The pitch is everywhere: layer generative AI onto your existing workflows and watch productivity climb. The logic breaks on contact with how most workflows actually run.
If a process has twelve handoffs and four of them are redundant, automating it does not remove the four. It runs all twelve at machine speed. You have industrialised the worst features of the process and stripped out the human friction that used to slow the damage down.
Automation is a multiplier. Point a multiplier at a broken system and you get more broken system per hour.
Complexity is not the problem. Complicatedness is.
The distinction most transformation programmes miss:
- Complexity is external and non-negotiable — markets, regulation, geopolitics. You don't get to opt out of it.
- Complicatedness is internal and self-inflicted — the committees, approval layers, and reporting lines every leader added for a sensible local reason and no one ever retired.
AI does nothing about the first and accelerates the second. So the answer is not "don't automate." It is: fix the sequence first, and give the sequence a structure disciplined enough to survive politics.
The structure: the Simplexity Backbone
That structure is what we call the Simplexity Backbone — six operating principles, run across four dimensions (People, Process, Data, Technology), for 24 coordinates. Every business problem maps to a specific one. It turns "simplify things" from a slogan into human judgment made structured and auditable.
The six principles:
- Purpose-Driven Action — every activity connects to strategic meaning, or it goes.
- Strategic Focus & Refusal — growth requires disciplined deletion.
- Anticipation & Scenario Readiness — lead on indicators, not historical reporting.
- Strategic Detours — architecture built to pivot, not to shatter under pressure.
- Market Sensing & Selective Intelligence — filter noise, kill fragmented data silos.
- Collaboration & Redundancy — redundancy in capability, not in bureaucracy.
Run the principles in the right order
Applied to the AI question, Principle 2 runs first. Activity-based costing exposes the hidden overhead; any workflow or handoff that does not move the business forward is explicitly removed — before a single tool touches it. Principle 1 anchors what survives to strategic purpose. Only then does automation enter, pointed exclusively at what earned its place.
Run it backwards — tool first, deletion never — and you have bought a faster version of the organisation you were trying to fix.
The test to run before you automate
Before you automate a function, run the harder test: if this workflow disappeared tomorrow, what would break? If the answer is "nothing," you have found your project — and its coordinate on the Backbone.
Complexity is inevitable. Chaos is optional. AI doesn't change that equation — it just raises the stakes of getting the sequence wrong.
Written by Geoffroy Vilbert, founder of Opteamum Associates.