Cut the Time, Skyrocket the Results: Find Your Value Proposition with AI in Minutes

Good morning. We concluded another quarter of the Stanford Class for Advanced Product Design this week. I am fortunate to be one of the coaches, and every year, it is so energizing to see how students bring ideas from concept to market and going through the thrilling adventures of an early product lifecycle.

Although it is not required to launch, many projects can show an intent to buy or raise money to pursue the project. Snapchat is a product you may have heard of. It came out of this class years ago.

Much of the work is user research and adapting the product direction. Although we didn’t see many AI-specific projects this year, AI helped the students with much of the research work behind the scenes.

After the wild success of building a Product Manager GPT together, today, we’ll do the same for the UX Researcher. We will discover many new tools augmented with powerful AI features.

This is a long post and likely more of an evergreen post you can return to when needed. I’ll keep it updated with the best AI-related tools for UX research.

So this post turns into an episode of…

7-minute-read

Story: The luxury of having a UX research lab

It was a sunny, warm Friday morning, and just like every week, our designers and I walked across the company campus in Sunnyvale from our desks toward our UX research lab. We were fully equipped with laptops and coffee and ready for our weekly, highly anticipated in-person user study.

As we arrived at the lab, we made our way to a room equipped with a couch, chairs, monitors, and a large one-way mirror. This would allow us to look into the room where our users would converse with our UX researcher and perform tasks in our apps.

This room was set up with various devices that our product would work on and several cameras and microphones.

Our researcher would receive our prototypes with a list of questions and things we wanted to learn. They took care of inviting and preparing the users and prepared the interview outline.

What a great feeling when everything runs smoothly like that.

When suddenly… I woke up. It was Monday morning, and I was back at a startup, and I had to do all this work by myself…

At least I had our recipe for a disciplined research loop that I could attempt to replicate without a researcher:

Our weekly rhythm was beautiful, and it is a cycle I highly recommend:

• Monday: Receive and review UX research notes, highlights, and key learnings. Design sync to discuss designs for the week and review user research feedback

• Tuesday to Wednesday: Get new designs ready

• Thursday: Setup clickable prototype and brief the UX researchers with questions

• Friday: UX research sessions or quantitative surveys go out

• And the cycle repeats.

The most important part was the discipline to keep that rhythm going. Having something ready on Thursday was our forcing function.

AI changes everything

AI gives me great hope that we can almost return to the full service we had in the past, even with a limited budget. Let’s see how we can use LLMs and AI tools to get into a good rhythm.

How can we replicate a successful UX research workflow with AI?

Many early-stage entrepreneurs and product builders struggle with the first step to get started.

• I have an idea for a product but don’t know where to start…

I am not making this up. See:

One smart Reddit user replies enthusiastically using not one but 3 exclamation marks: VALIDATE!!!

The goal is to find out if anyone cares about your idea because it would help them somehow. Confirm if you have a value proposition.

Validating an idea is doing the work of a researcher. You have a hypothesis and want to determine whether it is true or false.

It takes time and effort and can’t happen within a few hours… OR CAN IT NOW?

First, we will examine the steps needed to VALIDATE!!! and then see where AI can help put it in hyperspeed.

Let’s look at the product cycle.

The entrepreneurial and product journey starts with a very common set of steps that likely become second nature if you have done it a few times:

The outcome will be low—or mid-fidelity designs and a dataset that helps us answer whether our idea has any value and whether to proceed or take a new direction.

Within that part of the user research, we should also ensure that we keep these principles in mind:

• You want to meet the user where they are.

• What is their typical workflow? Getting them to pay for or even install extra apps or add extra clicks will be a hard sell. Is it worth it for them?

Ideation

Either you have an idea that you like or you don’t. If you don’t, and you are lucky enough to come from a clean sheet - just looking to explore what you could do, try this prompt to get your brain cells going:

You can always go more narrow if you are starting from a defined point. However, it is fun to play with this prompt because it shows that good ideas are not rare and easy to come by.

Another option is to scour X (former Twitter) or Reddit for threads around:

• “Help me…”

• “Help me avoid…”

• “I need to…”

Some of these should lead to an interesting opportunity that somebody could solve.

If your sandbox is smaller and you are, for example, ideating with an existing user base and set of tasks, the “Jobs to be done” framework by Clayton Christensen is a useful tool.

No need to dive into it too much, just apply this prompt:

The reply categorizes into the right jobs and gives tips on design direction. From here, you can always go deeper. Days of expensive workshops with mediocre “working-lunches” saved…

Please let me know in a reply if you are interested in diving into this framework more.

Need finding example

Before proceeding, we need to estimate the market and whether a viable need exists. The example below shows a quick and scrappy way to do this.

The first Answer my open-ended SaaS prompt above gave me:

• AI-powered creative tools for specific niches (e.g., crafting social media content for realtors)

Fiverr or Upwork are places where sellers and buyers meet to trade these services. Someone likely offers a niche service that a planned software product would solve. Almost any SaaS business exists as a manual service someone offers.

Fiverr even gives price transparency, which is useful data for evaluating value.

The example result below shows that he has had six reviews over the last six weeks and has saved his profile 53 times. It's not a huge crowd but a signal for a need. And there are a few more offers on the site.

A quick way to validate ideas and check if AI has the right sense of this need.

A lot deeper research has to follow, starting with customer discovery.

Customer discovery

To craft the value we want to provide, we need to understand our ideal customer, the target audience. After all, these are the people with whom we need to validate our ideas.

Target audience by AI

Use the LLM of your choice and type in the following prompt:

A really useful tool that will add an image and some style to the response (basically a pitch-deck-ready slide):

User Persona (Web app)

• A free service that creates personas including sub-categories

• See below an example of a user persona of this newsletter

For even more detail and to list out several personas, you can use

InstantPersonas (Web app)

• It lists out all possible different personas and describes them in detail.

• Free trial for 3 days, great to get the persona work done

Sidenote: Cool feature for content creators

InstantPersonas even gives you content examples that would be interesting for your target personas; it listed out a bunch of TikTok shorts for me:

Your AI UX researcher is becoming smarter.

With GPTs being able to remember context within the same chat and GPT-4o even across chats, doing all the research work in one chat puts you in a great position. Now that the AI knows your target users, you can extract more specific information from your AI UX Researcher within seconds.

Narrow down your audience and craft a killer offer for new customers.

Transferrable examples for a newsletter creator:

Understand the user lifecycle.

Meet the user where they are. What is their typical day, and where does your product fit in?

Wireframing and Prototyping

There is no BEST WIREFRAMING TOOL. It is the tool that lets you bring your thoughts into a visual representation that helps others understand with very little added context.

Tools I use: Pen & Paper, Whiteboards, Lucidchart, Figma

To create clickable prototypes, go with Figma.

Best Figma AI Wireframing Plugin

Figma plugins expand the features that Figma offers. There are a lot of AI-powered ones. I’ve tested a few of them and I stuck to:

WireGen - Generate Wireframes with AI

The results will look like a low-fidelity user interface that does exactly what you need: It lets users imagine what they would use the product for. (As long as you describe it well in the prompt.)

The tool is free, and you can generate 25 wireframes per month.

Once it puts them out, you can use Figma’s prototyping function to link buttons to actions and screens.

User interviews and surveys

Finding, recruiting, and scheduling users is the biggest time sink.

Even if you use existing tools, users may reschedule, not show up, or lack the background you expected. It’s a painful task.

Tools to recruit users are prolific.com or usertesting.com.

A very established tool for conducting interviews is lookback.com. I like the feature of a virtual observation room it creates. It reminds me of the time when we had our UX Research lab in a pre-Covid world.

AI in the mix: The most powerful recent addition is Eureka. Eureka is a sidekick that distills the highlights of hour-long user interviews into actionable takeaways within seconds—perfect for sharing with design and product managers.

The time savings here are so measurable. After the interviews, it would take one of our UX researchers about half a day to review all interview protocols. AI eliminated that.

Game changer: User research without recruitment

But fear not; the user study management pain described above could soon be solved.

The tool:

SyntheticUsers

Using human-like AI as user research participants.

This tool could be a potential game-changer for many time-consuming tasks.

You might be skeptical of how accurate these replies will be. I haven’t tested it, but the company is showing research that reaches an overall parity of 85%. This is a great starting point for getting quick answers, and it will only get better over time.

🔨 This concludes this episode of Tool Time. Happy researching.

If you know someone who might need this information, please help them and me out and pass this post along.

Have a great rest of the week,