What I check before I automate anything.

The use case library shows the workflows. These are the rules I run them through, in order. The first one is the one almost everybody skips.

Start with the data Browse the use cases

01 · The one most people skip

Connect the data first

Data foundation

Most automation fails for a boring reason: the data it needs was never connected. The numbers live in one app, the customers in another, the schedule in a third, and the truth in your head. Before anything gets automated, the sources have to agree on one picture.

When I assess

See the data

I look at where your information actually lives and how it moves between people and tools. That walk-through is how the bottlenecks reveal themselves: the double entry, the status-chasing, the "it is in the email somewhere." You cannot find the friction without following the data.

When I build

Wire it together

Then the sources get connected so they present one truth. Not a giant new platform: usually a few links between tools you already pay for. Automation built on connected data holds up. Automation built on scattered data quietly breaks.

See it in the library

OP5 Everything lives in my head A7 Typing the same thing twice S5 Feeding the CRM F1 The receipt pile on your desk RE4 Herding docs to close

02 · The question that comes first

Audit the workflow, not the tool

The question is never "what AI tool should I buy." It is "which recurring workflow is actually ready." Start from the work that repeats and frustrates, then decide what to do with each step. The tool is the last decision, not the first.

See it in the library

A1 Where are we on that? OP1 The schedule that falls apart by Tuesday PS1 The invisible clock

03 · The honest part

Most wins need no AI at all

The highest-value fixes, scheduling, onboarding, invoice follow-up, reminders, contain no AI. They are rules, templates, and one connected calendar. Reach for AI only where the job is genuinely reading, drafting, or summarizing. Spending on AI before fixing the plumbing is the most common and most expensive mistake.

See it in the library

A3 Calendar Tetris HR3 The calendar ping-pong PM4 The monthly rent chase

04 · The line that protects you

Keep a human at the high-stakes gate

AI can read, summarize, draft, and recommend. The safe ladder stops at "draft" or "recommend" wherever the stakes are real: a hiring decision, a legal citation, a clinical note. The machine does the typing; a person signs the consequence. That line is where the risk lives, so that is where the human stays.

See it in the library

HR2 The resume avalanche EEOC PS6 The research rabbit hole citations HW6 The phone that never stops HIPAA

05 · The unexpected deliverable

"Don’t automate yet" is an answer

Some workflows are too messy, too high-stakes, or too unclear to automate safely right now. Naming those is part of the value, not a failure to deliver. A good audit tells you what to fix first and what to leave alone, so you do not pour effort into automating a process that should be redesigned instead.

See it in the library

CS5 Complaint without a process F6 The shoebox handoff OP6 The quality that depends on who is working

The bridge

The scorecard is the first pass at these principles, run on your business.

It names the biggest leak across deals, time, and cash. If the leak is real, the Hidden Profit Review measures it against your data and turns it into a first pilot plan.

Take the scorecard

Already scored? See the measured review.

Connect the data · Audit the workflow · Most wins need no AI · Human at the gate · Don’t automate yet