AI Hub: A new distribution channel for Moss

Moss · 2026

A public website where finance professionals discover skills, tools and guides to get more done with AI, and where Moss builds reach and authority on AI for finance.

Role

Senior Product Designer. I covered research, positioning, concepts, testing and the design system, prototyped in code with Claude and handed over in Figma.

Explore page

Problem

LinkedIn was our main distribution channel, and every tool or skill we shared there got no traction. Our earlier top-of-funnel experiments had shown us how to grow organic search and AEO successfully. Based on those learnings we built a content engine that was producing plenty of content, but it was scattered across the Moss website and gave search engines and AI assistants no clear signal that Moss owns this topic.

Goal

Create one home for everything we’ve published and will publish, including our new MCP, which had no channel yet, and connect it to our content engine so the team can add new content easily. Since this is a top of funnel project, our aim is to reach people and create brand awareness, leading to conversion into qualified leads.

Opportunity

Competitor analysis showed that skills, MCPs and marketplaces were all aimed at tech-savvy people. But many finance professionals aren’t, and AI can feel intimidating to them. My main focus was building the AI hub for the people who feel left out, since that will automatically include everyone else too.

Challenges

  • An ambitious deadline

    2 weeks from start to launch.

  • A new audience mid-brief

    What started as a skills hub for existing customers became a public website, open for non customers and optimised for a less tech-savvy audience.

  • Quality at speed

    I had to build tools that enable the team to create high-quality content themselves, and be confident it would hold up.

Solution

  1. Search with examples
    Search and exploration in oneSolves new audience

    Visitors either arrive with a problem or just want to browse. Testing confirmed both, so the hub leads with problem-first search and still gives room to explore, in plain words, with no assumed knowledge.

  2. Design system: componentsParts of the hub design system: Copy prompt and Add as skill buttons, a Best match badge, format filter chips with counts, a Moss MCP card and tags
    A design system for the teamSolves deadline and quality at speed

    Templates for each content type, card rules, search, empty states and animation behaviour. This lets the team build fast and every page still comes out consistent and high quality.

  3. The MCP pageThe side panel of an MCP use case page: free, 2 minute setup, a note that the Moss MCP needs connecting and Copy prompt, Add as skill and Share buttons
    A home for our MCPComes from the opportunity

    Competitors built MCPs for tech-savvy people. I made ours understandable for everyone, and gave it a public home that LinkedIn never could.

Outcome

I designed the AI hub and its system in 2 days so engineering had enough time to implement everything. It took us a total of 2 weeks to ship in the end. I’ve already iterated on the MCP template, and a teammate used it to publish around 60 pages at a consistent quality. We are now starting to measure traffic against reach, authority and use.

Back to home