AI & Operations
AI Workflows: Where Founders Should Start
Find a practical starting point for AI in operations by mapping the work, choosing a bounded workflow and keeping ownership clear.
By Luca Cerullo · Published
Map the work before introducing AI
When someone asks where AI belongs in a business, I want to understand the existing work first. Follow one request from its arrival to its resolution. Note who handles it, which information they need and where another person has to step in to clarify the next action.
Use an actual recent example rather than an ideal process diagram. A workflow that looks simple in a meeting can depend on context that lives in a colleague’s head. Bring that context into the discussion before deciding which part could be supported by software.
Choose a bounded workflow with a clear owner
A useful first experiment has a recognisable input, a limited task and an output someone can check. Drafting a summary for review is a different proposition from letting a system make an irreversible business decision. Define the boundary explicitly and agree who can approve the result.
I would rather examine one workflow properly than announce AI across every department. Choose a person who understands the work and can explain what a good result looks like. Give them a way to flag missing information and keep the existing process available while the experiment is reviewed.
Review the workflow as an operating responsibility
Look at the full process around the output. Who notices an exception? Where does a correction go? What happens when the input is incomplete? Those questions belong to the operating plan, even when the underlying model can produce an impressive demonstration.
Set a regular review of examples and unresolved decisions. The purpose is to learn whether the workflow helps the people doing the work, not to defend the first implementation. If the process itself is unclear, improving that process may be the most useful next step.
Turn the next useful idea into working progress.
Bring a real problem, a clear audience and the next decision you want to make. Explore a hands-on product, operations or AI collaboration.
Let’s Build →