Strategizing for Success - scale your AI solution faster with rapid prototyping

Why prototype at all?

The baseline objective for any product manager is to meet the users’ needs. That means moving rapidly through the product development process – without sacrificing quality, of course! Prototyping allows you to explore and validate your AI product ideas rapidly so you can actually meet their expectations faster than the rest of the market.

Why is speed important in product development? Because of the ugly truth that we all know about:

• 90% of startups fail, of which 75% are venture-backed. (Startup Genome)
• 35% of startup failures happen due to lack of market need. (CB insights)
• 80% of features in the average software product are rarely or never used. (Pendo)

The best solution in the world will fail if the user can’t see the value, and prototyping helps you get to that value as quickly and cost-effectively as possible.

Then there’s risk management. As I see them, the ‘Big 4’ risks for any AI product are:

• Desirability – does the market want your product?
• Feasibility – can you actually make it happen?
• Viability – should you really build it?
• Usability – will users be able to use it?

Again, rapid prototyping is a key approach to mitigating these risks for your product and can be used at any stage of product development. The ‘classic’ development process has four stages: prototyping, minimum viable product (MVP), product-market fit, and scaling. Taken at face value, it would appear prototyping is only used in the early stages. However, at any point in your AI product’s development, you may be considering new design ideas, fresh features, or other user-focused changes. For example, when scaling your product to a wider market, new features can expand your target user base. Prototyping is ideal for rapidly testing these new ideas or features to confirm their feasibility (and desirability, viability and usability!) before investing in the new direction.

Take Advantage of the AI Revolution

Want to know how to use AI-powered solutions to optimize your development costs? Describe your needs, and our team will come up with a custom idea.

Marek Szydłowski

Generative AI Engineer

Olivier Halupczok

Generative AI Engineer

Get in touch!

Benefits of prototyping your AI product designs

Why build a prototype? There are a number of business reasons to quickly create a simplified or demonstration version of your product and use it to gather feedback; namely:

• Move faster than your competitors
• Incorporate user feedback as soon as possible
• Minimize development costs
• Mitigate risks

What’s more, a prototype (similar to an MVP in this way) is simply an impactful way to show the basics of your product to people and generate interest and enthusiasm, whether they are potential users, investors, or other stakeholders.

AI prototyping – getting it right

So, rapid prototyping…. how is it done? Let’s lay out the high-level steps in order…

1. Start with an appropriate strategy – You need to know both your current position in terms of development, investment, etc. and also exactly what you aim to achieve with your AI prototype serve (i.e. where you want to get to; what information do you aim to gather, etc.) Then you can lay out clearly what must be done, and in what order, to achieve your goals.
2. Get to know your customers / users – Often we believe we know our target users when really we don’t. Create a representative user persona, including their interests, goals, problems, preferred apps/tech, etc. (See below for more on user personas). By segmenting your potential users like this you can create a clear target audience.
3. Create assumptions and experiments – This is the heart of prototyping, identifying what assumptions or untested beliefs you have about your AI product, and constructing an appropriate prototype (wireframe, sketch, Wizard of Oz setup, etc.) to test them by experiment. (Again, see below for more detail).
4. Star with the riskiest – Prioritize your experiments, testing the assumptions that represent the greatest risk to your project first. Beware of testing the aspects of the product that appeal to you most – your favorite parts – because they are rarely the most important features or aspects for the user. In other words, avoid vanity testing.
5. Collect necessary feedback – Conduct your prioritized experiments and collect the resulting data.
6. Make sure that you work on the results – Analyze the results and act on them. This might seem obvious but too many startups get into a cycle of constant prototyping; testing, testing, testing and never pivoting. If you get negative results, accept them and act on them.

Tips for creating user personas

As mentioned in step #2 above, you can have an idea of what you want to achieve, to build, and to have a prototype to validate the inherent assumptions or unknowns, but you also need to be clear on who you’re building your AI product for.

A top tip is not to rely on simple demographics. After all, both King Charles and Ozzy Osbourne are the same age, live in castles, and are super-rich. But they probably have very different needs when it comes to AI software and apps!

User personas are based on particular needs, wants, pain points, etc. Think about what your product is aiming to address – those are the areas of commonality to look for.

I highly recommend two specific tools for constructing user personas easily and quickly: SparkToro and delve.ai. Both crawl through data on the Internet to create personas for your users based on the tech they use, the content and news they read, their business interests, the brands they like, the hashtags they use, their frequently used phrases, and so on.

Identifying assumptions & creating hypotheses

You have your persona. You know who you’re building your solution for. Now it’s time to experiment, but first you need to know what it is you’re testing. What are the assumptions or hypotheses that your experiments are focused on?

A useful tool for creating hypotheses for experimentation is Strategyzer, which breaks down the hypothesis statement into four elements:

1. “We believe that…” – This is the idea statement, your assumption.
2. “To verify that, we will…” – This is the nature of the experiment, how you will test the assumption, the method.
3. “And measure…” – Now, add your metrics; the specific measures that tell you what data you will collect.
4. “We are right if…” – This section lays out your success criteria, allowing you to decide whether the data you collect proves your assumption or not.

What tests can you use when prototyping?

Different tests produce different kinds of evidence – weak or strong – and this, together with time and cost considerations, drives your choice not only your choice of experimental method but also how far you can rely on the results.

Weak evidence is:

• Based on opinions
• What people say
• Lab settings
• Small investments

Strong(er) evidence is:

• Based on facts
• What people do
• Real world settings
• Large investments

For example, interviews tend to produce weak, opinion-based evidence but can give you a clear insight into user attitudes and mindset and can be used to engage with your target audience. Usability testing in which test participants interact with the prototype product or a mock-up give stronger evidence that can directly confirm (or not!) the value of the user experience.

Your choice of testing method depends on what assumption you are testing and what will move the development forward.

To generalize the qualities of some common testing approaches:

• Interviews – low cost; medium time commitment; weak evidence.
• Dedicated campaigns (such as generating interests via social media, Google Ads, etc. involving clicks and interactions) – medium to high cost; medium time commitment; medium-strength evidence.
• Trading features (letting customers prioritize the direction of development by asking them about potential new/enhanced features) – low cost; low time commitment; medium-strength evidence.
• Mash-ups (combining existing technology options to approximate the product feature to be tested; a kind of a Frankenstein solution) – medium cost; medium time commitment; strong evidence.
• Presales (getting people to sign up and/or buy before the product is ready to release) – medium cost; medium time commitment; strong evidence.
• Simulations (of UX) (e.g. Wizard of Oz tests where someone is ‘behind the curtain’ mimicking what the app/feature will do in a convincing manner to simulate the real experience for the user, who believes the experience is real; i.e. a human being writing replies in place of the AI) – low cost; medium time commitment; strong evidence.

Your choice of experiment depends on what you’re testing and what you need to get from the results.

Where can you find your tests?

Another highly useful tool I like to recommend is Lyssna (previously known as Usability Hub). This platform offers a wide range of prototyping tests that can validate your design decisions with real users. For the test participants, you can choose from Lyssna’s own database or you can draw on your own community of users and direct them to specific tests via the platform. Lyssna is an extremely powerful tool, with users that fit your persona, and provides good evidence via a variety of tests.

Design sprints – a rapid prototyping strategy

To put the ‘rapid’ into your rapid prototyping activities, the design sprint approach is well proven, following a set number of stages, often carried out at a rate of one per day, meaning you conduct the whole testing process in a single week:

1. Day One – Understand – Who are the users? What are their needs? What is the context? Competitor review. Formulation of strategy.
2. Day Two – Diverge - Envision. Develop many solutions. Ideate.
3. Day Three – Decide – Choose the best idea and storyboard it.
4. Day Four – Prototype – Build something ‘quick and dirty’ to show to users. Focus on usability and not on making it beautiful!
5. Day Five – Validate – Show the prototype to real users outside of the project. Learn what works and what doesn’t.

A design sprint is a simple and straightforward approach to testing solutions to user needs. For more on design sprints, check out our article: What are Design Sprints?

Continuous prototyping

To reiterate a point made at the beginning of this article, prototyping is not a one-off activity. In fact, it can (and arguably should) take place throughout the product development process, up to and including scaling your AI product for sustainability and stability in a wider market.

Everything we’ve covered so far is ultimately a case of collecting data to hone the development approach. A blanket term for this is ‘discovery’. Discovery should be continuous throughout the development process. At every stage, as you look at new design decisions, you use prototyping to test out the new direction or feature with your users.

For more on continuous discovery, check out our article, 5 reasons why Continuous Discovery is the new standard in product development.

Want to see your app integrated with AI?

Set up a call, and we will create a working proof of concept for you, even on the same day

Set up a call

Examples of Boldare’s prototyping

The following is a quick tour of some Boldare case studies – real projects in which we have applied prototyping principles with success (most of the prototypes outlined here were created as clickable prototypes using Figma, another highly valuable design tool).

A banking digital marketplace - The goal was to engage corporate customers by allowing them to exchange business opportunities supported by the bank’s financing, as well as benefit from commercial services and a loyalty program.

sonnen charger - The app syncs with the sonnen Charger device. Users can set the charging mode when charging an electric car. The charging status plays a key role in the app. That’s why an animation was used, which allows the user to quickly check the current status. Two animated moodboards were prepared to present ideas for the visual side of the app and the user experience.

Inceptua medical access - A clickable dashboard for a medical application. The prototype was created to demonstrate the form-building feature to potential users.

Rise ‘Start Right’ portal - A clickable dashboard for the Rise ‘Start Right’ portal to support graduates in the job market. The prototype was created to demonstrate the platform for students and future employers. The whole idea was to help students with their first step into the professional world after graduating from a university.

AI app prototype – the key takeaways

I know from experience that prototyping is a tool that should be used at every stage of product development. Its success for your AI product development (and your business goals) depends on having a clear strategy laying out what you aim to measure and why. The specific prototyping method you use will vary from project to project and will depend on the kind of evidence you need to gather, and the time and budget available. Finally, never be reluctant to prototype your AI product. The earlier in the process you begin testing your assumptions and unknowns, the better. Even if the results aren’t what you hoped for, they still offer valuable information – don’t put off the pivot!