Skip to content

SOP: Channel Deployment

Goal

Deploy a configured AgentBrain agent to one or more end-user channels (Slack, Microsoft Teams, LINE, Zalo, WhatsApp, web, API).

Prerequisites

  • An agent that is fully configured and tested in the AgentBrain console
  • Admin credentials for the target channel (Slack workspace admin, Teams admin, LINE OA owner, etc.)
  • A public hostname for AgentBrain (or a tunnel for non-public test environments)

Channel Selection

Pick the channel(s) based on audience:

ChannelWhen to pick
WebInternal portal or public-facing site
SlackInternal employees on Slack
Microsoft TeamsInternal employees on Microsoft 365
LINEAPAC customer audiences
ZaloVietnam-domestic customer audiences
WhatsAppInternational customer audiences
REST APICustom apps, mobile, IVR
WebhookTrigger from any system that emits events

Steps

1. Open Channels in the agent

In Agents → Your Agent → Channels, click Add Channel.

2. Pick the channel type

Choose from the supported list. AgentBrain shows the channel-specific configuration form.

3. Authenticate the channel

  • Slack: install the AgentBrain app to your workspace and authorize scopes
  • Microsoft Teams: sideload the AgentBrain manifest or distribute via your tenant
  • LINE: paste your Channel ID and Channel Secret from the LINE Developers console
  • Zalo: paste your OA ID and access token
  • WhatsApp: paste your business account ID and API token
  • Web: copy the embed snippet to your site
  • REST API: copy the channel endpoint URL and rotate keys

4. Configure routing

  • Map channel events to your agent: direct message, mention, scheduled trigger
  • Configure ignore patterns (channels, users to skip)
  • Set rate limits if you expect high volume

5. Test in a non-production channel

  • Send a test message from a non-production workspace, channel, or number
  • Confirm the agent replies as expected
  • Inspect the audit log for the channel.message and agent.invoke events

6. Roll out to production

  • Move the channel binding from staging to production
  • Announce the rollout to the affected audience
  • Watch the monitoring dashboard for the first day for errors

Verification

  • A test message produces the expected reply
  • Audit log shows the channel event and the downstream agent invocation
  • Rate-limit, retry, and error handling work under sustained load

Rollback

  • Pause the channel binding in Agents → Your Agent → Channels
  • Revoke the channel-side credentials if needed (Slack app, LINE token)