A Direct Comparison

Nine phases, or five stages and a foundation.

The EDRM* has served the industry since 2006, and EDRM 2.0, opened for public comment in the summer of 2026, extends analysis across the workflow and grounds discovery in a broader governance lifecycle. It is a real improvement. It also keeps the phase structure, adding AI as a capability inside each stage rather than asking which stages the underlying technology has made unnecessary.

EDRM (2006 / 2.0) The Living Discovery Model
Identification 01 Connected Corpus
Collection
Processing 02 Continuous Indexing
Preservation Foundation, not a numbered stage. Runs underneath all five.
Identification (search) 03 Retrieval-Augmented Discovery
Analysis
Review 04 AI-First Review with Validation
Production 05 Explainable Production
Presentation
Information Governance Assumed as a precondition of a governed, permissioned Connected Corpus, rather than a standalone left-side phase.
Why It Collapses

The phases were artifacts of a constraint that no longer exists.

Identification and Collection

These were two events because you had to find data before you could physically move it somewhere else. A permissioned, connected corpus does both at once, nothing moves, so finding and accessing are the same act.

Identification and Analysis

Search used to finish before understanding could start, search terms first, review second. Conversational retrieval collapses that sequence: a system can evaluate a document's relevance in the same turn it surfaces it.

Production and Presentation

A static export followed by a courtroom explanation made sense when the explanation had to be assembled by hand after the fact. When reasoning is generated alongside the document, the export and the explanation are one artifact.

Preservation

The one phase that doesn't collapse, it changes jobs entirely. Collection used to double as your defensibility anchor: copying data out kept it safe from deletion. Here, the data stays in place inside the corporation's own systems instead, so preservation has to hold two things at once, enforcing deletion holds on live production data, and logging exactly what the AI accessed. That's why it moves from a phase to the foundation everything else sits on.

A Note on EDRM 2.0
This is offered as an alternative framework, not a critique of the people building EDRM 2.0.

EDRM 2.0 was developed by roughly 150 practitioners across the industry, and it makes real improvements, extending analysis across the workflow and grounding discovery in the updated Information Governance Reference Model. The Living Discovery Model starts from a different premise: that the constraint EDRM was built to manage, data that has to move before anyone can look at it, no longer holds inside a properly governed, AI-native corporation, and that the model built on top of that constraint should not survive it unchanged.

* EDRM.net