top of page
Screenshot 2026-11-12 at 13.15.02.png

Panaseer Key Drivers

AI-powered investigation assistant that explains the why behind a metric change.

​

Role: Sr. Product Designer

Scope: Workshop facilitation, discovery, strategy, UX/UI, prototyping, and launch

Collaboration: CPO, CTO, Engineering, Data Science, and customers

Technology: AI 

 

Context: Key Drivers helps security analysts find the reasons behind a metric change in seconds. Instead of only seeing what happened, they understand why the number dropped and can close security gaps faster.

​

It was Panaseer’s first AI-powered feature, so I also defined how AI looks, feels, and earns user trust in the product.

INSPIRATION AND MOTIVATION

When I started, Panaseer had no AI features, while the market was filling up with AI-driven products. Attackers were using AI to move faster, too. I didn’t want us to fall behind, and the leaders went through another round of investment, so I strongly advised bringing AI into the product ASAP.

As a product experience leader, I facilitated the Future of AI workshop with our CTO, CPO, lead data scientists, and lead engineers. Rather than wait for a perfect long-term strategy, we agreed to start with one clear problem: slow, manual investigation. Our guiding question was:

How might we help security teams spot and close their control gaps faster, at a time when attackers are also using AI to move quicker?

OPPORTUNITY

key drivers feature.png

RESEARCH: AI-ASAP WHERE THE USER NEEDS IT THE MOST

“AI-ASAP” didn’t mean adding AI everywhere. It meant starting where users needed it most. Analysts spent days navigating dashboards to understand why a metric changed, so we started on the Metric Data page and learned from real users along the way.

​​​

Over almost a quarter, I ran user research with 20+ security stakeholders from the top finance, healthcare, and communications organisations worldwide. I started with about a dozen interviews with Data Security Analysts to understand their pain before we moved into delivery, then tested prototypes with them.

​

Two findings shaped the design:

  • They were excited about AI. Across very different security teams, users were keen to try AI, but for cyber teams it was unexplored territory, so my main goal became building trust.

  • They preferred words to charts. Most users preferred natural-language explanations to graphs or tables, because they could understand the change and copy it into their reports much faster.

AI DESIGN SYSTEM STEPPING STONES

In February 2025, I began exploring how to communicate an AI-powered feature alongside everything else in the product. There was no precedent, so every decision would become the starting point for the AI features that followed.

​

Two goals pulled in different directions. I wanted AI-generated content to stand out, so users could tell at a glance what was AI and what wasn’t. But I also wanted it to feel native to our in-house design system, because consistency builds trust. In a security product, an analyst who doubts an insight won’t act on it, and an AI feature that feels bolted on invites that doubt.

Screenshot 2026-10-19 at 09.51.06.png

How might we build user trust in an AI system?

Screenshot 2026-10-19 at 11.37.53.png

​​​

To build user trust through design and user experience, I treated the work as a set of small, reusable decisions rather than a one-off screen:

​

  • A recognisable AI identity: a consistent way to signal AI-generated content, using the common purple colors that were out there in the market - so users always know what they are reading.

  • Built from existing components with a twist: I extended the design system instead of working around it, so the feature felt familiar from the first click. On top of that, I used the AI spark icon - another recognisable symbol of an AI-powered feature. 

  • Natural-language insights as the core pattern: after usability testing showed users preferred plain explanations to graphs and tables, I designed around readable typography and new icons that will support the copy-ready summaries interaction.​​​​​​​​Patterns 

Screenshot 2026-11-12 at 14.59.46.png

PRODUCT DEVELOPMENT

Made to be reused

I documented each design choice so the next AI feature could build on it rather than start from scratch.

 

I created a Markdown pocket guide for the engineering team, and a year later it was used to build the Cyber Advisor.

​

These stepping stones, with the visual language and patterns first set here, then became the foundation for our Claude Code design system,

Screenshot 2026-11-12 at 15.00.21.png

PRODUCT DELIVERABLE & LEARNINGS

Metric Data page -Key drivers (1).png

Outcome and Key Learnings

 

- Key Drivers shipped to all Panaseer customers as the first product in the Panaseer AI suite. 

- I established the AI Innovation Lab program - early user test with and on live customer data.

- Its learnings led to the next Gen-AI features, MetricIQ and then the Cyber Advisor.

​​

​

​

Most importantly: what the launch taught us

​

In an AI innovation team, risk is part of the job, and the launch didn’t go as we hoped.

Adoption in Pendo was low; Customer Success and users struggled to understand how to use the feature, and customers raised questions about regulation since this was the first AI feature, with some opting out of AI altogether.

 

We responded right away:

  • Internal demos for CS and Sales, so they could confidently present the feature to customers

  • A Trust Centre on our website and Help Center, plus an AI governance document

  • A Pendo guided tour that I created to support first-time users

  • Feedback sessions with customers and SMEs (subject matter experts)

 

My main lesson: with AI, the launch and the trust around it need as much design as the feature itself. We gave the Gen-AI features that followed a longer launch process built on these steps.

  • LinkedIn
  • Facebook
  • Instagram

© 2026 designed by NBDF.

bottom of page