A Field of Practice
The structure underneath an organization's decisions: who can decide, what they can see, and what the organization remembers. DRI is the discipline of designing that structure, diagnosing where it has failed, and keeping it working.
Developed by Justin R. Greenbaum, a customer operations executive with more than twenty years leading care, technical support, and service transformation, and run as a practice at Greenbaum Labs. Pick the door that matches what you came for; the method and the research record are further down the page.
Decision & Responsibility Infrastructure® is the architecture of authority, accountability, and continuity that sits beneath organizational performance.
Training people to decide better does little when the structure around them will not let a good decision stand. DRI works on that structure: the conditions that let a sound decision get made, hold, and be remembered.
In practice that means who has the authority to decide, who carries the consequences, how decisions are remembered, how responsibility is assigned without being spread so thin that nobody holds it, and whether people can say what is actually happening without paying for it.
Organizations often depend on people to compensate for gaps in the structure: an agent finds a workaround, a manager carries a decision nobody clearly owns, a team keeps a broken process moving by hand. As more of that work is automated, those gaps become harder to hide.
The same three breaks show up across companies, sectors, and sizes. Each is easy to recognize once it has a name, and each comes with something specific to examine.
Authority separates from responsibility. A team is accountable for an outcome but needs approval for the decisions that drive it. Work queues behind senior calendars, the people closest to the problem stop proposing fixes, and the senior team becomes the default place where things get resolved. Worth examining: for the last outcome you assigned an owner, which decisions did they actually control?
Decisions are forgotten. The reason behind a policy leaves with the person who made it. The next team inherits the rule without the context, keeps it because nobody is sure it is safe to remove, and relearns the original lesson the expensive way. Worth examining: pick one standing rule and ask who can explain why it exists.
Reality gets expensive to say. The dashboard is green while the people doing the work know something is slipping. Raising it means a review, a name on a report, and a reputation for not handling things, so it does not get raised. Worth examining: what is happening right now that none of your metrics would show?
None of these needs bad intent. They are structural, which is why they survive reorganizations and new leaders, and why they get sharper once decisions about AI systems are added to the same structure: who can change a model's behavior, who answers for its outcomes, and whether the commitments made about it outlast the people who made them.
The diagnostic method for DRI™ is Coherence — a system for identifying, naming, and correcting structural failure in human systems.
Coherence diagnoses organizational breakdowns through three structural vertices:
Is reality being accurately perceived and spoken? Can the system see what is actually happening, and can participants name it without distortion, fear, or political cost?
Does decision-making power match responsibility? Are the people with the power to decide the same people who bear the consequences?
Are commitments, decisions, and narratives stable over time? Does the system remember what it decided and why?
When the three hold together, the organization can decide, act, and remember. When one drifts, people make up the difference by hand, and that cost rarely shows up on a report.
The same structural conditions look different depending on where you sit. Pick the chair closest to yours.
The taxonomy has two kinds of entries. Failure Modes are conditions that feed themselves and do not resolve without a change in structure. Functional Field Notes are the smaller, recurring conditions that usually show up first, before anyone would call them a failure.
Seventeen failure modes and twenty-one field notes, each with a definition, what it looks like from inside, and why it emerges. The failure modes are grouped into three tiers by where they sit in the chain of cause and effect.
Seven browser-based instruments. Each one observes, summarizes, and suggests; none of them decides, and nothing you paste leaves your device. The FM Scanner is the place to start, and it has a worked example built in.
AR-001 — Automation may observe, summarize, and suggest. Automation may not decide.
Detect failure mode patterns in organizational signals.
Surface early warning signals before failure becomes entrenched.
Orient on the diagnostic map. Identify where you are in the system.
Record when authority acted. Preserve decision rationale over time.
Prove fixes are real. Confirm structural change, not cosmetic adjustment.
Test if fixes hold over time. Monitor for structural decay.
Custody of meaning across time. Track how decisions survive handoffs.
The Coherence Record is the build record of DRI™ — published at the pace the work moves. Each edition documents what the diagnostic pipeline has found, what the methodology can and cannot see, and where the limitations are.
135 research runs of the diagnostic pipeline against public evidence on 31 organizations across six sectors. These are tests of the instrument, not client engagements. FM-01 (Responsibility Compression) appeared in 14 of the 15 organizations in the Edition 3 fleet read. The method's limitations are published alongside its results.
Building in the open means publishing failures alongside successes. Edition 4 disclosed that running the same organization through the instrument twice could produce different scores (a standard deviation of 0.114 on its primary dimension), which meant the scores were not yet measurements. The scoring was rebuilt, and the same inputs now produce the same result every time. That is repeatability, not proof that the findings are right, and the Record says so.
DRI™ is developed and advanced by Justin R. Greenbaum, powered by the infrastructure at Greenbaum Labs. The diagnostic method — Coherence — is published regularly on Substack, where Failure Modes and Field Notes are explored in depth.
Decision & responsibility infrastructure. Enterprise operations, applied AI, and the Coherence framework.
An independent advisory and applied-AI practice. Its Lab builds and runs the diagnostic pipeline.
Essays on coherence, decisions, and building infrastructure for what matters. The Coherence Record lives here.
The work is public. Explore the assessment, the tools, the research record, or the writing behind DRI.
Nobody notices infrastructure until it fails. The point of this work is to find the failure while it is still a question for a staff meeting rather than a crisis.