecosystem.

Agent groups
Agent groups are composed of individual agents, each designed to conduct specific activities within a broader workflow. These agent groups can efficiently connect to the enterprise data core, enabling them to access and retrieve data, regardless of the originating module or agent group.

The framework below illustrates eight core agent groups comprising individual agents organized to support key procurement transformation activities: sourcing, contracting, supplier onboarding, supplier risk, performance management, category management, purchasing, and invoicing and payments. Each agent group comprises agents tailored to execute specific activities. For example, contract review agent (under contracting agent group) rapidly extracts relevant contract clauses such as payment terms, termination conditions and liability from supplier documents for focused and efficient review. It streamlines contract analysis, highlights deviations, and identifies risk clauses versus organizational standards and prior agreements.

Organizations can string together multiple agents to build their own multiagentic AI ecosystems that address the organization requirements. 

S2P ecosystem orchestrator agent
The orchestrator agent will be the central coordinator that brings together the activities. The orchestrator can call different agent groups as needed, enabling seamless collaboration and end-to-end automation across the sourcing-to-procurement process. By managing the flow of information and tasks between agents, the procurement orchestration layer ensures that each agent operates in sync, leverages shared enterprise data, and contributes to a unified workflow.
S&P leaders, category managers, purchasing buyers, accounts payable and business users consume the information through the user interface powered by the orchestrator agents. The user interface can facilitate efficient interaction and decision-making across roles.

The orchestrator agent can connect with external data sources including market intelligence, supplier intelligence platforms, benchmarking services, and geopolitical analysis. This information can be used by the agent groups in decision-making. The orchestrator will be able to communicate with agents from suppliers and share specific information over a secured protocol.

Key considerations while adopting agentic AI

As organizations evaluate and plan for agentic AI transformations across S&P, leveraging frameworks available in the market, they should focus on the following considerations to enable effective and responsible deployment as part of a broader procurement transformation.

1. Governance, compliance and risk

Agentic systems make dynamic, independent decisions. Establish healthy governance mechanisms, clear accountability structures and compliance checks (data privacy, auditability, explainability) to mitigate potential risks, including bias, errors, or regulatory violations across autonomous procurement workflows.

2. Human oversight and collaboration

Define the role of humans in-the-loop or on-the-loop. It is crucial to set boundaries on agent autonomy and to establish escalation procedures, allowing agents to complement rather than replace human decision-making where needed in multiagentic AI environments.

3. Security and trustworthiness

Agentic AI may interact broadly across applications and data environments. Secure data access, protect confidential information and validate agent actions to prevent misuse, errors, or adversarial manipulation.

Transform your procurement organization through agentic AI

GenAI is positioned as a platform for data consumption and content generation—summarizing performance drivers, drafting narratives and answering questions—rather than providing authoritative solution or action recommendations. Agentic AI shifts the emphasis to procurement orchestration, coordinating multistep workflows across data, tools and stakeholders to move from insight to execution. A thoughtful framework for leveraging GenAI in procurement has been discussed in detail in our previous publication, “Transforming digital procurement through Generative AI.”¹

• Establish clear standards for interoperability that allow agentic AI solutions to connect with your existing ERP systems, procurement platforms and AI tools.

• Lay the groundwork for multiagentic AI by enhancing the quality and accessibility of procurement and other dependent data.

• Drive successful adoption by actively engaging and training key stakeholders including procurement teams, IT, finance and suppliers throughout the transformation journey.

• Define AI-first governance and accountability by establishing decision rights (including a clear responsible, accountable, consulted, and informed [RACI] matrix), risk tiers and escalation paths for agent-initiated actions across autonomous procurement workflows.

• Implement security and privacy controls including least-privilege access to ERP and procurement systems and protected handling of external data sources within the S2P ecosystem. 

Endnotes

1. Vinay Rajani and Mike Deng, “Transforming digital procurement through Generative AI,” Deloitte’s Business Operations Room blog, January 4, 2024.

Thank you to our contributors: Praveen Nemana, Prem Singh