After the Agent Writes the Code — Wellington AI Meetup, October, 2026

​Welcome to another Wellington AI Engineering meetup. We have two talks on what happens around the code your agents write — what your team keeps from every session, and how quality keeps pace.

​MC for the evening is Seb Krueger, CTO of ZunoSmart Labs.

​RUNNING ORDER

​5:00 PM — Doors open
5:10 PM — Pre-talk networking
5:25 PM — Kick off
5:30 PM — Talk 1: Aaron Rama, Keyhook
6:00 PM — Talk 2: Gisela de la Villa, Trade Me
6:30 PM — Pizza, drinks and more networking, courtesy of Joel from Younity Recruitment

​———

​TALK 1 — Aaron Rama, Co-Founder and CTO of Keyhook (20 min)

​Buy the Intelligence. Build the Advantage.
How we are turning off-the-shelf AI into a system that learns from every workflow, correction and mistake.

​We're all running the same coding agents. What separates teams is what happens after the session ends, and for many of us the answer is nothing. The correction you made this morning is gone, and the person sitting next to you hits the same wall next week.

​KADE is what we built to fix this for us. It's our agentic development environment, with every agent, diff, question and decision in one place, so running several agents at once is no longer a juggling act. Behind it sits the Keyhook Brain, where corrections get reviewed like code and shared with the team, so every feature and every bug fix makes the next one better. An agent should only ever make the same mistake once.

​———

​TALK 2 — Gisela de la Villa, Engineering Manager at Trade Me (20 min)

​The AI-First SDLC: Scaling Delivery Without Eroding Quality

​Integrating AI across the development lifecycle promises incredible speed, but faster code generation quickly pushes bottlenecks onto quality verification. To prevent this from eroding standards, we must shift quality practices earlier into the cycle and transform how developers and testers collaborate. I will share how Trade Me reviews our end-to-end pipeline to identify exactly where AI can actively participate, focusing on our shared "skills marketplace" and an automated test companion tool. Expect concrete techniques to help you accelerate software delivery while maintaining reliability.

​———

​IMPORTANT NOTE: Please do not come up to Level 2 until it's 5pm. One of our organisers will let you know when you can come up :)