Skip to main content

Quickstart Overview

Use this guide to move from an empty workspace to a support-ready Woes setup.

Setup Checklist

1

Create or open your workspace

Confirm the workspace name, support URL, timezone, and company details in Settings.
2

Add source context

Add at least one source in Context. Woes works best when the source includes endpoint paths, auth rules, examples, request bodies, response shapes, and error cases.
3

Create or tune an agent

Use Agents to attach source context to an agent. Each agent can have its own context, behavior, guardrails, and widget key.
4

Install a customer channel

Install the live chat widget, connect an email domain, or connect Discord.
5

Invite operators

Add team members in Settings → Users and assign the right role.
6

Review analytics

Use Analytics after the first conversations to inspect answer rate, skip reasons, confidence, surveys, and feedback.

Workspace Surfaces

SurfaceUse it for
InboxLive customer conversations across live chat, email, and Discord.
IssuesDurable support work, standalone tickets, linked conversation issues, notes, tags, and workflow state.
ContextAPI docs, repositories, specs, endpoint Markdown, auth configuration, and test readiness.
AgentsMulti-agent setup, context attachments, prompts, guardrails, learning, and widget keys.
AutomationsTriggered workflows for assignment, status, priority, tags, notes, SLA, snooze, CSAT, and auto-replies.
AnalyticsOperational reporting for conversations, agents, workforce, SLA, surveys, live chat, sources, clients, and feedback.
SettingsWorkspace configuration, channels, keys, users, roles, surveys, notifications, tags, macros, statuses, and install snippets.

What To Avoid

  • Do not paste production credentials into API source content.
  • Do not expose provider or model routing to customers.
  • Do not expect the agent to answer from memory when source context is missing.
  • Do not use public widget keys as account-level API credentials.
For best results, add one high-quality source first, test the agent with exact endpoint questions, then expand context coverage.