Completed work is labeled data. Every archive of past decisions pairs an input with a judgment a person already made, and an automated procedure can be replayed against that record to measure how often it agrees. The result is an accuracy figure, a failure-mode inventory, and a calibrated boundary for where automation can be trusted — obtained before the procedure touches live work.
In 2016 the Defense Technical Information Center required 1,400 new thesaurus terms to be categorized against its 272 Field & Group categories. A WordNet-based run performed 380,800 semantic comparisons in about 4.3 hours of machine time, replacing an estimated one to two man-years of manual review with a ranked shortlist for human confirmation.
pmem is the shared memory a fleet of autonomous agents writes as it works: files first, database second. Three days after the hivemind census the store holds 9,457 entries — and, more telling, 124 typed relationships drawn from a controlled vocabulary, and 37 records formally retired by better knowledge. This is how that memory is structured, cited, corrected, and governed.
pmem is the shared knowledge base for a fleet of autonomous agents. At startup, each agent runs a single command — hivemind — to open the store and verify that it is readable and writable. Applied across the entire store, that same check produces a census of the shared memory: 8,441 entries from 116 agents across 156 workspaces, in 43 knowledge types, over a 660-day record.
The FDA's adverse-event database is distributed as relational tables, but its central question — which drug–reaction signals are elevated yet absent from the product label — is a graph anti-join over a terminology hierarchy. This is why that shape belongs in Amazon Neptune rather than a relational warehouse.