Understanding supply chain relationships is a critical part of assessing a company or industry, but often that information is disjointed and missing context. A company may disclose its major suppliers without explaining which components each one provides or which products depend on them. For a financial analyst assessing supply chain exposure, that leaves important questions open. Does a supplier provide a widely available component or a specialized part that would be difficult to replace? Where does that part originate, and what happens to the customer’s production if supply is interrupted?
Answering those questions usually means piecing together evidence from different sources. A filing may establish a supplier relationship, while an earnings transcript discusses a change in sourcing. Expert information may provide more detail about the component involved. The analyst has to connect those findings and determine whether they describe the supply chain as it operates today.
That work becomes particularly important when assessing companies with just-in-time inventories. With limited stock to absorb an interruption, the availability of a single component can become a concern for an entire product line. Understanding the potential exposure requires tracing the component through the supplier network. Assessing alternatives requires further research into other suppliers and what they actually provide.
AlphaSense’s knowledge graph now includes supply chain relationships, connecting suppliers and customers to components, products, and countries of origin. These details are captured directly from the source material. Analysts can investigate these relationships through SuperAnalyst, AlphaSense’s always-on agent. They can ask supply chain questions in plain language, review the sources supporting the analysis, generate interactive visual dashboards, and follow up on a particular supplier or component. SuperAnalyst draws on the graph alongside document research and market data to explain the dependencies behind a company’s potential exposure.
How the Knowledge Graph is Built
AlphaSense continuously ingests public and AlphaSense proprietary documents, including filings, transcripts, and expert information. During ingestion, we extract supply chain relationships and check them against their sources before adding them to a temporal knowledge graph. The graph records when relationships were active, supporting both current exposure analysis and research into earlier periods.
As a result, SuperAnalyst can query relationships that are already in the knowledge graph, extracted and checked, rather than having to establish every connection from raw documents while simultaneously answering the analyst’s question. Two benefits of this structure are improved token efficiency (since connections aren’t established from scratch for each query) and improved recall. Because of how it's constructed, our supply chain knowledge graph identifies more than twice as many supply chain relationships than frontier LLMs operating on the same document inputs.
Tracing Exposure Through the Supplier Network
An analyst assessing Apple’s exposure to a disruption in a particular country could ask SuperAnalyst:
Which Apple products could be affected, and which components create that exposure?
When an agent builds an answer from retrieved documents, finding a reference to an Apple supplier is only the starting point. It still has to establish which component that supplier provides, where it originates, and which product uses it. With the AlphaSense knowledge graph, the agent can instead start with components recorded as originating in the affected country, then follow documented relationships to their suppliers and the Apple products that use those components. The answer can show the dependency behind a product’s potential exposure, with supporting evidence for each connection. The level of detail in Apple’s case depends on what the sources disclose. Where they identify a specific part, the analysis can follow that part from the country of origin through its supplier to the Apple product.
Once the analyst has identified a critical component in Apple’s supply chain, they can ask SuperAnalyst which other suppliers have documented evidence for the same part or component. These connections do not establish that an alternative supplier has available capacity or is qualified to provide a replacement for Apple. They simply give the analyst a more focused place to begin that diligence.
Tracking Supplier Relationships Over Time
A supplier relationship that was active in one quarter may have ended by the next. A company may add suppliers for a component, change the country of origin for a part, redesign a product, or move production to a different supplier.
A knowledge graph that stores only the existence of a relationship can therefore be misleading. Supply chain relationships in AlphaSense’s knowledge graph are temporal, allowing the graph to represent when a relationship was active and how it changed. Analysts assessing current exposure need the latest supported view of the supplier network. For research into an earlier reporting period, they can query the relationships recorded for that period.
What Makes Supply Chain Intelligence Trustworthy and Secure
An incorrect supplier relationship can distort an exposure analysis. AlphaSense combines automated testing of extraction models and prompts with source-level verification of each proposed relationship.
Before a model or prompt change goes into production, our automated experimentation framework compares candidate configurations against a golden dataset—a reference set of examples with expected results. The tests evaluate extraction accuracy for customer-supplier relationships, specific parts, and country-of-origin information. They also check for unsupported relationships and differences in performance across filings, transcripts, expert information, and private documents. This helps identify trade-offs: a model may improve part-level extraction while becoming less accurate at determining when a supplier relationship was active.
During ingestion, we use an LLM-as-a-judge approach to check the validity of each individual relationship. One model extracts the supply chain relationship. A second, independent model reviews the same source and evaluates whether the evidence supports the extraction. The relationship enters the graph only when the models agree and the evidence meets the required confidence level. Extractions whose evidence cannot be reconciled are withheld.
In addition to the verification layers, we built our supply chain knowledge graph with data security, compliance, and entitlement-aware retrieval woven in. Our knowledge graph has a policy layer built in that ensures it operates in a way that respects entitlements and compliance restrictions, so users only see things based on evidence they have access to.
Decision-Grade Supply Chain Intelligence
For the analysts assessing Apple’s exposure, AlphaSense’s knowledge graph makes it possible to follow an affected component through to the products that use it, check when the relevant supplier relationships were active, and investigate other documented suppliers. Those connections provide a starting point for assessing exposure and researching potential alternatives, with supporting evidence for each relationship.
Investigate supply chain dependencies in the AlphaSense knowledge graph. Request a free trial today.
