How to Move From AI Experimentation to Autonomous Supply Chain Operations

Article

August 4, 2026

Most organizations use AI to some extent in their supply chain operations, but far fewer have integrated it in ways that support autonomous decision-making.

Closing that gap is more about sequencing and integration than technology limitations. Moving toward an autonomous supply chain requires connecting AI to the ERP systems, product lifecycle management (PLM) platforms, and business processes that drive actual decisions; building visibility that reaches past tier-one suppliers to the parts and materials level; and mapping which exposures matter most before attempting to cover everything.

Kaitlyn Huissen, Vice President of Supply Chain Intelligence at Exiger, outlines what that progression looks like in practice with SupplyChainBrain managing editor Russell W. Goodman at the Orlando 2026 Gartner Supply Chain Symposium/Xpo.

Key Takeaways

Evaluating an AI investment comes down to one question: is this tool connected to the data our decisions require? That framing — rather than “can this tool answer our questions” — leads to different procurement decisions and different outcomes.
Most organizations already have the data they need to answer disruption questions. The problem is accessibility: it sits in spreadsheets and vendor surveys, not in a form teams can query at speed.
AI delivers its clearest value in supply chain by providing ontology — the structured connections between parts, products, suppliers, and risk factors — that converts a general question into an organization-specific answer.
Leaders who attempt to map their entire supply chain before taking action delay progress. Starting with the suppliers and materials that carry the most critical business impact is more productive and produces faster results.

The Difference Between Using AI and Integrating It

Autonomous supply chain operations do not require a different AI model. They require a different relationship between AI and the business processes that generate supply chain decisions.

The most common current pattern is using AI tools — ChatGPT, Claude, and similar systems — to answer specific questions on demand. An individual submits a question and receives an answer. That approach is useful for research and exploration, but it does not change how supply chain functions run, and each request remains isolated. The AI has no connection to the organization’s operational data, supplier records, or decision workflows. The answer is only as current and specific as the question.

Kaitlyn Huissen described beginning AI efforts in Supply Chain Readiness in the Autonomous Era to SupplyChainBrain managing editor Russell W. Goodman this way: “Many are using AI in a very immature way today. They’re asking a specific question for a specific answer rather than integrating it into their business process.”

“Many are using AI in a very immature way today. They’re asking a specific question for a specific answer rather than integrating it into their business process.”

- Kaitlyn Huissen, Vice President of Supply Chain Intelligence at Exiger

The organizations that have moved furthest toward autonomous supply chain operations, she observed, are those that have built what she called “decision intelligence integrated into their supply chain functions” — AI that is connected to the systems and data those functions already use, rather than operating alongside them.

What Autonomous Supply Chain Integration Requires in Practice

Autonomous supply chain integration means connecting AI to the systems where supply chain data lives and decisions get made. Huissen identified ERP and PLM connectivity as foundational since they contain the operational data that can answer what parts are in production, which suppliers provide them, what materials those parts require, and where those materials originate. A query tool without access to that data returns general information while an integrated system returns answers specific to the organization’s actual production line.

Supply chain leaders need to adjust the evaluation criteria for any AI investment. “Can this tool answer our questions” is the wrong test. The right one is: “Is this tool connected to internal and external data needed answer our questions.” Those two questions lead to different decisions about what to buy, how to implement it, and what to build first.

For teams currently experimenting, the next step is identifying which business processes generate the supply chain decisions that carry the most risk — tariff exposure analysis, supplier qualification, compliance screening — and determining where AI can connect to the data those processes depend on.

Building Visibility That Can Answer Parts-Level Questions

When disruptions hit — a new tariff, a trade restriction, a conflict affecting a key sourcing region — supply chain teams need to answer a specific question quickly: which of our parts, materials, and suppliers are affected, and what does that mean for production? Many organizations find that answering that question takes far longer than it should, not because the data does not exist, but because it is stored in formats that require manual aggregation to use.

The 232 tariffs on aluminum and steel made that problem painfully visible. Country-of-origin data existed inside most organizations. It was buried in spreadsheets and vendor surveys — present but not queryable. “When the CFO asks, what is the real impact to our business today, it took them weeks to get to that answer instead of hours,” Huissen observed. “Every hour, every minute that parts are not being manufactured on the floor is cost to top line.”

Closing that gap requires extending visibility past tier one suppliers to the parts and materials level, and structuring that data so it can be queried when a new disruption changes the question. Neither is a one-time project — trade conditions, supplier networks, and compliance requirements change continuously — which is why autonomous supply chain systems are built around living data rather than periodic audits.

How AI Brings Context to Supply Chain Decisions

Once supply chain data is integrated and accessible, AI provides the ontology — the map of relationships connecting parts, products, suppliers, materials, and the risk factors that determine how a disruption affects the business.

Huissen described it this way: “Whether you’re chasing spend visibility, sustainability initiatives, or the impact of tariffs, it’s really about what parts, what products, and what materials are ultimately crossing your production line. And so the value of AI is bringing that ontology and bringing that context to the supply chain questions that we’re asking.”

The difference between that and a standard query tool is context persistence: the system already holds the relationships, so each new question produces an organization-specific answer rather than a general one.

Start With Criticality, Not Comprehensiveness

When beginning to build toward autonomous supply chain operations, the instinct is often to start with a comprehensive map — cataloguing every supplier, every part, every material — before taking action, but it delays the outcomes the organization is trying to achieve.

A complete supply chain map is a useful long-term objective but a poor starting condition. With hundreds of suppliers and tens of thousands of parts, attempting full coverage before directing any intelligence toward risk means the highest-impact exposures wait alongside the lowest-impact ones.

Huissen identified this as the most common mistake she observes at the executive level: “They assume that they need to map everything to understand their level of risk within their organization. It’s not about mapping everything. It’s mapping those things that are most critical for your organization.”

Start with a focused question: which suppliers, parts, and materials would cause the most significant disruption if they became unavailable or affected by a new restriction? Building visibility and integrating AI around those exposures first produces faster results and a foundation for extending coverage over time.

That sequence applies whether the organization is beginning a visibility initiative, evaluating an AI investment, or designing a supplier risk management program. Supply Chain Command: What CSCOs Need Before the Next Alert Hits examines how prioritizing criticality connects to a faster, coordinated response when a disruption occurs.

Unlock the Next Frontier of Autonomous Execution

1Exiger^AI is the AI-native operating network for supply chain execution. It combines Exiger’s proprietary data with an organization’s internal systems to create a continuously updated view of the supply chain — mapped to the parts, products, and suppliers that matter most. Teams use it to accelerate procurement, reduce costs, and multiply capacity across risk assessment, compliance screening, document review, workflow execution, and supply chain modeling, routing work to human reviewers, AI teammates, or both.

Get in Touch

Learn how you can build a more resilient supply chain.

Frequently Asked Questions

In an autonomous supply chain, AI is connected to the operational data — parts, suppliers, materials, risk factors — that decisions depend on. It doesn’t wait to be asked a question; it surfaces what changed and what it means for that specific organization. Most organizations using AI today are doing the opposite: submitting questions to a tool that has no connection to their systems and receiving general answers in return.

Ontology is the map of relationships that gives AI organizational context. In supply chain, that means knowing how a finished product connects to the components inside it, which suppliers provide those components, where the materials originate, and which risk factors — country of origin, regulatory status, spend category, compliance flags — apply at each level. Without that map, AI answers supply chain questions generically. With it, the same question returns an answer specific to what that company makes and sources.

Most organizations have more supply chain data than they can use. The limiting factor is whether that data is reachable — connected to the systems and workflows where decisions get made — or sitting in spreadsheets and vendor surveys that require manual effort to interpret. When a disruption hits, the difference between hours and weeks often comes down to how the data is held, not how much of it exists.

Tier-one visibility identifies which direct suppliers are affected. Sub-tier visibility identifies which production lines, components, and cost lines are at risk — and why. That’s the difference between knowing a supplier is in an affected region and knowing which of your parts are in jeopardy, at what volume, and what it costs per hour to not have them. See also: How Dependency Warfare Is Rewriting Supply Chain Risk.

Comprehensive mapping treats all suppliers and materials as equally worth understanding before taking action — which means the highest-impact exposures wait in the queue alongside the lowest. CPOs who’ve moved furthest toward autonomous operations picked a defined set of critical suppliers and materials, built visibility and AI connectivity around those first, and expanded from a working foundation rather than a finished one.

Which of our active components are sourced from a newly tariffed country? Which tier two suppliers in our critical parts list carry a sanctions or forced labor flag? Which materials are single sourced from a region now under export restriction? If those questions still require an analyst to pull from spreadsheets, the integration work that enables autonomous supply chain operations hasn’t been completed.