The shelf-life problem
A conventional systematic review is a snapshot. It searches the literature up to a date, synthesises what it finds, and ships. From that point the review and the evidence drift apart: new studies appear, effect estimates shift, and the conclusions slowly age out of step with what is actually known. For fast-moving topics the gap can open within months.
The usual fix — commission an update every few years — is expensive, slow, and still leaves long windows where the best available synthesis is stale. The deeper problem is structural: the review is treated as a finished artefact rather than a system that is kept current.
What 'living' actually means
A living evidence system keeps the synthesis connected to its sources so that new evidence flows back into it. Studies, entities, and topics are held in a structured knowledge graph rather than frozen into a PDF, so when something new arrives you can see exactly which syntheses it touches and what changed.
The shift is from 'publish and move on' to 'maintain and surface'. The same infrastructure that produced the synthesis is the infrastructure that updates it — and the audience sees the current state, not last year's snapshot.
Why structure beats documents
Documents are where knowledge goes to stop being usable. A claim buried in a PDF cannot be re-queried, recombined, or traced to its source by a machine. The same claim held as a node in a knowledge graph can be linked, updated, and audited — which is what makes continuous updating tractable rather than a heroic manual effort.
It is also what makes the evidence legible to the AI answer engines that an increasing share of readers now ask first. Structured, source-linked knowledge is far easier for those systems to surface and cite accurately than prose locked in a report.
Responsible AI is what makes it sustainable
Keeping evidence current at scale is only realistic with AI doing some of the lifting — monitoring for new studies, flagging what changed, drafting updates for human review. That only stays trustworthy if the AI is used responsibly: validated, transparent, and kept under human oversight at the decisions that matter.