HomeSoftware EngineeringMethods to Use AI for Product Discovery and Writing Higher Consumer Tales

Methods to Use AI for Product Discovery and Writing Higher Consumer Tales


AI is revolutionizing how groups develop merchandise. On this weblog we’ll introduce a set of prompts you should use to know your customers, put together to interview actual customers, generate a product backlog, add acceptance standards, and consider your personal person tales.

All through this weblog, I’ll use a constant instance so you possibly can see how outcomes from one immediate feed into the subsequent. For our instance, let’s think about our group is creating a brand new product for valet-attended parking garages. We’ll validate our product by promoting initially to impartial parking garages comparable to at boutique inns.

For the examples right here, I will probably be utilizing ChatGPT’s GPT 5.2 Considering mannequin.

Constructing a Persona

It stays crucial for groups to know a product’s customers and clients by interviewing, surveying, and observing them. A group can complement these interactions by coaching an AI on what’s been realized after which interacting with a user-proxy AI. Crew members might, for instance, ask their user-proxy AI questions comparable to:

  • What do you consider this characteristic?
  • How typically would you utilize this characteristic?
  • What might make this characteristic extra helpful?
  • Which of those options would you discover extra helpful?
  • Would you be prepared to pay for this characteristic?

The final type of our immediate to create a persona is:

Our product is [product description]

Construct me a persona of [ persona description]. Record their hopes, issues, emotional triggers, and resolution standards for selecting our product.

Right here’s what I used for the parking valet app:

Our firm is creating software program for valet-attended parking garages. To validate our product we are going to promote initially to impartial operations comparable to boutique inns.   

Construct me a persona of the proprietor or supervisor of such a parking storage. Record their hopes, issues, emotional triggers, and resolution standards for selecting our product.

The Outcomes

I used to be given a persona, Carlos Martinez, who owns and operates a valet-attended storage with 80–200 areas at a boutique lodge. Carlos has 10–25 years in parking and hospitality. He has a group of 5–25 valet attendants, who’re a mixture of full-time and high-turnover part-time workers.

The purpose about high-turnover part-time workers is an perception I may need ignored. I take into consideration a lodge I go to typically: there are 4 attendants I see each journey, and I’d acknowledge them away from the lodge. The opposite attendants blur collectively. I wager they’re the high-turnover part-time ones.

You’ll be able to view the complete persona description within the popup beneath.

I significantly like seeing our persona’s hopes and aspirations. My character is such that I all the time wish to get proper all the way down to what a person wants, however understanding their hopes and aspirations is useful. There have been a couple of issues on this persona I won’t have recognized by myself:

  • Lowering the dependency on one “supervisor who is aware of every little thing”
  • Win or renew lodge contracts extra simply (I may need considered this later however I’d have been initially too centered on working the storage to have considered Carlos’s large concern: renewing the lodge’s contract with him)
  • Predictable labor prices. If our product consists of the precise analytics, we are able to undoubtedly assist with this characteristic I won’t have considered.

Interviewing the Persona

As soon as AI has created the persona, inform it:

Please act as [persona name] and let me interview you. 

You’ll be able to see an excerpt of my interview with Carlos.

The downloadable immediate pack consists of some pattern questions I’ve discovered helpful you should use to get began.

An Interview Information for Actual Customers

You may also use AI to organize interview questions of actual customers or clients who match a persona. To try this, I take advantage of this immediate:

I plan to interview some precise customers who match this persona. Please put together an interview information I can use.

Embrace key factors I ought to make initially about confidentially, time dedication, function, and that there aren't any proper or unsuitable solutions. 

Establish questions I can ask in regards to the present course of and ache factors with it. As applicable, embody appropriate follow-up questions that probe for extra depth. 

Embrace questions that can assist me perceive their desired future state together with objectives and success standards. 

Embrace questions on constraints (regulatory or different) that I ought to ask. 

Lastly, embody questions or statements I can use to wrap up the session.

You’ll be able to see the interview information for the owner-operator of the valet-attended parking storage. I’ve ready nicely over 100 of such interview guides through the years. this makes me understand how a lot better ready I might have been had AI been obtainable.

Writing Consumer Tales

On an agile mission, we finally wish to get to the extent of writing particular product backlog objects, sometimes person tales and job tales. To try this, we’ll use this immediate:

You're a enterprise analyst chargeable for writing clear, high-quality person and job tales. 

Every person story should strictly observe this precise format: 

As a [USER ROLE] I need [WHAT] in order that [WHY]. 

Every job story should strictly observe this format:

When [SITUATION], I wish to [MOTIVATION] so I can [EXPECTED OUTCOME].

Every person and job story should fulfill the INVEST standards as follows: 

- Impartial: one characteristic or want per story. 
- Negotiable: open to dialogue, keep away from overly detailed options. 
- Helpful: clearly advantages the person or enterprise. 
- Estimable: scoped nicely for effort estimation. 
- Small: completable inside one dash. 
- Testable: indicate verifiable outcomes (acceptance standards not required). 

Use clear, concise language appropriate for agile growth groups.  

Produce tales within the precise format beneath with no numbering, bullet factors, commentary, or clean strains. 

Select between person and job tales for every want as applicable. 

Don't deviate from these codecs or add anything. 

The product is [describe the product].

Deal with this useful space of the product: [describe a subset of the product].

Your job is to generate as many tales as wanted to for [attach a persona PDF or name the persona].

The vast majority of this immediate offers steering on what person and job tales are, describes the INVEST standards of excellent tales after which instructs the AI to provide tales with out additional info or formatting. This permits them to be simply copied and imported right into a backlog administration system.

The previous couple of strains of the immediate are the place issues get attention-grabbing.

Product Description

Describe the services or products being developed. You’ll be able to alter the immediate to incorporate this as a file you probably have an in depth description.

Useful Space

Instruct the AI to deal with a subset of the product. In the event you’re creating a phrase processor, for instance, you may deal with printing. The method might be repeated for every space of the product.

Persona

Lastly, point out the persona (or person position) for whom you wish to write tales. In the event you created the persona within the AI chat, you possibly can simply title the persona (e.g., “for Carlos”).

Nonetheless, because you’ll seemingly repeat this course of, it’s helpful to save lots of the persona description in a file and add the file.

Tales for the Valet Parking Instance

For our ongoing parking valet instance product, right here’s how I stuffed within the final a part of the immediate:

You're a enterprise analyst … [same as above]

The product is a brand new software program product to be offered to valet-attended parking storage proprietor/operators. To validate our product, we are going to initially promote it to impartial operations comparable to boutique inns. 

Your job is to generate as many tales as wanted for the persona hooked up. 

I didn’t instruct the AI to deal with a subset of the storage proprietor/operator’s wants on this case. I felt just like the proprietor’s wants had been sufficiently small I might let the AI deal with all of them. Right here’s what I bought:

  • When automobiles are dropped off by visitors, I need every handoff recorded clearly so I can cut back disputes about accountability.
  • As an owner-operator I need proof of auto situation at drop-off in order that I can shield the enterprise from false injury claims.
  • When a valet retrieves a car for a visitor, I need the retrieval time tracked so I can monitor visitor wait instances.
  • As an owner-operator I wish to know which valet dealt with every car in order that I can maintain workers accountable if points come up.
  • When peak check-in or check-out durations happen, I need the drop-off and retrieval course of to remain quick so I can meet lodge service expectations.
  • As an owner-operator I need visibility into automobiles at present in valet custody in order that I can reply lodge or visitor questions confidently.
  • When a visitor studies an issue throughout pickup, I need entry to a transparent file of the car’s historical past so I can resolve the difficulty shortly.
  • As an owner-operator I need the drop-off and retrieval workflow to be easy for valets in order that new or momentary workers can use it appropriately.
  • When I’m not on-site, I wish to see what automobiles are being dropped off and retrieved so I can really feel in command of operations remotely.

You’ll be able to see the immediate gave AI sufficient element that it might appropriately fluctuate between person and job tales. It selected job tales when a set off or scenario was extra essential than the particular person performing a narrative. For instance when automobiles are dropped off and when I’m no on-site.

Including Acceptance Standards

I choose including acceptance standards as a separate step after I’ve tales for one a part of the product. AI can simply add these as a easy bulleted checklist or in Gherkin format (given-when-then).

I discover the simplicity and terseness of a easy checklist preferable after I’ll evaluate the tales with customers. You may also ask AI to transform them into Gherkin format later.

To generate acceptance standards inside the identical AI session, you should use this immediate:

Please add acceptance standards to the tales you wrote expressing them as [just a list][in Gherkin format].

Instance Acceptance Standards

Right here’s the instance acceptance standards as a easy checklist and in Gherkin format for this story:

As an owner-operator I need proof of auto situation at drop-off in order that I can shield the enterprise from false injury claims.

Record

  • Car situation proof is captured at drop-off
  • The proof is related to the right car file
  • The proof might be seen after the car is returned

Gherkin

Function: Seize car situation at drop-off

Situation: File car situation proof throughout drop-off

Given a car is being dropped off by a visitor
When the valet completes the drop-off course of
Then the system captures proof of the car’s situation
And the proof is related to the right car file

Situation: View car situation proof after drop-off

Given a car has been dropped off with situation proof recorded
When the owner-operator views the car file
Then the car situation proof is displayed
And the proof is clearly identifiable as taken at drop-off

Given a injury declare is reported by a visitor
And car situation proof exists from drop-off
When the owner-operator critiques the declare
Then the owner-operator can entry the drop-off situation proof
And the proof features a timestamp indicating when it was captured

Having AI Consider Your Personal Tales

Synthetic Intelligence doesn’t have a monopoly on creativity. So naturally you have to to write down some person or job tales your self. Once you do, it may be useful to have AI consider your tales. You’ll be able to even have AI consider its personal tales–in spite of everything, it’s not excellent.

We’ll have AI consider our tales towards the well-known INVEST standards. These point out that tales must be:

  • Impartial
  • Negotiable
  • Valauble
  • Estimable
  • Small
  • Testable

Right here’s the overall type of the immediate for evaluating tales:

You're an professional agile coach specializing in person story high quality. Your job is to judge person and job tales towards the INVEST standards beneath:

- Impartial: Story stands alone with out dependencies.
- Negotiable: Story is open to dialogue and refinement.
- Helpful: Story delivers clear person or enterprise worth.
- Estimable: Effort might be moderately estimated.
- Small: Story might be accomplished inside one dash.
- Testable: Story has clear, verifiable completion standards.

[Optional: Examples of good and bad output are in the attached file.]

For every person story and its acceptance standards

1. Establish solely the INVEST standards that the story does NOT meet.
2. For every unmet criterion, present a concise, particular clarification (1-2 sentences) centered on why the story fails it.
3. Recommend one concrete, actionable change that may make the story absolutely meet all INVEST standards, instantly addressing the unmet factors.

Format your response precisely as follows:

Story: [repeat the story]

INVEST Standards Not Met: [criteria separated by commas]

Clarification:

Recommended Enchancment: [Concrete, actionable change]

Don't present suggestions on standards the story already meets.

Keep away from verbose or generic explanations. Recommend just one centered change.

Listed here are the tales and their acceptance standards:

[Stories followed by their acceptance criteria]

This immediate helps optionally together with a file containing examples of excellent and dangerous person tales and acceptance standards. I recommend initially evaluating your tales with out bothering to do that. The outcomes will seemingly be ok.

Nonetheless, it’s simple to slowly assemble a doc of excellent and dangerous examples by including one or two every time you do that. The AI Immediate Pack: Consumer Tales reveals an instance format for the examples.

You’ll be able to see how AI evaluated three of its personal tales.

A Sensible Associate, Not a Substitute

AI received’t change the necessity to discuss with actual customers, make judgment calls, or train product sense. What it can do is dramatically cut back the friction between these actions. It helps groups transfer quicker from obscure concepts to concrete artifacts, exposes blind spots we would in any other case miss, and raises the baseline high quality of our work—particularly when time or expertise is proscribed.

Used nicely, AI turns into a tireless collaborator: one which remembers each persona element, by no means runs out of endurance for rewrites, and may immediately shift from big-picture considering to express backlog objects. The prompts on this weblog will not be meant to be adopted blindly, however tailored, challenged, and refined—similar to any good agile apply.

If there’s one mindset shift to remove, it’s this: don’t ask whether or not AI can change a part of your product discovery or backlog refinement course of. Ask the way it might help you present up higher ready for the conversations that also matter most—with customers, stakeholders, and your group. That’s the place its actual worth lies.

Want Assist Utilizing AI to Speed up Your Crew?

Teaching groups on instruments and approaches like that is what we do. If you wish to focus on how we might help your group incorporate AI into its agile work, let’s discuss.


Final replace:

January thirteenth, 2026

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

- Advertisment -
Google search engine

Most Popular

Recent Comments