LobeHubFrom the builder
Proactive AI Agents: When a Bot Decides to Speak Up
LobeHub v2.3.0, released on 1 November 2026, makes proactive participation the default: bots read watched channels and decide whether to speak, stay quiet, or wait instead of waiting for an @ mention. The essay argues that speaking unprompted is easy, while making mistakes reviewable is the real product job, because unsolicited posts multiply interruption cost across a channel.
The release adds Wake-policy controls for duties and pace, and writes each judgment into auditable records people can mark right or wrong to track proactive wakes and error rates per channel. Device-local CLI-quota agents still start with wake off, and sensitive channels can disable proactivity.
LobeHub removed it in v2.3.0, released on 1 November 2026. Bots now read the channels they watch and decide for themselves whether to speak up, stay quiet, or wait.
Saved . A shortened excerpt from the official blog; the original publisher retains authorship.