Channel routing
Decide which channels the bot watches, how it behaves in each, and where escalations land.
The model
DuggAI thinks about Discord in three layers: servers (one or more Discord servers connected to your project), watched channels (specific channels the bot reads in each server), and per-channel rules (autonomy, escalation routing, allowed topics).
Picking watched channels
Good candidates:
#help,#support,#questions#installation,#bugs- Any channel that's implicitly "ask the team"
Bad candidates:
#general— too much chitchat, will dilute your resolution metrics#announcements— read-only, no questions- Off-topic / community channels — false positives waste your team's queue
Per-channel rules
Autonomy
See Discord install → Tuning autonomyfor the three modes. The default ("reply when confident, else @-mention only") is right for most channels.
Topic scope
For mixed-purpose channels, set a topic scope. The bot only replies if it classifies the message as on-topic. Example: in a channel that mixes "help me" and "show off your build," only reply to help requests.
Escalation role
When the bot escalates in a channel, it pings a role. Set this per-channel — your billing team for #billing, your engineers for #bugs. Default is the role you set during install (typically @support).
Working hours
The bot can hold escalation pings until your team's working hours so engineers aren't paged at 3am. Set time zone and hours per-server.
Multi-server setups
Common pattern: one server is your public community (looser tone, the bot is friendlier and chimes in more), one is for paying customers (tighter, more business-like, faster escalations). Configure per-server independently.
Conversation history
Every Discord thread the bot touches lands in your dashboard inbox. You can take over from the dashboard — your reply posts back into Discord as the bot, so the conversation stays in Discord for the user.