The Art of the User Story (and Where AI Fits In)
This is the first part in a three-part series to introduce using AI as a tool to help you create what the team needs to deliver the product you envision.
The three parts walk you through three levels of engagement with AI and where it can lead.
Part 1 - AI as a story Reviewer: how to use AI to refine your user stories.
Part 2 - AI as a story-writing assistant: how to use AI to create stories.
Part 3 - AI as a planning Partner: How to use AI to define features, epics, and stories
So, let’s get started with part 1:
There’s an art to writing a user story. The story itself is the easy part. What makes it great is everything around it, and that’s where the art comes in. It’s also where AI can help.
We all know what a user story is: it describes a piece of functionality, and it’s what the team builds from. It follows a standard format.
As a (who), I want/need (what) so that I can (why).
Let’s use this one for the rest of the article:
“As a commercial trucker, I need to see what a shipper is willing to pay me to deliver their goods so I can decide if I want to take the load.”
It has all the pieces of a good story. The who: a commercial trucker. The what: see what a shipper is willing to pay. The why: so I can decide whether to take the load.
It’s a good story. But one sentence can’t carry everything, and the team will come back with questions. What currency do I show the rate in? Where does it appear? What happens if there’s no rate at all?
That’s where AI earns its keep. I ran the story through an AI tool, and here’s what came back:
“As a commercial trucker browsing available loads, I want to see the offered rate and key load details before I commit, so I can quickly decide whether a load is worth taking.”
Why is it better? It swapped “what a shipper is willing to pay” for “offered rate.” That small change shifts how I think about the whole rate feature. “Willing to pay” suggests the driver has no say. The “Offered rate” suggests I can counter. We may not build counter-offer functionality now, but I want to leave the door open for later, and the wording quietly protects that option. That’s the kind of detail that shapes how the team architects the system.
Where it went too far. It also added “key load details” to what the trucker sees. That might matter, but it’s more than I want the team taking on in this story. AI tends to do this: it gives you more than you asked for.
The bonus: a first draft of acceptance criteria. The other thing it handed me was a starting set:
-
The offered rate is visible on the load listing without the trucker having to open or claim the load.
-
Rate displays both total payout and rate per mile.
-
Each listing shows the pickup location, drop-off location, distance, weight/equipment type, and pickup window, along with the rate.
-
If a shipper hasn’t set a rate, the listing says so clearly (e.g. “rate negotiable” or “make an offer”) instead of showing a blank or $0.
-
The trucker can sort or filter loads by rate and rate per mile.
These are strong. But, same as before, not all of them belong to this story. And that’s the thing to remember about working with AI: you always check the work. It adds things you didn’t ask for, and if you leave them in, they create more confusion than clarity.
So what do I actually do with all this? I break it apart into clean, focused stories.
Story 1: “As a commercial trucker browsing available loads, I want to see the offered rate before I commit, so I can quickly decide whether a load is worth taking.”
-
The offered rate is visible on the load listing without the trucker having to open or claim the load.
-
Rate displays both total payout and rate per mile.
-
Rates are displayed in US dollars.
Story 2: “As a commercial trucker browsing available loads, I want to see the key load details before I commit, so I can quickly decide whether a load is worth taking.”
-
The key load details are visible on the load listing without the trucker having to open or claim the load.
-
Each listing shows the pickup location, drop-off location, distance, weight/equipment type, and pickup window, along with the rate.
Story 3: “As a commercial trucker browsing available loads, I want to sort by total rate and by rate-per-mile before I commit, so I can quickly decide whether a load is worth taking.”
- The trucker can sort or filter loads by rate and rate per mile.
Story 4: “As a commercial trucker browsing available loads, if there’s no offered rate, I want to see the words ‘rate negotiable’ in the rate field, so I know I can work with the shipper on a price.”
- If a shipper hasn’t set a rate, the listing clearly indicates that (e.g., “rate negotiable” or “make an offer”) instead of showing a blank or $0.
The moral of the story. Using AI didn’t just sharpen my original story. It surfaced new ones I might not have thought of on my own. Yes, I had to do some work to clean it up and make the calls about what stayed and what went. But it got me further, faster. AI can’t replace you as a product owner. It can make you a lot more productive.
Watch for Part Two of this Three-Part Series.
Creating Stories through AI Dialog