What are we looking at?
When a system works badly, can “making it smart” save it?
Whenever a problem shows up in a system or a service, one suggestion lands on the table straight away: let’s make it smart. But can smartening, automation or adding AI solve the problem by itself? The short answer: no.
Technology cannot rescue a bad process
Technology is an enabler, not a miracle worker. If our processes, the order of our actions and the logic of our decisions are flawed, even the most advanced intelligent system will only speed up the existing faults, not fix them.
To explain it, I’ll borrow an example from Jānestān–Kābolestān, a travelogue by Reza Amirkhani.
The story of two toilets
In this travel book the author describes how people use the toilet in Afghanistan. It looks trivial, but I think it holds one of the deepest lessons in systems understanding.
The Afghan routine goes like this: first you flush, then you do your business, then you wash and leave. In Iran the order is usually reversed: first the business, then washing, and flushing last.
In terms of components and actions, the two are identical: flush, use, wash. In terms of order and process they are completely different, and that subtle difference changes the final experience: in Iran the room usually stays clean, and in Afghanistan, even though every component is present, the place often smells.
Now imagine automating the flush
Suppose we want to make the Afghan setup “smart” and say: “Let’s install an automatic system that flushes after every entry and exit!” The idea looks appealing, but it does not solve the problem, because the order of actions is still wrong. We are putting technology on top of a faulty process, and the result is that the system’s fault runs faster and more often.
The lesson: process matters more than components
In many organisations and projects exactly this happens. An ERP is installed without redesigning the organisation’s processes. Software robots (RPA) run on systems whose processes are inefficient or redundant. Recommender systems go live without anyone understanding the user’s need or the context of the data. In all of these, advanced technology doesn’t merely fail to help; it produces “wrong output, automatically, continuously and faster.”
So what should we do?
Before any digitisation, automation or AI deployment, we should answer a few questions:
- What exactly is our process?
- Where are its bottlenecks?
- Seen from the user’s experience or the real goal, how sensible is the order of actions?
- If we ran this same process with paper and pen, would we still get the result we want?
If the answers aren’t clear, any amount of smartening will just spread the same bad smell faster.
Technology is not a band-aid
Technology shouldn’t be treated as a temporary patch for an unhealthy process. First take system design, careful process analysis, re-engineering the order of actions and optimising decisions seriously; only then will any technology you add actually help. Technology is only an amplifier: if your process is flawed, automation means amplifying the flaw.
Is there a place in your system where only an “automatic flush” has been installed, yet the smell of the problem remains? If so, perhaps it is time, before innovating, to revisit the process.
process-designautomationERPRPAartificial-intelligenceprocess-analysis