accenture

Practically AI

A lightning talk by Rick Whalley

Introduction

Who am I?

  • Manchester Engineering Practice Lead
  • AI-Native Engineering Lead

Accenture is hiring

AI in Enterprise Architecture

  • Brownfield Work
  • Topologically Complex Enviroments
  • Tribal Knowledge

The Approach

  • Build living harnesses
  • Valuestream Context
  • Skills, Skills, Skills
  • Hot takes

Living Harness

  • Init is your start point
  • Claude.md is a target
  • Progressive disclosure is king

Value Stream context

  • Microservices context is siloed
  • Psuedo-mono repo using meta-repo
  • Flow based implemention

                        // .meta file (using the `meta` tool)
                        "projects": {
                            "services/repo-a": "https://github.com/org/repo-a",
                            "services/repo-b": "https://github.com/org/repo-b"
                        }

Skills, Skills, Skills

  • Cheapest investment for biggest results
  • /skill-creator
  • Superpowers
    • Brainstorming
    • Sub-agents

Hot Takes

  • CLI > MCP
  • Defining your own agents is pointless

Thanks

fourohfour.dev