At Knowledge 2026, @Henrik Hiort and I ran a 90-minute hands-on lab, Design Faster, Smarter: Prototyping with Horizon, Figma, and AI.
The room was full of admins, implementors, and builders: people who configure and extend ServiceNow every day but don't necessarily think of themselves as designers. That's exactly who this lab was for.
We've all been there: the requirements looked fine on paper, you built exactly what was asked, and then someone tried to use it. The form was confusing, it solved the wrong problem, or you heard "that's not what I meant." The rework takes just as long as the original build, except now you're also managing stakeholder disappointment.
This lab was about catching those problems before they become expensive ones.
What the lab covered
This wasn't an advanced Figma tutorial or a strict AI demo. It was about learning a repeatable workflow you can take back to your team the next time someone says "we need a better way to do X."
We structured the 90 minutes around four steps:
- Understand the problem
- Identify a solution
- Make it tangible
- Validate with your users
Step 1: Understand the problem
We started by sharing a scenario featuring a fulfiller persona: someone who processes and resolves work on behalf of others in a ServiceNow Workspace. The scenario described a real role, a task they perform, and a specific friction point they run into.
Participants then used an AI tool of their choice (Claude, ChatGPT, or whatever they preferred) to quickly define the user's primary goal, biggest frustration, and a "How Might We" statement to anchor their design thinking.
The key idea: AI is great at helping you articulate a problem clearly and fast, so you don't spend your prototyping time solving the wrong one.
Step 2: Identify a solution
Next, we introduced Horizon, ServiceNow's design system and design enablement platform. Instead of jumping straight into Figma, participants explored Horizon's app frameworks, patterns, and components with their scenario in mind, looking for existing, proven solutions to address their user problem.
The goal wasn't to memorize every component. It was to see what the platform already makes possible, so you're designing within reality rather than outside it. Participants returned to their AI chat, which already had context about their persona and problem, and asked it to help map their "How Might We" statement to specific Horizon patterns and a rough screen flow.
Step 3: Make it tangible
This was the build step. Working in pairs with Figma Make, participants created 2–3 clickable screens that told a story from the user's trigger to at least one resolution point.
A mantra we kept coming back to: clarity over polish. A rough prototype that tells the story is more valuable than a pixel-perfect screen that doesn't.
Step 4: Validate with your users
Now, to figure out if the solution we designed works for the people we built it for. Pairs swapped prototypes and gave each other focused feedback using a simple structure: "I understood…" and "I wasn't sure about…" It's surprising how much you learn from watching someone click through something you built, even when it's rough.
The workflow
If you take one thing from this post, it's the four-step workflow:
Understand → Ideate → Build → Test
- Understand your user's problems, goals, and constraints.
- Ideate within what the platform makes possible.
- Build something you can test — fast.
- Test. Validate your approach with actual people before you commit time and energy to implementation.
This isn't a framework that requires design expertise. It's a thinking workflow that helps anyone — admins, developers, architects — make better decisions about what to build and why, before investing in the full build.
Try it yourself
We built a lab guide so you can run through the entire workflow on your own or with your team. It includes sample scenarios, copy-paste AI prompts for each step, links to Horizon patterns, and guidance on how to evaluate your output along the way.
🔈 Important note: Never input company secrets or sensitive info into third party tools!
Take a problem your team is currently solving and run it through the workflow. You just need a scenario, an AI chat, Horizon, and Figma. The lab guide walks you through all of it.
If you try it, we'd love to hear how it goes.
