Start with operational fit
A flashy automation can hide heavy judgment, frequent exceptions, and little tolerance for a wrong answer. Those traits make a poor first build.
A strong first workflow passes five tests.
A strong first workflow is recurring, measurable, and owned. Use five tests before choosing the first controlled build.
A flashy automation can hide heavy judgment, frequent exceptions, and little tolerance for a wrong answer. Those traits make a poor first build.
A strong first workflow passes five tests.
It recurs on a schedule or a trigger. Daily, weekly, per order, or per inbound lead. Unpredictable tasks that happen for different reasons need a different design.
It follows a pattern a person could write down. Ask the person who does the work to explain it. If they can say “first I check this, then I copy that, then I send this unless X,” you have a candidate. Repeated “it depends” answers signal too much judgment for a first build.
You can measure it. Count how often it runs and how long it takes, including the time spent redoing it when it comes back wrong. If nobody can say what the workflow costs today, nobody will be able to say whether the automation worked.
It has an owner. One person who knows how the work really runs and will sit in the approval seat once it is built. Orphaned workflows produce orphaned automations.
The judgment separates from the repetition. The machine handles the repeatable part and a person keeps the decisions. If you cannot draw that line on paper, the build will blur it in production, which is worse.
My first build replaced the repeated work of copying numbers into a morning report, structuring the same sections, and formatting the same output. The repetition exposed a stable pattern.
The system now drafts the brief from defined sources. I review the exceptions and decide what happens next.
Rank your candidates by hours consumed per month, weighted up for every time a human has to rescue the output. Then pick the smallest one you can see whole, not the biggest one on the list.
The first build's job is to prove the loop works inside your company: the trigger fires, the draft appears, a person approves it, the result gets measured. Once that loop runs on something small, the bigger workflows have a path to follow.
When the first build is unclear, the AI Workflow Audit maps the current state, ranks the backlog, and supports the choice with evidence.
The free assessment is written. You describe the drag, I reply with where I would start and why.