How to spot intelligence loss

Every organization leaks intelligence — expertise, data, and judgment that never reaches the decisions where it would matter. These are the twenty-eight patterns we've identified. Pick where you want to look, and we'll show you what to watch for.

Where to look

AI-specific only

Tier 1

Context stripped in the handoff

Looks like

· Receiver spends hours rebuilding context before real work starts

· Rework traces back to info the sender had but didn't pass along

Sounds like

"I didn't know that was a factor — it wasn't in the brief"

"Can you give me the backstory?"

Tier 2

Reasoning behind decisions never recorded

Looks like

· Decision records contain only outcomes or approvals

· People reopen settled questions — can't find original logic

Sounds like

"The notes just say approved — no idea what the thinking was"

"She left last year — I think it was her call"

Tier 2

Institutional knowledge unreachable

Looks like

· Teams duplicate work completed months ago

· Search tools return wrong results

Sounds like

"I feel like we've done this before but can't find it"

"The knowledge base is basically useless"

Tier 2

Knowledge that never left someone's head

Looks like

· Post-step interviews reveal uncaptured knowledge

· Systems use dropdowns that exclude qualitative judgment

Sounds like

"I had a feeling but there was nowhere to put it"

"Nobody asked me"

Tier 2

Nuance compressed across audiences

Looks like

· Summaries turn conditional findings into absolutes

· Slide decks reduce findings to single scores

Sounds like

"The research said something more nuanced than that"

"By the time it got to the exec team it was a different finding"

Tier 1

Expert attention spread too thin

Looks like

· Review durations far shorter than genuine assessment requires

· Hardest cases get same rushed treatment as easy ones

Sounds like

"She looked at it for maybe two minutes"

"I just don't have time to go deep anymore"

Tier 1

Knowledge trapped across boundaries

Looks like

· Teams duplicate work that exists elsewhere

· Solutions re-discovered instead of retrieved

Sounds like

"The London office did something like this but I can't get their files"

"It's in Confluence somewhere — good luck"

AITier 3

AI capability the workflow doesn't use

Looks like

· AI tools used for a fraction of their features

· Workarounds for problems the tool could solve

Sounds like

"I just use it for the basic stuff"

"Can it do that? I had no idea"

Tier 2

Expertise consumed by low-value work

Looks like

· Senior people spending time on formatting and gathering

· Work needing their expertise gets the least time

Sounds like

"I spend Mondays building the deck instead of thinking about strategy"

"Most of what I do doesn't require my experience"

Tier 3

Work routed to the wrong person

Looks like

· Work assigned by capacity, not expertise

· Generalists handling specialist problems

Sounds like

"They gave it to us because we had capacity"

"I did my best but I'm not the expert here"

Tier 2

Frontline observations never reach decision-makers

Looks like

· Problems visible at the front line for weeks surface as crises

· Leadership surprised by things the floor saw coming

Sounds like

"We've been saying this for months but nobody listens"

"I told my manager but it didn't go anywhere"

Tier 3

Review steps that go through the motions

Looks like

· Review durations far too short for real assessment

· Approval rates near 100%

Sounds like

"I just glance at it — if nothing jumps out, I approve"

"It's a checkbox step, not a real review"

AITier 1

Alternatives suppressed by anchoring

Looks like

· Very few alternatives considered at decision points

· AI recommendations accepted near 100%

Sounds like

"The model says X, so let's go with that"

"Does anyone disagree? No? Moving on"

AITier 3

Decision authority migrated to AI

Looks like

· AI acceptance rates above 90%

· Nobody can name who owns the AI's configuration

Sounds like

"The system recommends it, so that's what we go with"

"I could override it but I'd have to justify why"

AITier 2

AI echoing what the team already believes

Looks like

· AI outputs consistently match existing beliefs

· Contradictory output treated as error, not signal

Sounds like

"The AI confirmed our analysis"

"It came back with what we expected — good sign"

AITier 3

Stale models shaping live decisions

Looks like

· Model training dates months or years old

· Business conditions shifted in ways the model doesn't reflect

Sounds like

"When was this model last updated? Not sure"

"The numbers feel off but the model says it's right"

AITier 1

Plausible-but-wrong AI output

Looks like

· Nobody can describe what a wrong output looks like

· No verification step exists in the workflow

Sounds like

"It looked right — the writing was really good"

"How would I know if it's wrong?"

AITier 2

Everyone's output looks the same

Looks like

· Outputs from different teams share unusual similarities

· Range of ideas narrowed since AI adoption

Sounds like

"Everyone's decks look the same now"

"We all arrived at the same conclusion — or did we?"

AITier 1

Experienced people losing their edge

Looks like

· Difficulty performing tasks without AI that were routine

· Declining catch rates for AI errors

Sounds like

"I used to do this in my sleep — now I have to think about it"

"When the system went down it took us twice as long"

AITier 2

Juniors producing work they don't understand

Looks like

· Junior workers can't perform core tasks without AI

· Conceptual understanding lags behind output quality

Sounds like

"I've never actually done one from scratch"

"I'm not sure why it works, but the output looks right"

AITier 2

Teams stopped thinking together

Looks like

· Meetings are serial presentations, not working sessions

· Debate and pushback decreased

Sounds like

"Critiques used to be where real work happened — now it's show and tell"

"I already worked it out with AI, so I'm just presenting"

AITier 3

The interesting parts of the job are gone

Looks like

· Declining effort in AI-affected roles

· Increased attrition among strong performers

Sounds like

"I basically just edit AI drafts now"

"The part I loved about this job doesn't exist anymore"

Tier 3

Nobody checks the premise before executing

Looks like

· Work begins immediately without questioning the brief

· Rework traces to flawed starting assumptions

Sounds like

"We just jumped in — nobody asked whether we should"

"The whole approach was wrong from the start"

Tier 3

Functions contribute in sequence, never combined

Looks like

· Cross-functional involvement serial, not simultaneous

· Output reflects the last function's perspective most

Sounds like

"We each review it in our silo and pass it along"

"By the time it gets to us, the real decisions are made"

Tier 3

Outside information doesn't enter the process

Looks like

· Decisions without reference to external conditions

· Competitive intelligence ad hoc

Sounds like

"We didn't know the competitor launched that"

"Clients are asking — has product heard?"

Tier 1

Nobody checks if decisions were right

Looks like

· No post-decision review process exists

· Same mistakes recur across projects

Sounds like

"Did that feature work? I don't think anyone checked"

"We keep doing it this way — no idea if it's effective"

Tier 2

No safe way to disagree

Looks like

· Consensus reached quickly without visible disagreement

· Dissent expressed privately after meetings

Sounds like

"I had concerns but the room moved on"

"Nobody wants to slow things down"

AITier 2

Nobody mapped where AI works and where it doesn't

Looks like

· AI applied uniformly without differentiation

· Nobody can say where AI is reliable

Sounds like

"We use it for everything — it's easier"

"Sometimes great, sometimes off — no idea why"

Tested Experimental evidenceGrounded Strong theory + case evidenceEmerging From practice, ahead of published research

Set your minds free