# Seth Gerard > Design leader for AI-native products. Former VP of Design at Pax8. 19+ years in complex B2B platforms, with a specialty in data-dense visualization. ## Contact - Email: seth.gerard@gmail.com - LinkedIn: https://www.linkedin.com/in/sethotron - Resume (PDF): https://sethgerard.com/Seth-Gerard-Resume.pdf - How I lead: https://sethgerard.com/how-i-lead.md - Contact page: https://sethgerard.com/contact.md ## Selected work ### Managing digital labor at scale - Org: Pax8 (2026) - Role: VP of Design · Design lead - Timeframe: 6 weeks · 2026 - Status: Prototype · early access targeted Oct 2026 - Focus: AI-native product design, Agentic workflows, Data visualization, Rapid prototyping, Visual design - Outcome: 100K+ AI workers it's built to manage, across hundreds of clients - Access: summary public; the full case study unlocks with a password from Seth's application, or on request - URL: https://sethgerard.com/work/fleet A control surface that grows with the provider, from a first AI worker to a fleet across hundreds of client businesses. A hundred thousand AI workers. One question: what needs me today? Pax8 is building for a new kind of provider: one that takes on a small business's AI workers, its digital labor, and runs them as a service. The providers already doing it had no way to see their whole fleet, only one customer at a time. As VP of Design and design lead for Pax8's new agent-first business unit, I pitched a control surface that grows with the provider, from a first visit to a fleet across hundreds of clients, and built the prototype myself in Claude Code. Pax8 is taking it to early access in October 2026. - **100K+** AI workers it's built to manage, across hundreds of clients - **2 weeks** from pitch to working prototype ### Building the two-sided marketplace - Org: Pax8 (2023–25) - Role: Sr. Director → VP of Design · design direction - Focus: B2B SaaS & marketplaces, Product strategy, Visual design, Design leadership - Outcome: 98% opt-in adoption of the new marketplace - Outcome: 70 → ~90% vendor self-service completion, pilot to today - Access: summary public; the full case study unlocks with a password from Seth's application, or on request - URL: https://sethgerard.com/work/marketplace Turning a sales-driven cloud distributor into a product-led marketplace, for the MSP partners who buy from it and the vendors who sell through it. A distributor takes orders. A marketplace makes connections. The Pax8 platform was built to sell cloud software to its partners: thousands of managed service providers (MSPs). In 2023 we set out to turn that distribution platform into a true two-sided cloud marketplace, a dual flywheel. On one side, software vendors reach new markets through Pax8's partners. On the other, partners get the tools to build, sell, and manage solutions for the small businesses they serve. I set the design direction for both sides, starting with design-led workshops that turned the CEO's vision into prototypes, and led the designers who built them: first a partner marketplace that replaced the old platform for nearly every partner, then a self-service marketplace for vendors, launched as an MVP. ![Pax8's marketplace at the center, surrounded by the MSP partners it serves. One arm reaches to vendors: products and prospects flow in, recurring revenue flows back, and the hook is a complete small-business go-to-market. The other arm reaches to small businesses: managed solutions and support flow out, recurring revenue flows back, and the hook for partners is a solutions and marketplace platform.](https://sethgerard.com/work/marketplace/two-sided-marketplace.svg) ### Maturing design at Pax8 - Org: Pax8 (2022–26) - Role: Sr. Director → VP of Design - Timeframe: 4½ years · 2022–26 - Focus: Design leadership, Org building, Design systems & tokens, UX research, Content strategy - Outcome: 3 → 20+ design, research, and content - Outcome: ~90% design system adoption across product verticals - Outcome: 4 directors - Access: summary public; the full case study unlocks with a password from Seth's application, or on request - URL: https://sethgerard.com/work/org How I built design, research, and content into one organization at Pax8, with four directors, a shared design system, and a seat in planning. From three designers taking requests to a core function the business plans around. Pax8 had grown fast as a sales-driven cloud distributor and was turning itself into a product-led technology company. I joined in 2022 as Senior Director of Product Design, alongside a new head of product, and together we designed the product organization design would live in. Over four and a half years I built product design, UX research, and content strategy into one organization with four directors, a design system that most of the product was built on, and the rituals that made design part of how work got planned rather than where it ended up. I became VP of Design in 2024. - **3 → 20+** designers, researchers, and content strategists, plus 7–10 contractors - **4** directors across design, research, and content, three grown into the role - **~90%** design system adoption across product verticals ### Seeing cyber threats in real time - Org: ProtectWise / Verizon (2016–21) - Role: Lead UX Architect; later de facto design lead for Verizon's security products - Focus: Data visualization, Visual design, UX research - Access: public - URL: https://sethgerard.com/work/protectwise A network-security product for analysts, built around dense, real-time visualization with a film-grade look. An interface that looked like it came from a film, for analysts who needed to find a live attack in seconds. ProtectWise recorded a company's entire network traffic in the cloud and continuously replayed it against new threat intelligence, so analysts could see attacks as they happened and find ones they had missed. Its Visualizer was one of the most complex, data-rich products I've worked on. I designed across the whole of it, and my part was making that density work for the people who used it every day. #### The Visualizer The Visualizer's look came first: its visual direction was set by a designer known for film interfaces, and it was deliberately dense, dark, and alive. Rings of protocol and threat activity spin around a globe, events stream in from the edge, and everything moves in real time. It was striking, and it was a lot. The design work was to make that complexity usable for analysts working an incident under pressure, without flattening what made it powerful. [Video: A tour of the Visualizer: the HUD's rings of traffic and threats, a kill-chain ring, an attack traced across a map, Explorer's network graph and connection curves, and a 3D view of traffic](https://sethgerard.com/work/protectwise/visualizer-tour.mp4) *ProtectWise Visualizer tour: HUD, Killbox attack trace, Explorer, and a 3D traffic view. Video capture courtesy of Richard Trott.* I joined as one of two Lead UX Architects reporting to the VP of Engineering. We made the case for a full-time UX researcher, and her field research with analysts, who walked us through real incident-response scenarios, grounded the redesigns that followed. When the other designer moved to a longer-range labs effort, I took over design for the whole core product. I built it hand in hand with the lead UX engineers, Richard Trott and Jason Johnston, whose craft is a big part of what you see here. The Killbox is the analyst's inbox for threats as they happen, and the most-used part of the product. In 2018 I redesigned it with the product manager and our lead UI engineer around what customers had told us for years: letting analysts group events, by host for example, so a burst of related alerts reads as one story, not fifty rows. ![The Killbox: threat events grouped by IP address, each row with a threat score, observations, and hosts](https://sethgerard.com/work/protectwise/killbox-1.jpg) *The Killbox redesign, grouping events by host.* ![Explorer: a network graph of hosts and connections above a table of flows](https://sethgerard.com/work/protectwise/tour-explorer.jpg) *Explorer: following an attack across hosts. Capture: Richard Trott.* Not everything shipped. I sketched hive plots as a way to show a day of threat observations across three axes at once (attack stage, device, and protocol), then turned one axis into a timeline and made it filterable. ![Four hive plots of threat observations, colored by severity and protocol, with one device selected](https://sethgerard.com/work/protectwise/hive-plots.jpg) *An exploration: hive plots for threat observations. A concept, not shipped.* #### Sensors Everything in ProtectWise depended on sensors: lightweight software on a customer's network that captured the traffic. If a sensor went quiet, the customer was blind, so sensor health had to be impossible to miss. The original sensor page showed a card for every sensor. That made sense with a handful. Large deployments ran dozens, across regions, and the cards turned into a wall you had to scroll and read one by one to find the one in trouble. [Video: The old sensor page: a grid of identical cards, one per sensor, scrolled through one by one](https://sethgerard.com/work/protectwise/sensors-before.mp4) *Before: one card per sensor.* [Video: The new sensor dashboard: sensor groups by region with rolled-up health and throughput, failing sensors in red, expanding to individual sensors](https://sethgerard.com/work/protectwise/sensors-after.mp4) *After: groups roll up, trouble shows in red.* I designed the new dashboard with the lead engineer and product manager, bringing in subject-matter experts as we needed them. Sensors roll up into groups, by region or however a customer organized them, and each group shows its health, throughput, and capture at a glance. Anything offline or struggling shows in red, so trouble surfaces before you dig in. Then you drill in: open a group to see its sensors, and a sensor to see its charts and manage its configuration. > Design for the deployment a customer will have, not the one they have today: roll up to spot trouble, drill down to fix it. ![The sensor dashboard with sensor groups for Asia Pacific and Europe, per-sensor bandwidth and capture charts, and one sensor flagged as needing a restart](https://sethgerard.com/work/protectwise/sensors-big-chart.jpg) *The shipped dashboard: regions roll up, sensors drill down.* #### Bring Your Own Intel ProtectWise detected threats with IDS rules in the Suricata format, mostly from commercial and open-source feeds like Emerging Threats. Those covered most threats, but intelligence analysts always had special cases: rules they wanted to tune for their own systems, or rules they had written themselves. Bring Your Own Intel opened the platform to them. Analysts could upload and manage their own rule lists, tailor what the product watched for, and see how their rules were performing. I designed three parts: the workflow for adding intel, list management, and a performance dashboard showing which rules were firing and what had changed. ![The intel performance dashboard: observation counts over time, a list of rule changes, and contributing rules with a detail chart](https://sethgerard.com/work/protectwise/byoi-dashboard.jpg) *The intel performance dashboard. Names blurred.* ![An intel list: rules with their message, kill-chain stage, category, severity, and confidence](https://sethgerard.com/work/protectwise/byoi-list.jpg) *Managing a rule list. Names blurred.* We also talked about where it could go next: a marketplace where analysts could buy and sell intelligence sets. We never built it, but the idea stayed with me. #### What came next In 2019 Verizon acquired ProtectWise. I was pulled in to work with the owner of Verizon's unified security platform, ATLAS: a large effort, already under way, to bring dozens of security products, most of them with little or no interface, under one experience. The direction I gave it was "unified access, not unified interfaces": a central security home, single sign-on, a shared navigation and product switcher, and a common component library as a starting point rather than a mandate. I shaped that direction for a while, and the team was building some of my recommendations when I moved to a different team at Verizon. Looking back, the lesson I carry from ProtectWise is about complexity. People can handle far more than we give them credit for. Nobody expects to fly a 747 the first time they see the cockpit, and even pilots retrain when they move to a new plane. Software is the same: people learn through repetition, and their understanding grows with use. So I'd rather test whether people can learn something than whether they get it at first glance, and design to bring them along: roll-ups before details, a calm first step, and an escape hatch when it gets to be too much. It's the same bet this site makes: a sky to explore, and a plain version to read for anyone who'd rather not.