UC Berkeley AI Instructor | ex CIO & CPO

You have an AI workflow idea. Maybe it's automating client onboarding, streamlining reporting, or building an intelligent triage system. The natural instinct? Jump straight to tools — pick a platform, start prompting, see what happens.
The problem: without structural thinking first, you're building on sand. You don't know what decisions the workflow actually makes, what data it needs, where it can fail, or which parts should be automated vs. human-guided. The result? Fragile demos, over-engineered agents, and weeks of rework.
This workshop teaches the most critical step in the Business-First AI Framework: Deconstruct. But you won't just learn a method — you'll use it. The framework is embedded in agent skills that you install in your AI tool of choice. The skills guide you through the deconstruction conversationally — asking the right questions, surfacing what you're missing, and assembling the artifact as you go.
You'll deconstruct your own workflow during the session and leave with a complete Workflow Definition artifact — the essential input for designing and building any agentic AI workflow. Plus the agent skills installed and ready to use on your next workflow.
Introductions, networking, and share what workflow you're bringing.
Workshop overview, what to expect, how the session is structured.
Install the Business-First AI Framework skills for your AI tool or AI code editor.
The 7-step framework overview. Where Deconstruct fits. The Workflow Design Matrix. Three entry points for any workflow.
Apply the deconstruction method to your own workflow using AI-powered agent skills. Instructor coaches live. Stress-test, iterate, and optimize your workflow definition.
Evaluate your completed Workflow Definition. Check for gaps, missing decision points, or steps that need further breakdown.
Share insights, discuss what surprised you, and see how others uncovered hidden complexity in their workflows.
Resources for continuing the work.
You'll go from "I have an AI idea" to holding a structured, buildable specification — the foundation every successful AI workflow needs
Set up the framework's agent skills in your AI tool or AI code editor — Claude Code, Cursor, Copilot, or similar
Leave with the skills installed and ready to deconstruct your next workflow independently, without instructor support
Position workflows by execution mode (augmented vs. automated) and autonomy level (deterministic vs. guided vs. autonomous)
Make informed decisions about how much human involvement and AI decision-making each workflow step requires
Use agent skills to conversationally surface hidden steps, decision points, data flows, and failure modes in your workflow
Apply the method from three starting points — a defined process, a pain point, or a desired outcome
Produce a Workflow Definition artifact — the essential input for designing and building your agentic AI workflow

UC Berkeley AI Instructor | Former CIO/CPO | Trained 5,000+ Leaders
Leaders scoping AI initiatives — You need a structured way to move from "we should automate this" to a concrete specification.
Builders planning AI workflows — You're ready to build but the process you're automating isn't defined well enough to start.
Operators drowning in manual work — You know every edge case and workaround. Now translate that into a spec AI can help you build.
The framework's agent skills run inside your AI tool and guide the entire deconstruction process conversationally.

Live sessions
Learn directly from James Gray in a real-time, interactive format.
Lifetime access
Go back to workshop content and recordings whenever you need to.
Community of peers
Stay accountable and share insights with like-minded professionals.
Framework agent skills
Business-First AI Framework skills — installed and ready for your AI tool
30-minute optional pre-workshop meet & greet
Connect with other participants before the session starts. Share what workflow you're bringing and hear what others are working on.
Certificate of completion
Share your new skills with your employer or on LinkedIn.
Completed Workflow Definition artifact
The key deliverable — a structured specification capturing every step, decision point, data flow, context need, and failure mode. This is the essential input for designing and building your agentic AI workflow.
Real-time instructor coaching
Get unstuck fast. The instructor coaches you through the deconstruction as you work, helping you surface complexity you'd miss on your own.
Maven Guarantee
Your purchase is backed by the Maven Guarantee.

$250
USD