Duplicate work
The same information typed into more than one system.
Insights
How to identify business processes that should be automated with AI.
Don't look for places to use AI — look for expensive, repetitive, measurable friction. Four patterns, ten diagnostic questions, and a test for whether AI is even the right tool.
Short answer
Don’t start by looking for places to use AI. Look for work that is expensive, repetitive and measurable: information re-entered between systems, reports people wait on, data that disagrees across systems, and decisions made with the same rules every time. Then check whether AI, simple automation or a process change is the right fix.
Four patterns
Where the best opportunities usually hide.
Reporting delays
Managers waiting on IT or an analyst for answers that already exist in the data.
Disconnected systems
ERP, CRM, accounting tools and spreadsheets that disagree about the same fact.
Repeated decisions
People applying essentially the same rules over and over.
Ten diagnostic questions
Ask these about your own operation.
- How often does your team re-enter information from one system into another?
- How much time is spent researching information needed for routine work?
- Do employees manually read, sort or route emails, documents, orders or requests?
- When leaders need a new answer, do they wait for IT or an analyst to build a report?
- Do your ERP, CRM, accounting tools or spreadsheets disagree about the same business fact?
- Do people make repetitive decisions using essentially the same rules each time?
- Are there processes that depend on one employee knowing “how we do it”?
- Do you struggle to see margin leakage, slow turns or avoidable exceptions quickly?
- Could the same team handle more volume if routine work were automated?
- Have useful AI experiments failed to become reusable workflows or agents?
Is AI the right tool?
Sometimes the answer is no.
Find the one repeated, per-unit action driving the cost. In the bulk receiving case study, the fix wasn’t AI at all — auto-sequencing barcodes removed a repeated motion and cut labor 75%.
AI earns its place when the work involves reading, identifying or researching unstructured information: photos, documents, spec sheets. In the AI Intake case study, a photo returned a full product record in under 30 seconds.
Before building, confirm the project moves a measurable outcome — cost, speed, accuracy, human touches, revenue or profitability — and test it cheaply first. That test took one late night with five items.
Start with a conversation
What problem is costing you more than it should?
Let’s spend 30 minutes looking for the first problem worth solving.