Find the hidden system
Read spreadsheets, side files, and workarounds as evidence of missing requirements.
A field guide for real-world automation
Before YouAutomate It
Find the hidden rules, exceptions, and spreadsheets that actually run the work.
Mac Ibale
← All books
First draft complete · Reader review next
Before You Automate It
A Field Guide to Finding the Hidden Rules, Exceptions, and Spreadsheets That Actually Run the Work
For developers, analysts, operations leads, and process owners asked to automate work before anyone has fully explained how that work survives exceptions.
Not for sale yet. The manuscript is being refined before release.
The central idea
“The hardest part of automation is rarely writing the code. It is discovering the real process hiding behind the documented one.”
Assign the right responsibility
Use AI for interpretation, code for guarantees, and people for accountable judgment.
Design for trust
Make exceptions, failure, ownership, validation, and recovery part of the first useful version.
Inside the book
Nine chapters from discovery to dependable delivery
- 01
The Process That Exists Only in PowerPoint
Follow the work instead of trusting the cleanest diagram.
- 02
The Excel File Nobody Mentioned
Treat unofficial tools as evidence before replacing them.
- 03
Ask What Happens on a Bad Day
Use difficult cases to reveal assumptions and exceptions.
- 04
The Person Who Is Secretly the Database
Find important decisions stored only in someone’s memory.
- 05
Turn Clues Into Rules and Responsibilities
Name sources, states, rules, owners, and evidence.
- 06
AI, Code, or a Human With Context?
Give interpretation, guarantees, and accountability the right owners.
- 07
The Design Must Fit Through the Door
Treat data, access, security, and support as architecture.
- 08
When Automation Needs a Full-Time Babysitter
Count the human work surrounding an automated run.
- 09
Build the Smallest System People Can Trust
Start narrow, close the reliability loop, and expand carefully.
Free excerpt
The question before “Can we automate this?”
“Can we automate this?” sounds like a technical question. It invites technical answers.
Should you write a script? Build an application? Connect an API? Use an AI agent? How long will development take?
Those questions matter, but they arrive too early.
Before deciding how to automate a process, someone has to determine what the process actually is. Not what the diagram says. Not what happens during a prepared demonstration. Not what the procedure looked like when it was approved two years ago.
What happens on an ordinary, slightly inconvenient Tuesday?
Consider an automation that appears to be correct. It follows every documented rule, moves information through the expected steps, and produces the required result.
Then someone asks whether it also updates an Excel tracker.
What Excel tracker?
The workbook did not appear in the requirements or workflow diagrams. Yet it contains information the official process does not: real statuses, reference values, manual corrections, special cases, and colors that apparently carry legally binding emotional authority.
The automation understands how the process is supposed to work.
The spreadsheet understands how people are actually completing it.
This does not make the spreadsheet bad or the automation a failure. It means another part of the system has been discovered—one operating quietly outside the official description.
That discovery is the beginning of the real engineering work.
Automation magnifies your understanding
Automation makes a process faster and more consistent. That is useful when the process is understood.
When it is not, automation makes incomplete assumptions faster and more consistent.
A person can notice that something feels wrong, open another file, ask a colleague, or delay a decision. Ordinary code follows the path it has been given. An AI system may interpret an unclear case, but it cannot recover rules nobody included in its context.
The quality of an automation therefore depends on more than code. It depends on the quality of the investigation that happened before the code.
Before you automate the work, earn the right to describe it.
Read the ideas behind the book
Before You Automate It
Get the sample and launch update.
Join the newsletter for one useful excerpt during editing and a single message when the finished book is available. No daily launch countdown.
Unsubscribe anytime. Powered by Buttondown.