Load human context: make the agent brief you on your terms

Matt Wynne has been programming since the mid-1980s and advancing agile practices since discovering the C2 Wiki in 2001. He joined the Cucumber open source project in 2008 and co-authored The Cucumber Book with Cucumber's creator. In 2013 he co-founded Cucumber Ltd, growing it to roughly $1.5M in annual revenue while training thousands in Behaviour-Driven Development, and developed the Example Mapping practice along the way. After SmartBear acquired Cucumber in 2019, he led the open source project until 2023.

Aldric Giacomoni has been building with AI agents since 2023, and for the past 18 months his agentic interface has been his primary way of writing software. He taught the Web Development Immersive at General Assembly and, with Zach Marcin, built a complete training program for Mechanical Orchard's mainframe-modernization product. He is the author of org-gtd (a GTD package for Emacs), Stone (an agentic harness framework), and rakka (BEAM/Erlang primitives on top of Racket) — all built entirely with agents since early 2025.

Jeremy Lightsmith cares about complexity. Across 25 years in tech he has stayed deeply technical while managing structural complexity not just in codebases but across teams and organizations — as a developer, consultant, agile/lean coach, startup founder/CEO, and now executive coach. Most recently he spent three years at Mechanical Orchard anchoring delivery teams and building tools as they used LLMs to modernize mainframes for Fortune 500 clients. Today he helps senior leaders manage new forms of complexity, navigate accelerating change, and rebuild trust as engineering loops are rewritten.

The most effective software practitioners make their whole team better, not just themselves. From pair programming to consulting to teaching I've spent my entire career focused on effective knowledge sharing — growing and helping those around me grow. Lately that's meant learning, using, and teaching how to effectively use AI to modernize mainframes.

Used well, an LLM is a huge multiplier for one engineer. Share that skill across a team and it goes exponential. Build a system that improves along with your team's knowledge, and it goes hyperbolic. That's what I'm excited to share: how to build and improve agentic systems you and your team can rely on for trustworthy, high-quality outcomes.