Implementing AI tools in SkillsWave's design process

Building AI into the design process without losing the user-centred thinking behind it.

My Role: Manager of Product Design

2 Designers, 45-Person Company

October, 2025-present

I implemented AI tools in our design process in a meaningful, sustainable, and collaborative way

As AI started changing how product teams worked, I wanted to make sure our design team was learning alongside it without losing the user-centred thinking behind our process. I introduced a series of experiments to understand where AI could genuinely improve the way we explore, prototype, and make product decisions, then worked with development to turn the most useful ones into part of our everyday design practice.

In short, I led the design team to meaningfully integrate AI into our design process, making it part of our everyday tool belt without losing core design and user-centred thinking.

First, I organized a workshop and planned two experiements

I organized a workshop with the head of design at Plotly to come in and explain his process using AI tools. For context, in October 2025, designers just beginning to venture into creating PRs, fully fleshed "production-like" prototypes, and many experiments that fell somewhere between Lovable prototypes and real working code.

After that workshop, I set a few principles for how I wanted the team to approach AI. I wanted it to support our design thinking, not replace it; I wanted us to learn quickly by experimenting and sharing feedback; and I wanted the team to know that nothing was permanent. If something wasn’t useful, we could always revert back. These principles gave us a bit of a guardrail before we started figuring out where AI could actually fit into the design process.

Then from there, I created two experiments:

Experiment 1: Try AI Prototyping in a product-like environment with the ability to share prototypes (Explore & Check)

Experiment 1 would solve our immediate problem with analytics and could be a game changer for us making our ideas more tangible, including testing interactions and visuals quickly. The measure for success would be: Could we communicate our ideas more efficiently and effectively using these tools, or were our traditional Figma methods still winning out?

Experiment 2: Design ships small PRs with dev approval (Make & Implement)

The Result: We had a tangible way to move forward testing AI tooling to see what would work for our design process

Next, I found more opportunities for design to contribute, starting with admin roles and permissions

Once the team had a clearer way of working, I started looking for places where design could help SkillsWave make better product decisions earlier. I did this because I wanted to showcase design's value, with our new process, as fast as possible. The original "dev-only" admin roles and permissions project became my next opportunity.

Changing roles and permissions from "a dev project" to a growth project

At first, roles and permissions looked very technical, but I saw roles and permissions as critical for future product growth. Roles and permissions would shape how SkillsWave scaled operationally and could influence future customer-facing admin experiences. I also saw this as an opportunity for one of my newer designers to learn more about the product and grow in system analysis skills.

I created an admin inventory board and gave design a meaningful space to contribute

An image of the admin inventory board I created and collaborated with my designer on. The left says "My Problem Refinement and Context setting", the middle says "My structure, an 'Admin Inventory', and the right says "My team member's exploration, each square was a different part of admin!"

I carved out time before roles and permissions for design to work on this project before development would start. I started with creating an “admin inventory” board, planned out the system analysis pieces, then told my designer to document every feature and learn how it worked.

She discovered that roles and permissions would be integral to many internal team workflows. Almost every other team outside of product used admin. If we removed important permissions, we could disrupt business. Design became integral not only to creating future-proof roles and permissions, but creating the internal permission sets themselves. My designer took on putting the permission sets together and designing the roles and permissions page that was later re-used in our customer facing self-serve offering.

The Result: Design made key contributions to a previously "dev-only" project and the work was re-used later in a customer facing feature. Design's work on admin roles and permissions resulted in smoother admin workflows, a more scalable permissions model, and a meaningful growth opportunity for a designer on my team.

Then, I jumped in with design strategy to create clarity around PLG

Then, I jumped in with design strategy to create clarity around our self serve (PLG) offering

When SkillsWave spun out, there was a clear business goal we were heading towards: introduce a self-serve (product led growth) option for mid-market employers, nicknamed PLG.

At the beginning of the project, everyone had a different definition of PLG. Product thought it was onboarding, marketing was focused on pricing and acquisition, and some teams thought it was a new feature launch.

What is Product Led Growth (PLG)?

I saw an opportunity for design to create shared understanding before the company moved too far into siloed execution

With the PLG project, I saw an opportunity to foster alignment and move forward as a group. This would prove design's value more, and show the value of design strategy in the most impactful project. My goal was to thoroughly understand PLG as a general business model, the user experience pieces related to that model, and then create a shared language, understanding, and a journey that everyone could map their activities to when thinking about PLG and SkillsWave.

A section of my PLG research Miro board, taking in inputs from across the company, secondary research, and more.

To do this, I took all of the research and inputs and created a living document called The Stages of PLG. Each stage had a part of a user's self-serve journey accounted for, from searching and browsing our marketing site through to paying for upgrades. This gave us a simple way to bucket all of the work, user goals, business metrics, KPIs, and more into tangible stages of what our product would look like in a PLG model.

The first draft of the high-level stages of PLG for SkillsWave. These stages have been updated since this initial draft, but further details have been omitted due to confidentiality.

I also mapped SkillsWave’s existing product experience to those stages, highlighting where users already experienced value, where the journey broke down, and where new product work would be needed. This helped turn a broad business strategy into some actionable next steps.

A product journey map of SkillsWave, key moments of existing value, gaps, and opportunities, all mapped to the PLG stages.

After this research, I presented it to all of the senior leadership team, starting with the stages of PLG and then talking specifically to the CEO and product team about the customer journey map. After seeing the stages of PLG, leaders understood where their work fit in our PLG initiative. People started using the language and stages outlined in the presentation, and people moved forward with clarity, framing their work in the stage of PLG they were focused on.

With that in hand, the product team began confidently working on the Setup and Showcase phase. The stages of PLG and the customer value journey map allowed us to move forward with confidence that "Set up and Showcase" would unlock the ability to test our PLG offering in the market. It also helped us stay aware that more work needed to be done after this stage to provide value within our product during the PLG workflow.

A designer on my team owned the detailed design, while I helped set the PLG direction through critique and working sessions. I pushed for the experience to feel simple, encouraging, and celebrate moments of success, while also being measurable enough for us to monitor and iterate after launch.

Result: My design strategy and framing gave the company shared language for framing their work in the context of the whole PLG journey. The first phase of our self-serve offering launched in June 2025 and has helped support new pilot clients, enabled sales to give their prospects a free offering to try, and is in the process of helping open a door to a large confidential client opportunity.

From there, I made user research a repeatable team practice

SkillsWave didn’t have a dedicated research team, so once I had a couple of instances proving the value of the design team, I knew I needed to focus on research. Research could no longer on D2L's solid user research team or longer timelines. We needed practical, quicker ways to learn continously from our users.

To solve this, I started building research into the team’s rhythm through lightweight studies, client partnerships, and quick but structured synthesis. Additionally, I wanted to build the team's confidence planning and facilitating research and get us all in the rhythm of regularly looking for opportunities to learn from users with our available resources.

SkillsWave Guide Focus Group and 1:1 Interviews (our first study)

SkillsWave didn’t have a dedicated research team, so research could not depend on good conditions, a user research team, or long timelines like we had at D2L. I knew the design team needed practical ways to learn continuously from users.

I started building research into the team’s rhythm through lightweight studies, client partnerships, and quick but structured synthesis. Additionally, I wanted to build the team's confidence planning and facilitating research and get us all in the rhythm of regularly looking for opportunities to learn from users with the resources we had.

I partnered with Client Success to reach a large existing client to run ran a series of focus groups and interviews, and helped translate the findings into product improvements. I also looped one of my designers into facilitating the 1:1 interviews and analyzing the results of this study.

To start, I partnered with Client Success to reach a large existing client and run a study to understand how they were using SkillsWave Guide. After some discussions, they agreed to a series of focus groups and 1:1 interviews on the condition that we would share back some of the findings. We ran 5 focus groups and 7 1:1 interviews asking people about their experience using SkillsWave Guide and walking through their current set up of it.

This study was completed by myself and another designer on my team. I set up the logistics, organized the sessions, and came up with most of the activities for the focus groups, and my designer focused on the 1:1 interviews and the post-study analysis.

SkillsWave Guide Study Results: A very successful study that led to practical SkillsWave Guide and other improvements including better skill searching, better search in our catalog, and better priority of recommendations

This is a Miro board we used for the SkillsWave Guide research. It includes background questions and ice breakers, and a Rose, Bud, Thorn activity where users could mark different parts of SkillsWave Guide with what they liked and what they struggled with.

SkillsWave Guide Study Results: A very successful study that led to practical SkillsWave Guide improvements and general product improvements including better skill searching, better search in our catalog, and better priority of recommendations. The designer in my team also learned a lot about facilitating sessions and session analysis that they have brought into other studies since then.

After the study, we improved search on our skill set page by adding skills to the search context and highlighted those skills within the skill sets after search. Before this study, only skill set names could be searched.

Second Example: L&D "System Implementation" Research

I recruited through internal networks and ran journey-mapping sessions with L&D leaders to understand how they evaluate, adopt, and roll out new tools. This was done through "advertising" my plans at multiple senior leadership calls. Senior leaders reached out to me with contacts and I interviewed them using this live journey mapping activity below as a guide.

This was a "live journey mapping" activity I created for L&D leaders to walk through a time when they implemented a new tool or system in their company. I took notes and built the journey map as they were speaking. Details removed due to confidentiality.

L&D "System Implementation" Research Results: The report became an input into our self-serve PLG initiative to start to understand how L&D leaders think about implementing new software systems in their companies.

Additional benefit: Team Research Habits and designers who can all run usability research sessions

I coached designers through research planning, interview guides, facilitation, note-taking, synthesis, and sharing insights back to the team. Now all of my team members have planned, facilitated, and analyzed at least 2 usability studies each, and one of my team members created templates for future research recruiting, planning, and analyzing.

The Results: Research has become a regular part of how we design, and the business strongly supports us continuing to look for ways to connect with our users. We regularly connect with client success on opportunities for research and everyone on my team knows how to plan, facilitate, and analyze a usability study. We've run 5 studies so far! Our latest study used AI prototypes made with Claude 🎉

Where we are now: A trusted design team with a strong voice in shaping product strategy and direction

The design processes, strategic frameworks, and team research practices I put in place during SkillsWave’s first years helped establish design as a respected function across the company. We now play a key role in shaping new features and workflows, and we are known for bringing clarity, polished experiences, and strong cross-functional collaboration. The foundation I built has also created new opportunities for the team, including budget for AI tooling, more direct access to clients, and multiple usability studies with both clients and non-clients.

The design processes, strategic frameworks, and team practices I put in place during SkillsWave’s first years helped establish design as a respected function across the company. We now play a key role in shaping new features and workflows, and we are known for bringing clarity, polished experiences, and strong cross-functional collaboration. The foundation I built has also created new opportunities for the team, including budget for AI tooling, more direct access to clients, and multiple usability studies with both clients and non-clients.

Since 2024, the design team has designed and helped launch:

A self-serve product-led growth model
SkillsWave Guide
A new learner profile
Improved semantic search and skill tagging
A new employee widget
Multiple improvements to provider and approval workflows
And more! (and more on the way!)

What's next: Incorporating AI tools meaningfully into our design process

AI tools are new and exciting with tons of potential, but I am working on being intentional about where they are best used in our design process. Similar to my question at the beginning of this case study, can we use AI to move meaningfully faster, without losing user-centred rigour?

A sneak peak into where I've started 👀

All designers at SkillsWave have local dev environments on their machines and access to Claude and Cursor. With AI we've been able to experiment with making direct contributions to small PRs, realistic and more feasible prototypes for developers as we are fleshing out designs, and fully interactive user research prototypes for our usability studies.

Where we've found AI to be most meaningful so far is in idea exploration and experimentation. It has given us an ability to see detailed interactions without having to slack message a developer. Our AI prototypes are also a lot faster to build for user research (no more Figma noodle arms)

These were two explorations made completely from AI. These are not difficult to create in Figma, but we were able to fully utilize the exact same graphing library that the developers would be using. This is also one of MANY analytics and graphing explorations I explored when putting these widgets together.

Some areas I'm still exploring right now:

How can we move faster, and where should we be intentional about spending time/not moving too fast?
I'm seeing some potential in AI helping us upskill in accessibility. I'm in the process of building an accessibility Claude skill that can help with identifying accessibility gaps in our existing product and point out tips in Figma designs for making designs more accessible.
Is AI helpful when it comes to small polish fixes in our product?

✨ All that and more coming soon!