Pre-Conference Workshop

PRE-CONFERENCE WORKSHOP

Beyond the Basics: Starting Your Software Factory

Tim Rayburn

Your Instructor: Tim Rayburn

Vice President of Consulting at Improving | 11x Microsoft MVP

Tim Rayburn is a Vice President of Consulting with Improving Enterprises, a software and project management consulting company in Dallas, TX. He is a passionate advocate of apprenticeship and mentorship in corporate environments. He has been awarded the Microsoft MVP in recognition of his expertise and community leadership 11 times.

He is also a conference organizer, an author, and can be found speaking across the nation on agile processes, software design, and career management. He lives in the North Dallas area with his amazing wife Kate and their loyal dogs Dresden and Radagast.

When: Full Day (9:00 AM - 5:00 PM)

Where: Improving - 11675 Rainwater Dr #100, Alpharetta, GA 30009

Level: Intermediate to Advanced

Register Now

What You'll Build

In this full-day hands-on workshop, you'll design and deploy a working Orchestrator Harness: a custom multi-agent system that takes a feature request from idea to shipped, tested, and documented software with minimal human intervention. By the end, you'll have a functional template you can bring back to your team.

The Challenge

The problem you're solving. You've adopted an AI coding tool. Your developers love it. But orchestration is still manual: copy-pasting requirements between tools, manually triggering each phase (design → code → test → review), waiting for human judgment at every checkpoint. You're treating a production pipeline like a playground. The promised move from "ideas take weeks" to "ideas take hours" never materializes because the human bottleneck moved, not disappeared.

The workshop approach. This is not a lecture. We will build incrementally, starting with a simple request-to-delivery pipeline and progressively adding sophistication.

Key Outcomes

  • Build a working Orchestrator Harness
  • Design multi-agent systems
  • Implement quality gates and failure recovery
  • Add human-in-the-loop decision points
  • Create observable, trustworthy pipelines
  • Adapt the template to your own team

Skills You'll Gain

  • Multi-agent orchestration patterns
  • Work-item routing and task breakdown
  • Automated quality gates
  • Failure handling and self-correction
  • Pipeline observability
  • Team customization strategies

Prerequisites (sent to attendees ahead of time)

  • Required: An active subscription to any AI coding CLI tool (Claude Code, GitHub Copilot, Devin, or equivalent) with API access enabled
  • Bring: Laptop with Python 3.10+, Node.js 24+, Git, and a code editor (VS Code recommended)
  • Comfortable with: Python scripting, async/await patterns, basic git workflows
  • Nice to have: Familiarity with multi-agent systems or agentic loops (explained as we go)

Setup help available 30 minutes early (8:30 AM) for anyone who needs assistance. Clone the workshop repo, test your API key, and ensure you can make API calls to your chosen AI coding platform before the workshop.

Full Day Agenda (9:00 AM - 5:00 PM)

9:00-9:45 AM

Foundation: Understanding the orchestration problem

Live walkthrough of what goes wrong when you chain agents naively. Hands-on: set up your development environment (local harness, API credentials, test workspace). Outcome: Everyone's harness boots and runs with a working baseline.

9:45-10:45 AM

Building the pipeline I: Planner and Implementer agents

Design the work-item router (idea → task breakdown). Build a Planner agent that turns raw requirements into a work breakdown, and an Implementer agent that turns a task into code. Write and iterate the prompts yourself.

10:45-11:00 AM

Break

11:00 AM-12:00 PM

Building the pipeline II: Quality gates and the end-to-end run

Add a simple quality gate (tests pass/fail → retry vs. escalate), then test the pipeline end-to-end with a real feature request. Outcome: Your orchestrator routes a feature from idea to merged code before lunch.

12:00-1:00 PM

Lunch

1:00-2:15 PM

Scaling patterns: Handling failure, specialization, loops

How to handle hallucinations and bad code without manual debugging; when to use parallel agents vs. sequential. Add a Code-Review agent that forces architectural decisions before implementation. Wire feedback loops so failures self-correct before escalating.

2:15-3:00 PM

Human-in-the-loop and observability

Where to place human decision points (approve the plan, approve the merge, approve the escalation) and how to make the harness observable enough to trust: run logs, per-agent transcripts, and a dashboard. Hands-on: add an approval gate and a run log to your pipeline.

3:00-3:15 PM

Break

3:15-4:30 PM

Adaptation lab: Point the harness at your own work

Adapt the template to your own context: swap in your team's standards and conventions, adjust the quality gate to your test stack, and run a feature request shaped like your real backlog. Outcome: The take-home template is already customized before anyone leaves the room.

4:30-5:00 PM

Wrap-up and next steps

Walk through the architecture you've built: what generalizes and what's team-specific. Planning: how to adapt this for your team's codebase, standards, and tooling. What to tackle first when you get home, plus Q&A.

You'll Leave With

  • A working orchestrator harness (yours to customize for your tool choice)
  • A mental model for when to use parallel vs. serial agent execution
  • Patterns for quality gates, failure recovery, and human-in-the-loop decision points
  • A local GitHub repo template you can extend and share with your team
  • Four concrete "Monday morning" next steps tailored to your team size and current AI maturity

Ready to Automate Your Software Factory?

Secure your spot in this exclusive workshop. Limited seats available!

Register for Workshop Now