Åström's own 1965 introduction claims his control-theoretic result generalizes to other decision problems, offering queuing theory as an example

control-theorypomdpkarl-astromhistory-of-sciencecross-domain-bridgellm-agents

Immediately after stating his paper's control-theoretic motivation (see claim-astrom-1965-paper-uses-control-vocabulary-not-agent-language), Karl Åström writes, in his own introduction: "Although the problem has arisen from study of control systems, its solu-tion may have applications in other fields, In queuing theory, [to] mention one example, it will thus be possible to treat queues so that service is made to depend on the past status of the queue."

This is a genuine self-authored anticipation of broader use, not one read into the text by a later reader: Åström explicitly recognized that the mathematical structure he built — optimal action selection under a hidden Markov state, given only noisy observations — generalized past its originating domain of industrial process control. He reached for queuing theory, an established branch of operations research, as his own example of where else the result might apply. The word "agent," and any framing resembling artificial decision-making, is absent from this sentence and from the rest of the paper.

The distinction matters for how the vault reads the later chain from A.A. Fel'dbaum through Åström into Kaelbling, Littman & Cassandra's 1998 canonization (claim-kaelbling-1998-cites-astrom-1965-as-pomdp-origin) and on into 2026 LLM-agent formalizations (claim-2026-comap-paper-formalizes-llm-agent-as-pomdp): the generalizing move was not invented by later readers projecting AI relevance backward onto a narrow control paper. Åström made the generalizing claim himself, in his own second paragraph, in 1965 — he simply did not have, and did not need, the word "agent" to state it.