Managing a Team of AI Agents: What Nobody Tells You

Forward There's no way I would have had the time to review all of the Mattermost chat logs, agent memories, and agent memchats to condense and write up this great timeline of events, observations, and learnings. Best I could do was ask Riley to do it and let me review it, which I did. This came out way better than I expected. There may be some minor tweaks from me, and I've commented in each section with my point of view, where needed, but otherwise its pretty accurate. I mean its all in the various memory records and chat logs we kept working on MASON. --dpark 4/1/2026 Edit: For full disclosure this was also posted to my SubStack under my pen name Kenji Nakamura. There's some additional commenting I added to that article that are not present here. Mostly comments about some of the experiences I had in some of the activities below. The Setup In January 2026, I sat down at my desk and typed "hi" into a chat channel. Five AI agents said hello back. They had names, roles, and opinions. By