Agent Studio consolidates agent management into a single workspace. You no longer need to jump between separate sections for each agent type.
Access Agent Studio
Navigate to AI Agents > Agent Studio in the left sidebar. The Agent Studio dashboard displays all existing agents with their status, type, and performance summary.Agent types
Agent Studio supports three core agent types:Conversation AI
Text-based agents for SMS, live chat, Facebook Messenger, and Instagram DM. Handle FAQs, book appointments, and qualify leads automatically.
Voice AI
Phone-based agents that answer inbound calls and make outbound calls using natural speech. Ideal for appointment confirmations, lead follow-up, and after-hours support.
Content AI
Content generation agents that produce on-brand marketing copy, social posts, email drafts, and blog outlines using your business context.
Create a new agent
Open Agent Studio
Go to AI Agents > Agent Studio and click the Create Agent button in the top-right corner.
Choose an agent type
Select the type of agent you want to build: Conversation AI, Voice AI, or Content AI. Each type opens a purpose-built configuration wizard.
Name and describe your agent
Give the agent a clear name (e.g., “Sales Chat Bot” or “After-Hours Voice Agent”) and add an internal description so your team understands its purpose.
Configure behavior
Set the agent’s personality, tone of voice, and operating rules. For Conversation AI, choose between Suggestive and Auto-Pilot mode. For Voice AI, select the voice profile and call handling rules.
Assign a knowledge base
Link one or more knowledge bases so the agent draws on accurate, up-to-date information. Go to AI Agents > Knowledge Base to create or manage knowledge sources.
Set channel assignments
Choose which channels the agent operates on. Conversation AI agents can be assigned to SMS, live chat, Facebook, or Instagram. Voice AI agents are assigned to specific phone numbers.
Agent templates marketplace
Agent Studio includes a template marketplace with pre-built agent configurations for common use cases:- Lead qualifier — asks discovery questions and scores leads before routing to sales
- Appointment scheduler — checks calendar availability and books meetings
- FAQ responder — answers common questions using your knowledge base
- After-hours agent — handles inquiries outside business hours and logs callbacks
- Review responder — drafts replies to Google Business Profile reviews
Managing multiple agents
The Agent Studio dashboard provides a centralized view of all agents:| Column | Description |
|---|---|
| Name | Agent name and internal description |
| Type | Conversation AI, Voice AI, or Content AI |
| Status | Draft, Active, or Paused |
| Channels | Assigned channels or phone numbers |
| Conversations | Total conversations handled (last 30 days) |
| Last edited | Timestamp and editor name |
Testing and debugging
Open the test console
Click the Test button on any agent card to open the built-in test console. This simulates a real conversation without affecting live contacts.
Run test conversations
Type messages as if you were a customer. The agent responds using its current configuration and knowledge base. Review the response quality, accuracy, and tone.
Review conversation logs
Open the Logs tab to inspect past conversations. Each log entry shows the contact message, the agent’s response, the knowledge source used, and any actions triggered.
Best practices
Start with Suggestive mode
Start with Suggestive mode
When launching a new Conversation AI agent, use Suggestive mode first. This lets your team review every response before it reaches the customer. Switch to Auto-Pilot once you are confident in the agent’s accuracy.
Keep knowledge bases focused
Keep knowledge bases focused
Create separate knowledge bases for different topics (e.g., pricing, product features, support). Assign only the relevant knowledge base to each agent to reduce confusion and improve response accuracy.
Use descriptive agent names
Use descriptive agent names
Name agents by their function and channel, such as “Website Chat — Lead Qualifier” or “Phone — After Hours Support.” This makes the Agent Studio dashboard easy to scan when managing multiple agents.
Monitor and retrain regularly
Monitor and retrain regularly
Review the Conversation AI dashboard weekly. Look for conversations where the agent handed off to a human or gave low-confidence answers. Use these as training opportunities to improve the knowledge base.
Set clear escalation paths
Set clear escalation paths
Every agent should have a defined handoff action. Configure a Stop Bot action so contacts can reach a human when the agent cannot resolve their question.