Building an AI Practice and 340+ Skill Agent System at Artium

Overview

Two parallel workstreams at the intersection of AI product strategy and hands-on delivery.

Skillet: AI Agent System

Skillet started as a personal productivity experiment and grew into a 340+ skill AI agent system powering Artium's consulting practice. Each skill encodes a reusable consulting workflow -- from discovery facilitation to story writing to client coaching -- with structured inputs, knowledge references, and output templates.

The system transforms how product teams integrate AI into their work: instead of ad-hoc prompting, consultants reach for purpose-built skills that embed best practices and institutional knowledge. Skillet covers product management, engineering practices, design, facilitation, client engagement, AI adoption, and business development -- organized as a searchable catalog with cross-references and quality standards.

I designed the skill architecture, wrote the authoring conventions, built the knowledge canon system, and created the intake pipeline for new content. The system runs as a Claude Project with canonical reference documents, prompt templates, and a content governance model that keeps skills accurate and current.

Helia: Clinical Trial Simulator

Led product on Helia, a clinical trial startup simulator built for Fortrea, a $2.7B contract research organization. Fortrea runs thousands of clinical trials annually; startup (the phase before patient enrollment begins) is one of the most complex and delay-prone stages.

Helia applied deterministic AI to model site selection, enrollment forecasting, and startup timelines. I authored the PRD, defined the simulation logic with clinical operations SMEs, designed the output format for trial managers, and shaped the product's position within Fortrea's broader digital transformation strategy.