Skip to content

Details

What changes when a software team adopts AI from the start, and what remains essential? Drawing on his work with the team building Dreamcatcher at Conquer, Paul Hammond will explore two sides of AI-first development: how the team works together, and how the software is designed and tested. He’ll show how the team uses Talat, interactive explainers and visual Given/When/Then examples to develop shared understanding and agree behaviour before implementation.

He’ll then demonstrate the testing approach and hexagonal architecture behind the application, including a CLI that gives developers and AI agents another way to exercise the system. Through practical examples and demonstrations, the session will explore what has helped, what has needed correcting, and where human judgement continues to matter.

Outline of the session:

- 𝗧𝗵𝗲 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝘁𝗵𝗲 𝘁𝗲𝗮𝗺: a look at Dreamcatcher and what adopting AI from the beginning has meant in practice.

- 𝗪𝗮𝘆𝘀 𝗼𝗳 𝘄𝗼𝗿𝗸𝗶𝗻𝗴: using Talat, interactive explainers and visual Given/When/Then examples to explore ideas, uncover misunderstandings and agree behaviour.

- 𝗧𝗲𝘀𝘁𝗶𝗻𝗴 𝘄𝗵𝗮𝘁 𝗮𝗴𝗲𝗻𝘁𝘀 𝗯𝘂𝗶𝗹𝗱: acceptance testing, test-driven development, mutation testing, and how we assess the quality of the tests themselves.

- 𝗛𝗲𝘅𝗮𝗴𝗼𝗻𝗮𝗹 𝗮𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 𝗶𝗻 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗲: separating domain behaviour from infrastructure, enforcing architectural boundaries, and exercising the system through the application and CLI.

- 𝗟𝗲𝘀𝘀𝗼𝗻𝘀 𝗹𝗲𝗮𝗿𝗻𝗲𝗱: what has changed about the work, what still requires engineering judgement, and where the approach is continuing to evolve.

ABOUT PAUL

Paul Hammond is a software engineering consultant and owner of Pack Software Limited, with around 20 years of industry experience and more than a decade working with Extreme Programming. His work centres on test-driven development, domain-driven design and software architecture, with a particular interest in how fast, trustworthy feedback helps teams build better software.

He is working with the team at Conquer on Dreamcatcher, a collaborative application that helps engineers and domain experts explore and understand business domains together. Alongside building the product, the team is exploring how AI can support collaboration, learning and implementation, while keeping people responsible for agreeing behaviour, making design decisions and judging the results.

- LinkedIn: https://www.linkedin.com/in/paul-hammond-bb5b78251/
- GitHub: https://github.com/citypaul
- Technical blog: https://feedbackdriven.dev/

You may also like