Inside an agent
Understand the six working areas inside a Studio agent: Overview, Settings, Flows, Conversations, Analytics, and Channels.
Use Overview to confirm identity, status, owner, model, temperature, and preview behavior before changing settings or channels.
How an agent is organized.
Read Overview, Settings, Flows, Conversations, Analytics, and Channels as one agent system: identity, behavior, orchestration, evidence, and publishing.
Configure the agent's instructions, model/provider, temperature, reasoning limit, knowledge, tools, playbooks, deployment, and chat behavior here.
Check which workflow responsibilities the agent owns: start flows, lead capture, intents, variables, and draft/live flow status.
Confirm where the agent can talk to customers or teammates before routing sample conversations through it.
Configure the agent's instructions, model/provider, temperature, reasoning limit, knowledge, tools, playbooks, deployment, and chat behavior here.
What you will learn
This page explains what each tab inside a Studio agent is for and what decisions belong there.
Use it after creating an agent, or when you inherit a workspace and need to understand where identity, behavior, conversations, metrics, and channels live.
Overview
Overview is the agent identity and readiness card. It shows the agent name, Agent ID, selected model, created date, status, owner, temperature, and a live preview chat.
Use it to answer the basic control questions: which agent is this, is it live, who owns it, what model is it using, and does the basic chat behavior look sane?
Settings
Settings controls behavior. This is where you edit instructions, model/provider, temperature, max reasoning iterations, Knowledge Bases, Tools, Playbooks, Deployment, and Chat Customization.
Before connecting channels or workflows, make sure the instructions define role, goal, tone, allowed context, allowed actions, escalation rules, and expected output format.
Settings subareas
AI Agent is where the core behavior lives: instructions, model, temperature, and max reasoning iterations.
Knowledge Bases decides which approved material the agent can use. Tools decides which system tools, custom tools, or connected integrations are available to the agent.
Deployment controls whether the agent is Live or Offline and exposes the secret key controls. Chat Customization controls the first message, placeholder, launcher alignment, profile image, nickname, and preview.
Flows
Flows defines the agent's structured conversation paths. The screen shows flows such as Start or Lead Capture, trigger type, Draft/Live status, search, filters, and Create Flow.
Use Flows when the agent needs to ask questions in order, capture information, classify intents, or store variables before a workflow or human handoff.
Conversations
Conversations is the evidence layer. It shows real conversations by contact and channel, sent and received messages, session status, Audit, and the option to respond manually.
Use it to review quality: whether the agent understood the context, asked for the right information, stayed in scope, escalated at the right time, and needs operator intervention.
Analytics
Analytics shows whether the agent is working. Review total conversations, total messages, average messages per conversation, AI credits, activity by time/day, and model usage.
Use it to spot adoption, cost, peak usage, channel activity, and whether instructions, flows, or channels need tuning.
Channels
Channels defines where the agent can talk. This is where you configure Live chat, WhatsApp, Instagram, or Messenger, plus website embed, share URL, split long messages, multimodal capabilities, auto-close, and feedback.
Use Channels late in the setup process: first make behavior and flows reliable, then expose the agent where users or customers will interact with it.
Simple operating order
Configure Settings, connect or review Flows, enable Channels, audit Conversations, and improve with Analytics. Keep Overview as the quick identity and readiness check.
Do not change everything at once. If a response is wrong, start from the Conversation, then inspect instructions, available context, flow logic, channel state, and analytics.
Launch checklist
The agent has a clear owner, status, model, role, instructions, safe temperature, required context, tested flows, one intended channel, and a way for a human to review or respond.
After launch, keep Conversations and Analytics open as your feedback loop. They tell you what users actually ask, how the agent responds, and whether the setup is becoming more useful.
Transcript
Open searchable transcript
FAQs
Which tab should I open first inside an agent?
Open Overview first to confirm you are editing the right agent. Then use Settings for behavior, Flows for conversation logic, Conversations for evidence, Analytics for performance, and Channels for where the agent can communicate.
What is the difference between Settings and Flows?
Settings defines the agent's behavior and access. Flows defines structured conversation paths such as Start, Lead Capture, intents, and variables.
When should I check Conversations?
Check Conversations before and after launch. It is the clearest evidence of whether the agent understood the user, stayed in scope, and handed off correctly.
When should I connect Channels?
Connect Channels after the agent behavior and flows are ready to test. Channels expose the agent to users, so they should come after configuration and review.