AI Enablement Engagement

Session 3 — AI-Assisted Development Foundations (Detailed Agenda)

Total Duration: 2 hours (120 minutes) Phase / Week: Phase 2, Week 3 Format: Hands-on workshop — developers on their machines Required attendees: All Turner developers participating in the engagement

Note. Hands-on exercises target the sister product's codebase, not the pilot's. Each developer needs the sister product cloned locally. If pilot spike code exists, have it available as a secondary target.

Pre-session prep (Turner)

  • Claude access provisioned for every developer attending
  • Each developer has cloned the pilot product's repo locally
  • Recommended IDE extensions installed (per a setup guide Nubitz sends one week prior)

Agenda

0:00 – 0:10 (10 min) — Setup verification and Claude product surfaces tour

  • Lead: Nubitz
  • Format: Demo + smoke-test
  • Confirm everyone is logged in and working. Tour Claude Code (CLI/IDE), Claude.ai chat, and how they differ. When to reach for each.

0:10 – 0:30 (20 min) — Prompt engineering fundamentals

  • Lead: Nubitz
  • Format: Demo with audience prompts
  • Anatomy of a productive prompt: goal, context, constraints, examples. Iteration patterns. Live demonstration with a real code sample from the pilot product. Audience suggests the next prompt; Nubitz drives.

0:30 – 1:00 (30 min) — Exercise A: "Explain and refactor"

  • Lead: Developers (Nubitz floats)
  • Format: Hands-on
  • Each developer picks a file from their cloned pilot product repo. Use Claude to explain what it does, then propose a small refactor. Compare results in pairs. Capture three prompts that worked well for the Developer Reference Guide.

1:00 – 1:10 (10 min) — Break

1:10 – 1:35 (25 min) — Exercise B: "Tests for a real function"

  • Lead: Developers (Nubitz floats)
  • Format: Hands-on
  • Pick a pilot product function lacking tests. Drive Claude to produce a test suite. Run it. Iterate on failures. Discuss what Claude got right and where it needed correction.

1:35 – 1:55 (20 min) — Exercise C: "Second opinion on a real PR"

  • Lead: Developers (Nubitz floats)
  • Format: Hands-on
  • Each developer pulls an open or recently-merged PR from the pilot product. Ask Claude for a review. Compare its findings against what the human reviewers said. Note where AI added value and where it missed context.

1:55 – 2:00 (5 min) — Debrief and capture for Developer Reference Guide

  • Lead: Nubitz, facilitating
  • Format: Synthesis
  • Each developer shares one prompt or pattern that worked and one that didn't. Nubitz captures the keepers for the reference guide. Preview Session 4.

Materials

  • Prompt-pattern handout (Nubitz)
  • Setup verification checklist (sent in advance)
  • Shared note-doc for capturing keeper prompts

What Nubitz produces after this session

  • First draft of the Developer Reference Guide's "Foundational Patterns" section, populated with Turner-specific examples

Risks to manage

  • Setup drift: if any developer can't connect to Claude in the first 15 minutes, pair them with another participant and resolve offline rather than blocking the room.
  • Demo creep: keep Nubitz demo time tig