Fix the process before you automate it
· 5 min read
There is a strong temptation to take an existing process and simply add AI on top of it. It feels fast and low-risk. But most processes were never designed; they grew over years of workarounds. Automating them as they are locks in every unnecessary step.
Map how the work really flows
Start by drawing the process as it actually happens today, not as the procedure document describes it. Follow a single request from start to finish and write down every step, every handoff between people, and every place it waits.
Most teams are surprised by what they find: approvals nobody remembers the reason for, data entered twice into two systems, and requests that sit in an inbox for days between steps that take minutes.
Remove before you automate
For every step, ask a simple question: what would happen if we stopped doing this? Often the honest answer is "nothing". Removing a step costs nothing to run and never breaks. Frequently the biggest gains in a project come from steps that were eliminated, not from steps that were automated.
- Eliminate approvals that add delay without adding control
- Enter each piece of data once, in one place
- Reduce handoffs between people and teams
- Standardize the inputs so fewer cases need special handling
Then add AI where it earns its place
With a cleaner process, it becomes clear where AI actually helps: reading and classifying incoming requests, extracting data from documents, drafting replies, or moving information between systems. These are the steps that are repetitive, high-volume and well-defined.
Keep people responsible for decisions that need judgment, and design clear points where a person reviews or approves.
Pilot, measure and then scale
Run the redesigned process with one team and compare it to the baseline you captured while mapping. If cycle time, error rates or workload improve, expand it with training and a simple dashboard that shows the impact. If they don't, you have learned that cheaply.
