Why agentic AI presents a turning point for business resilience

Companies have achieved and continue to work towards many admirable goals with sustainable business practices. These include less wasteful product packaging, more ethically sourced materials and a responsible approach to product testing. It’s less common to see sustainable practices and business models that simultaneously elevate both the planet and the bottom line.

This is because sustainability principles and traditional business thinking are often contradictory, especially for people working in sourcing, pricing and product development. For example, if pricing models are generally focused on paying as little as possible for a minimally compliant product, it’s not easy to switch gears and create pricing and cost models for a more sustainable product with a non-linear supply chain.

Furthermore, the mechanisms required to adopt rigorous sustainable sourcing practices could be spread across multiple enterprise domains, systems and stakeholders. Teams in procurement, product development, logistics and marketing all need to act in concert. While these teams may easily handle data and business intelligence in routine situations, tackling questions around material traceability, vendor ethics and labour or trade compliance is a different challenge entirely.

What’s needed is a way to break through these entrenched boundaries. That’s where agentic AI enters the picture.

In agentics, AI systems work together to autonomously gain new knowledge and orchestrate solutions to complex problems. This creates a way for businesses to make breakthroughs. With agentic AI, AI agents are sharing information, learning and reasoning. This frees them to uncover new correlations that human stakeholders may miss.

With their autonomy, learning and orchestration capabilities, multi-agent AI models can help businesses explore patterns and discover sensitivities that challenge business-as-usual thinking. This is what is needed to successfully design low-footprint, high-recoverability, long-lifecycle, sustainable business models.

Think of these agents as custodians of specific domains (such as taxation rules or export compliance) or systems of record (such as procurement or product lifecycle management applications). They are trained to interpret and draw inferences from that domain.

But unlike traditional AI systems, which rely on predefined rules and algorithms, these systems can dynamically adapt to changes in their environment. They can learn from new data and make informed decisions with less human oversight. Multiple agents work across organizational silos to interact, mediate and deliver intuitive reasoning that is not programmatically created.

We see three ways for multi-agent systems to move the needle on sustainable and resilient business practices:

1. Understanding the cause-and-effect relationships of resources

In most modern enterprises, activities like raw material sourcing, energy use, product structure design, manufacturing planning and logistics are all managed through disparate systems and in siloed organizations.

Agentic AI systems can help businesses identify, learn and reinforce the cause-and-effect relationships across these systems and design operating processes that achieve sustainability targets.

Take a question like, “What is the trade-off on ethical wages and the price of a part on a handbag?” The data exists within various domains for business stakeholders to answer the question. The problem is, they are not accustomed to looking at data across silos and drawing out insights to find the answer.

AI agents, however, can confer with and correlate multiple data sources in an autonomous – not rules-based – way. Much of the burden of reasoning is transferred to these agents, reducing the need for elongated interactions among human stakeholders. Even better, the agents will likely find correlations humans will miss.

Consider another question: “At what volumes should we divert sourcing from Supplier A because of deforestation stress?” This can be answered by expert analysis on a one-off basis. But to establish controls across multiple environmental factors across large product portfolios? That would require agentic AI-enabled automation.

2. Gaining proactive insights in the supply chain

Companies typically rely on periodic audits and third-party reporting to ensure supply chain resilience. However, these isolated efforts result in a disjointed view of potential disruptions. With fragmented audit, inspection and certification processes, it’s difficult to get a holistic view of the supply stream, or the potential infractions or stresses that could result from changing geopolitical issues.

Multi-agent AI could be deployed to monitor systems and transactions to recognize patterns, thresholds and anomalies. With these insights, businesses could refine the supply base and cultivate trusted chains through better and proactive scenario analyses. For example, an AI agent trained to interpret satellite imaging could confer with an agent managing the procurement of agricultural produce to alert supply chain managers of deforestation concerns.

3. Reducing the burden of multi-dimensional compliance and governance

Strong compliance and governance frameworks promote trust, integrity and long-term sustainability. However, companies often contend with multiple intersecting frameworks, reporting models and audit trails. As a result, the same record can present itself in multiple compliance frameworks, and vice versa – most compliance frameworks require data to be assembled across multiple systems of record. Reporting on labour compliance, for example, would span export control, sustainability, intellectual property protection and diversity reporting.

With agentic AI, agents could work together to ensure the data sets are semantically compiled into the right kind of compliance and disclosure models, thus automating a large part of the external disclosure and reporting effort.

For example, data sets on environmental or material safety often need to be interpreted across multiple statutory models and consumed by many company stakeholders. Agentic AI would reduce the dependence on expert personnel and provide contextual answers for people to use information in the context in which they need it.

In addition, businesses could set up observance mechanisms to highlight discrepancies and data or audit gaps that would otherwise lead to non-compliance or reputational risk.

Agentic AI can help businesses transcend entrenched thinking that hinders their sustainability efforts. Multi-agent systems offer a transformative opportunity to design operating processes that can scale up to self-fulfilling, sustainable business models capable of delivering benefits to the top and bottom lines.