Evidence | Case Files

What the work looks like

Five AI Diagnostics from actual engagements, shared with client permission. Industries represented: construction, real estate, financial services, fleet operations, and residential security. Each score is out of 100; a higher score indicates greater unmanaged exposure.

Client identity is never disclosed or confirmed, under any circumstance, including to prospective clients. That standard doesn't bend for a sales conversation.

78/100
Regional construction firm | February 2026

High AI embedded in bid estimating, subcontractor pricing, and safety documentation, with no data handling controls and no governance layer of any kind.

Read the redacted diagnostic →
76/100
Homeowners association | April 2026

High AI integrated into daily speed enforcement and gate access decisions with no defined ownership of the classifications it was generating.

Read the redacted diagnostic →
70/100
Fleet operations company | April 2026

High AI active across sales, operations, and service functions, expanding through individual adoption with no centralized coordination or accountability structure.

Read the redacted diagnostic →
68/100
Financial services firm | April 2026

High AI influencing internal reporting and client-facing outputs in a regulated environment, without review standards or defined ownership before outputs became action.

Read the redacted diagnostic →
42/100
Real estate firm | April 2026

Moderate AI in use across client communication and internal workflows, review practices inconsistent across the team, no formal policy governing data handling or output use.

Read the redacted diagnostic →
The common thread

One pattern, five industries

Across every case, the score tracks the same thing: AI embedded in consequential decisions with no defined owner, no validation, and no record. The number climbs wherever governance is absent, not wherever AI is used most.

How the gap gets closed →
Fellowship Research

What we study, and where it goes

The diagnostics on this page are one input. Across engagements we track where AI is showing up in consequential decisions, where oversight tends to fail first, and which patterns hold across industries rather than one-off cases. That research does three things: it sharpens the methodology behind our own advisory and governance work before it ever reaches a client, it feeds the arguments we publish in The Wrong Default, and it shapes the AI literacy curriculum we bring to students through Still In Charge. Same question in every direction: who owns the decision, and did anyone check before it mattered.

Start here

The Diagnostic

Fixed fee. Time-boxed. Engagements begin September 1, 2026.

A structured conversation and a written read of where AI is already operating in your business, where the exposure sits, and whether anything more is warranted. If it isn't, we tell you.