Web3 / Crypto

Fintech

Mobile App Design

UX/UI Design

iOS

Workday

The Modern Cash Account, Rebuilt with Crypto

Context

Brookwell is a modern cash account that is designed with blockchain at its core. It routes your balance into stablecoin holdings while letting you move freely between cash and stablecoins for free, so you don't lose value shifting funds around while getting a higher yield.

My Role

Product Advisor Intern

Duration

May - Aug 2026

Team

Workday GO


Design the end-to-end app from the ground up

Objective

001 THE PROBLEM

Agent quality had no home outside of GitHub

THE PROBLEM

Our team was building dozens of automated “Skills” — repeatable workflows that turn customer documents into structured deliverables (benefits workbooks, job catalogs, integration plans, and similar). The library was growing fast, but we had no single place to answer basic program questions:


  • Which Skills are production-ready vs. still experimental?

  • Are we building them the right way, or cutting corners?

  • Who’s contributing, and what changed recently?

  • Where should a program lead focus review time?

THE ISSUE

Manual spot-checks don’t scale. Spreadsheets go stale the day you export them. We needed something that stayed current by default.

So what?

Teams had no centralized place to see what agents, who built them, or how mature they were.

No Centralization

GitHub's commit history and file structure weren't built to represent agent autonomy, or maturity.

Confusing

Signal

Without visibility, Workday risked duplicated work or agents shipped without proper evaluation.

Business

Impact

002 THE SOLUTION

A self-updating quality dashboard for a library of AI-powered implementation workflows

WHAT IS IT?

The Program Health dashboard is a single HTML page that consolidates deployment domain skills into one view surfacing health scores, autonomy levels, maturity stages, and recent activity, so teams can see what's built, what needs attention, and who's responsible for it without digging through GitHub.


It shows:


  1. Summary stats — total Skills, average health score, pass/fail counts, recent commit activity

  2. Autonomy & maturity — how hands-off each Skill is, and where it sits on an Experimental → GA ladder

  3. Health compliance — a ten-metric quality scorecard per Skill

  4. Contributions — who built and improved what over the last 90 days

  5. Development activity — a timeline of recent commits (and open PRs when GitHub CLI is available)



It sits alongside a sibling page, the Skills Map, which answers “what do we have?”

Program Health answers “how healthy is it?”

003 THE BREAKDOWN

Dashboard Overview

  • Skills in library — total count, broken down by Domain / Shared / Builder


  • Health — how many Skills pass evaluation (e.g. 32/40) and pass rate %


  • Needs attention — count of Skills with evaluation errors (clickable — jumps you to the Health section)


  • Activity — number of repo commits in the selected time window (7 / 14 / 30 days)


Summary: Four at-a-glance numbers for program leads: library size, overall quality, what’s broken, and how active development has been. Click “Needs attention” to drill straight into problem Skills.

Autonomy & Maturity

Contributions

Click to see an old iteration!

AUTONOMY

Shows how hands-off the selected Skill category is when it runs. A program lead can see at a glance whether the library leans toward full automation or still needs human involvement

MATURITY

Maps where each Skill sits on the readiness ladder. Click any Skill name to open its full detail panel.

HEALTH

Uses a ten-metric evaluation rubric to cross reference agent health. Surfaces which Skills need review and how serious the issues are. Warnings are fixable gaps; errors are blockers on the quality rubric.

More coming soon…

You caught me! Still working on this one, be back soon. In the meantime, please visit another part of my site.

Let’s talk!

anjali_garg@berkeley.edu