Automate a Broken Process and You Buy a Faster Broken Process
Automation makes a process permanent. Any step you encode without questioning becomes structurally harder to remove than it was the day before.
Michel ZozayaTechnical AuthoritySpeed is not the same as improvement
A property management firm that automates work-order dispatch without first defining an SLA does not get faster service. It gets ambiguous dispatch at four times the volume, and complaint handling scales with it.
The typical finding is that 30–40% of steps in a long-standing process are compensating controls for problems that no longer exist — a double-entry check for a system replaced in 2019, three signatures on a €200 invoice, a status spreadsheet duplicating a CRM field.
The correct sequence
Document the process as it actually is, with an owner against each step. Mark every step as creating value, preventing a real risk, or neither. Delete the neithers. Then automate what survives.
Deleting is free and instant. Automating is neither. Once a step exists in code, removing it requires a change request, a developer or a vendor, and a regression test. A step performed by a person can be dropped in a meeting.
The failure mode specific to AI
A language model will cheerfully replicate a nonsensical rule and even rationalise it in its output. The moment of human friction that would have prompted somebody to ask "why do we do this?" never arrives.
A useful cost signal: if your automation specification carries more than about a dozen conditional branches, you are encoding accumulated exceptions rather than a process.
Take the process you most want to automate and remove three steps from it using nothing but a decision. Run it that way for a week before anybody writes a line of code.
Forty years as a Technical Director across shopping malls, offices, hotels and industrial assets.
Meet the team →Questions we get asked
Should I re-engineer my processes before automating anything?
No — a six-month process re-engineering programme is as wrong as automating blindly. The rule is narrower and cheaper: delete steps rather than redesign them. Two weeks of deletion, then build.
How much of a typical business process is genuinely necessary?
In long-standing processes, 30–40% of steps are usually compensating controls for problems that no longer exist — checks for retired systems, approvals set at thresholds fixed years ago, spreadsheets duplicating fields.
What warns me that a process is not ready to automate?
More than about a dozen conditional branches in the specification. That means you are encoding years of accumulated exceptions rather than a process, and the automation will be expensive to maintain.
Related insight
The AI Assistant That Does Everything Does Nothing
An agent's reliability collapses as you add systems it must be simultaneously correct about. The only agents that reach production…
AutomationYour Software Is Fine. The Gaps Between It Cost You Forty Hours a Week.
The recoverable waste in most operating companies sits in the transfers between systems, not inside any one of them. No vendor is …
AutomationTwenty Missed Calls a Day Is a Revenue Line, Not an IT Problem
Voice automation pays for itself on the calls that currently ring out. It only pays at all if the call ends up as a record in your…
Where is your business still running manually?
We review your processes, identify the highest-impact automation opportunities and estimate their financial impact — before anything is built.