Documentation

Administrator Help for 123AIChat

This guide is designed for deployment owners, security leads, and platform operators running 123AIChat in LAN or on-premise environments.

On this page

Setup and Access Control AI Models and Routing Strategy Channel and Bot Operations OpenClaw Readiness and Approvals Meeting Workflows Fast Troubleshooting Order

Setup and Access Control

First-run sequence: create admin account, complete setup, then configure AI in Admin > AI.
Use role boundaries intentionally: admin for system control, moderator for channel governance, member for participation.
Confirm registration policy before rollout by checking system configuration.

AI Models and Routing Strategy

Verify at least one available model before expecting bot replies.
Routing precedence is global default -> channel override -> bot override.
Start with one stable global default and add scoped overrides only when behavior requirements are clear.

Channel and Bot Operations

Use @BotName for direct bot invocation and @bots for multi-bot brainstorming.
Keep each channel role-focused to reduce signal noise and improve decision quality.
When creation fails, check license/seat limits and remove unused bots first.

OpenClaw Readiness and Approvals

Readiness checks should follow order: installed -> running -> agents available.
Use approval mode for controlled execution in sensitive channels.
Approval states normally move through pending -> running -> success/failed/rejected.

Meeting Workflows

Lifecycle: start -> kickoff round -> next round -> end -> summary -> history/detail.
Round topics can auto-trigger @bots when configured.
Summary generation is designed for ended meetings; fallback behavior may appear when AI generation fails.

Fast Troubleshooting Order

Check prerequisites first: auth/session, model availability, runtime status.
Then validate configuration: routing, channel settings, bot settings.
Then validate permissions and scope: admin/moderator/member boundaries.
Finish with explicit verification checks to confirm recovery.

Deep-Dive Pages

AI Routing Strategy

Production-safe routing model across global, channel, and bot scopes.

Open Page

OpenClaw Approvals

Readiness checks and state model for approval-driven execution.

Open Page

Troubleshooting Runbook

Recovery sequence for auth, routing, permissions, and verification.

Open Page

Frequently Asked Questions

Why does a bot exist but not reply in channel?

Most cases are model readiness or routing issues. Confirm an available model is configured and that routing resolves correctly at global/channel/bot level.

What is the safest way to start AI routing in production?

Set one stable global default first, validate behavior in a pilot channel, then add channel or bot overrides only for explicit requirements.

How should admins handle OpenClaw in sensitive workflows?

Enable approval-driven execution, verify runtime readiness before sessions, and keep moderator/admin responsibilities explicit in channel governance.

Review Platform Features Review Architecture Talk to an Implementation Specialist