Why Your Investing Agent Should Read Filings From Companies You Do Not Own
September 5, 2026
Altsets
Research by Altsets Research
Use economically connected customers and suppliers to build a selective external-document research queue around portfolio holdings instead of reading only the filings of owned companies.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- Apple at 8.75% of SK Hynix revenue gives an SK Hynix investor a relationship-based reason to follow Apple disclosures even when Apple does not quantify the supplier relationship in its own filing.
- An agent can use relationship materiality to choose which external filings deserve review, then keep company filings, supplier lists, Altsets metrics, and inference as separate evidence layers.
A fundamental investor usually reads the filings of companies they own. A supply-chain-aware investing agent can widen that document universe selectively by reading filings and disclosures from economically connected companies whose results can change the thesis.
SK Hynix and Apple illustrate the idea. Altsets associates Apple with 8.75% of SK Hynix revenue, which means an SK Hynix investor has a reason to follow Apple disclosures even without owning Apple. Apple's own 2025 annual report discusses product transitions, component supply, limited-source inputs, supplier concentration, global manufacturing risk, and the possibility that supply constraints or price increases can affect results. Those disclosures do not quantify SK Hynix exposure, but they can change the context around a customer relationship that Altsets already identifies as economically meaningful.
The relationship graph decides which external filings deserve time
Without an economic map, an investor can easily end up reading dozens of adjacent companies because they share an industry label. The relationship data creates a better filter. If a customer represents a meaningful portion of supplier revenue, the customer's filing can deserve more attention than a larger company that merely operates in the same sector.
This is where an agent can save real research time. It can start with the portfolio, retrieve important counterparties, and build a document queue that is specific to the holdings rather than a generic list of industry filings.
External filings answer questions the holding's filing cannot
A supplier can explain its own revenue, margins, capacity, and customer concentration, but it cannot fully disclose the customer's product roadmap, demand assumptions, capital allocation, or risk factors. The customer filing can provide that second viewpoint, even when it never names the supplier.
For an SK Hynix investor, Apple disclosures about component availability, product transitions, sourcing risk, geographic manufacturing exposure, and demand conditions can all be relevant to the Apple relationship. None of those statements should be converted automatically into an SK Hynix forecast. Their value is that they help the investor understand what can change on the other side of a material customer relationship.
Public supplier documents can fill a different evidence gap
Apple's historical supplier list separately names SK hynix and identifies Apple-related manufacturing locations in China and South Korea. That document does not provide current relationship percentages, but it supplies external evidence that the commercial connection existed.
An agent can keep those evidence layers separate. The relationship dataset supplies current economic measurements, the customer filing supplies company-level business context, and the supplier list supplies historical relationship evidence. A good research summary states which claim comes from which layer rather than blending them into one source.
The agent should search for changes, not summarize entire filings
A 10-K can contain hundreds of pages that have no bearing on the portfolio thesis. The agent becomes more useful when it reads with a relationship-specific question, such as whether the customer changed its sourcing language, product-transition risk, geographic exposure, inventory posture, or expected demand.
That is different from asking for a generic filing summary. The agent already knows why the external company matters, so it can focus on information that could alter the economic relationship or the assumptions the investor is making about it.
Connected-company reading can work in both directions
A supplier investor can read customer filings for demand context, while a customer investor can read supplier filings for capacity, pricing, order patterns, and technology transitions. The useful direction depends on the thesis.
KLA's fiscal 2026 annual report, for example, discusses a highly concentrated customer base and says customer investment patterns can affect orders, revenue, and margins. A Micron investor who knows KLA is a mapped supplier can use that disclosure as upstream context, while a KLA investor can use Micron disclosures as customer-demand context. The relationship tells the agent why the other company's documents belong in the research set.
An agent can maintain a document graph instead of a ticker list
The practical advantage is not that the agent reads more documents. It is that it can maintain a structured connection between each document and the portfolio thesis. A filing from an unrelated company stays out of the queue, while a filing from a material customer or supplier can be tagged to the holding and the specific relationship that made it relevant.
That makes the resulting research easier to audit. The investor can see that a conclusion came from the holding's filing, a connected company's filing, an Altsets relationship, or an inference made after comparing them.
This workflow is especially useful when disclosures are asymmetric
One company can disclose a relationship in detail while the other barely mentions it. A customer can name a supplier in a public list while the supplier reports only aggregate customer concentration. A supplier can discuss product shipments while the customer discusses only broad sourcing risk.
The supply-chain data versus company filings guide explains why no single document contains the full dependency picture. An agent can make the fragmented evidence easier to work with without pretending the fragmentation disappears.
The research queue should remain economically bounded
The point is not to read every filing from every node in a large graph. The agent should prioritize based on relationship materiality, thesis relevance, and the type of event under investigation. Structural-only relationships can be retained as lower-priority context, while quantified relationships can receive deeper monitoring when the exposure is economically significant. The Altsets documentation explains how relationship data can be retrieved for this workflow, while the LLM grounding guide covers the evidence discipline needed when an agent combines proprietary and public information.
For relationship definitions and evidence limits, read the Altsets methodology. Browse Supply-Chain Data Use Cases for other fundamental-investing research workflows.