While I've posted here and there over the past couple years, I have largely taken a sabbatical from speaking and creating content. It's been a lot of reflection and gathering my thoughts on things, but it seems that time has come for me to emerge from hibernation. So, last month, I had a chance to present at one of my favorite groups in the DC area: RefreshDC!

It was hosted by the good folks at Goodshuffle who not only provided a beautiful penthouse view of the city but also provided food and drinks.

About the talk

Given the audience was a mix of professionals from the tech industry, I wanted to cover a topic that would hopefully be more relatable across the various disciplines (i.e., engineering, design, product) while not being so basic it's boring. So, after some brainstorming and back and forth with the organizers, here's what I ended up with:

  • Title: TDD? ADD? DDD. The Missing Methodology in the Age of AI
  • Description: The AI conversation is flooded with methodologies for product development: ADD (Agent-Driven Development), which promises "autonomous agents" across your entire workflow, and TDD (Test-Driven Development), which promises agents won't ship broken code. However, both are fragile without the discipline nobody wants to talk about, but the one that matters more than ever in the AI era: DDD (Documentation-Driven Development).

This talk was inspired by the realization that there are a lot of acronyms (BDD, TDD, SDD, etc.) that have been going around with various hyperbolic claims of how it's the best thing since sliced bread.

Key concepts

I'll be doing a separate post on the details of the talk, but as far as some key takeaways, here are the top four:

  1. While there are a lot of people selling the idea of "autonomous" agents, be cautious of such promises as I've yet to see a money-back guarantee with any of these sales pitches.
  2. Methodologies like TDD and SDD are great, but they're not silver bullets. They require discipline and sufficient context for them to forge a resilient process and culture in a team / company.
  3. Documentation Driven Development (DDD) is a methodology that has gone largely ignored by most because it was easier to get away with having fragmented / poor documentation. With the introduction of AI, this is no longer the case and poor docs end up amplifying the magnitude of errors that users and AI agents will make.
  4. While AI has given us new ways to accelerate parts of our workflow, best practices from enterprise products are still relevant today. Rather than reduce the need for accountability and ownership, it is more critical than ever for discipline and deep understanding because our decisions have much wider impacts than ever before.

Unfortunately there is no recording of the talk, but here are my slides to give you a peek into it.

Closing thoughts

Just when I thought things couldn't be more digital with all the screens we have, the advent of AI has made these in-person events even more valuable than ever.

A huge shoutout to Rochelle, Faz, Nick, and the rest of the RefreshDC crew for helping to organize a great event. I had a phenomenal time getting to meet new people in the area while also getting to catch-up with past co-workers and friends!