There are two ways to bring AI into a piece of work. The first is to take the process you already have and add AI to one of its steps. The second is to ask what the process would look like if you were designing it today, knowing what AI can do, and then build that. The first is bolted on. The second is built in.
Both can be done with the same tools. They lead to very different places.
What bolting on looks like
Picture a request that arrives by email, gets copied into a tracker, waits for someone to review it, goes to a manager for approval, and finally gets acted on. A bolted-on approach adds AI to the review step. The reviewer now gets a summary instead of reading the full request.
That helps the reviewer. It does not help the request. It still waits in the same queue, still passes through the same hands, and still takes as long to finish. The bottleneck was never the reading. It was the waiting, the copying and the handoffs, and none of those changed.
This is why so many AI pilots look promising in a demo and disappointing a quarter later. The step got faster. The process did not.
What building in looks like
A built-in approach starts somewhere else. It asks what the request actually needs: understanding what is being asked, checking it against the rules, gathering the right information, and a decision from someone accountable.
Designed around AI, the work can be arranged differently. AI reads the request as it arrives, checks it against the rules, pulls what is needed and prepares a recommendation. Routine requests that clearly meet the rules move straight through. The ones that need judgment go to a person, with everything already assembled and the open question stated plainly.
The tracker, the copying and most of the queue disappear. The person's time goes to the decisions that need a person. The process is shorter, easier to see and easier to improve.
Why teams default to bolting on
Bolting on feels safer. It changes less, it fits inside existing roles, and it can be done by one team without asking anyone else. Building in means touching the whole process, which usually crosses teams and challenges how responsibilities are divided.
That is exactly why it pays off. The value of AI sits mostly in the handoffs and the waiting between steps, not inside any single step. You only reach that value by redesigning the flow.
How to start
Choose one process with a clear beginning and end. Map it as it really runs. Mark every place where work waits or gets copied. Then design the version where AI carries the routine path and people own the decisions. Run that version alongside the old one until the team trusts it, and then retire the old one.
It is more work than adding a feature to a step. It is also the difference between using AI and being changed by it.