Content engine: A UX playbook for skills and AI tools

Moss · 2026

UX principles, an automated quality check and onboarding blocks that cover the whole journey of a skill and AI tool, from finding it to understanding its value.

Role

Senior Product Designer. Research synthesis, UX principles, the skill UX checker, the block system and the workflow. I built the blocks in the production website code with Claude Code and Prismic.

The Get started block

Problem

We wanted to offer skills and AI tools to finance professionals. To do that well, we had to look at the whole journey: people have to find a skill, trust it, install it, use it and make sense of what it gives them. There was no playbook yet for that, so I set up a system that covers every step and solves for what people need at each one.

Goal

My team’s hypothesis was: skills and AI tools can work as awareness pieces for Moss, if we design for what holds finance professionals back: not understanding what a skill or an AI tool is, not trusting what it can access and not knowing whether it worked. And if every skill and tool meets the same quality bar.

Success metrics

  • UsablePositive feedback after using a skill or tool, and download and install numbers.
  • Consistent and high qualityEvery article follows the same template, and every skill and tool passes the UX check.
  • FastTeammates produce an article page quickly, because the system carries the design knowledge, and quality is checked by the UX check, not by hand.

Challenges

  • No playbook yet

    No research or guidelines were available to understand what defines good UX in a skill.

  • Knowing a skill worked

    Only once a finance professional has used a skill or tool and got the output, we would know if it had worked.

  • My first engineering work

    The first time I created a schema and UI in our content management system, Prismic.

  • Non-tech-savvy users

    Most users don’t have technical skills to understand the ins and outs of AI.

Solution

  1. Three findingsThree research findings: not understanding what a skill is, not trusting what it can access (69% had avoided AI for finance over privacy or security concerns) and not knowing whether it worked
    ResearchSolves no playbook

    Desk research on adjacent evidence (Anthropic documentation, public discussion and trust studies from comparable products) and surveys. 69% of finance professionals had avoided AI for finance at least occasionally over privacy or security concerns. The research showed where people get stuck across the journey, which I translated into 11 UX principles.

  2. The skill UX checkerA Claude chat: the skill UX checker reports one blocker, two improvements and passed checks for a variance commentary skill, each with a fix
    The skill UX checkerSolves no playbook

    The 11 UX principles as a pass/fail audit to check a skill or AI tool: one concrete fix per fail, blockers and suggestions for improvements.

  3. The trust blockThe Security and privacy block: data goes from the providers to your computer to the dashboard, is stored locally and covers only usage and billing
    Before the downloadSolves readers and engineering

    A design system called 'Blocks' that is additive to an article, while the existing prose and H1 stay the same, since the articles rank well already.

  4. A skill’s outputA Claude chat: the skill has written a variance commentary as a Word document to download, asks whether it came out right and after Looks good offers to talk to Moss
    After the outputSolves knowing it worked

    Output that finance professionals can put their name on in a shareable format. Collecting feedback at the end, which turned out more challenging than anticipated.

Outcome

A research-backed playbook, including the UX check for skills and tools which the team finds useful and uses regularly. Teammates now build articles on their own that follow the guidelines. The problem we ran into once again: the articles had no real distribution, so few people found them. That’s where the AI Hub comes in as a new distribution channel for Moss.

Back to home