Configure an agent
Define the role, instructions, tools, data boundaries, and workflow responsibilities that make a Studio agent production-ready.
Configure the agent's instructions, model/provider, temperature, reasoning limit, knowledge, tools, playbooks, deployment, and chat behavior here.
How an agent is organized.
Read Overview, Settings, Flows, Conversations, Analytics, and Channels as one agent system: identity, behavior, orchestration, evidence, and publishing.
Verify the agent role, configuration entry point, and whether it is ready to connect to tools or workflows.
Verify the agent role, configuration entry point, and whether it is ready to connect to tools or workflows.
What you will build
A production-ready agent configuration for a narrow operating role: instructions, model/provider choice, allowed tools, workflow responsibility, connected context, and review path.
Use this as the companion step after creating the agent shell. Creation gives the agent an identity; configuration defines how it behaves before it touches channels or customer records.
Write behavior as operating policy
Instructions should cover goal, tone, qualification or resolution criteria, escalation rules, allowed actions, required context, output format, and what the agent must never do.
For an AI SDR, include the qualification rubric, the required questions, the handoff triggers, and the fields the downstream workflow expects.
Connect the minimum context
Attach only the knowledge bases, tools, playbooks, tables, and CRM context needed for this agent's job. More context is not automatically safer; it can make behavior harder to test.
Use Tables for operating rules such as qualification thresholds. Use CRM for customer memory. Use Channels only after instructions and workflow behavior have been reviewed.
Common mistakes
Do not let a broad agent handle every conversation. Start with one role, one workflow, and one channel.
Do not publish after editing instructions without testing Preview, Conversations, Logs, and the handoff path.
Do not give the agent tools or data sources it does not need for the operating job.
Expected result
The agent has an inspectable role, safe instructions, minimum useful context, a defined escalation path, and a workflow-ready output format.
Once the role and tools are clear, connect the agent to workflows. Workflow Builder makes the trigger, AI step, routing, communication, and handoff path visible to the team.
Prerequisites
Start with an existing agent draft or a safe demo agent. Do not edit a live customer-facing agent unless you have a rollback plan.
Prepare the operating rules before opening Settings: role, boundaries, qualification or resolution criteria, required output, and escalation conditions.
Step-by-step configuration
1. Open the agent from Agent Builder.
2. Open Settings and review the current instructions, model/provider, temperature, knowledge, tools, playbooks, deployment, and chat controls.
3. Write instructions as operating policy: role, goal, tone, required context, allowed actions, escalation rules, and output format.
4. Attach only the knowledge bases, tables, tools, and playbooks the agent needs.
5. Keep Deployment offline or draft while testing.
6. Test realistic happy-path, missing-context, and escalation scenarios before connecting channels or workflows.
What to configure in each Settings area
AI Agent: write the operating instructions, choose the model, set temperature, and limit reasoning iterations so the behavior is predictable.
Knowledge Bases: attach only the approved sources this role needs. If the agent does not need a document or policy to answer, do not attach it.
Tools: enable system tools, custom tools, and integrations only when the agent needs to take that action. A sales lookup agent may need CRM search; a support reply agent may need ticket context; a drafting agent may not need write access.
Deployment: keep the agent Offline or in a safe draft state while testing. Move to Live only after preview responses, workflow behavior, and channel exposure are reviewed.
Chat Customization: set the first message, placeholder, nickname, launcher image, and preview so the external-facing experience matches the agent's job.
Developer note
If the agent needs to call an external service, attach a bounded custom API-call tool instead of asking instructions to perform unsupported actions. The Public API supports API_CALL tools with GET, POST, PUT, PATCH, and DELETE methods, headers, query params, body, and typed arguments.
Do not publish live API keys in examples. Use placeholders such as {{apiKey}}, and do not describe public creation of WhatsApp template tools because that surface is not exposed for public tool creation.
Transcript
Open searchable transcript
FAQs
What makes a Studio agent production-ready?
A production-ready agent has a clear role, explicit instructions, limited tool access, data boundaries, and defined workflow responsibilities.
Should every agent have every integration?
No. Connect the minimum tools required for the agent's operational job.
What is a Frontline agent?
A Frontline agent is an operational AI teammate with identity, behavior, prompts, connected context, channels, flows, conversations, analytics, and permissions.
When should I create a new agent instead of reusing an existing one?
Create a new agent when the operational role, tool access, escalation rules, or expected output is meaningfully different. Reuse an agent when the same job simply needs another workflow entry point.
How should prompts connect to workflows?
Prompts should describe the agent's job and output in a way the next workflow node can use. If a workflow branches on the result, ask the agent for structured values or a clear decision.
What memory should an agent have?
Give the agent the minimum useful memory: approved resources, relevant CRM context, table data, and current workflow state. Too much memory makes behavior harder to test and audit.
How do I test an agent before production?
Test the agent with realistic conversations, missing context, edge cases, escalation scenarios, and expected workflow outputs. Review tone, accuracy, tool use, and handoff behavior.
What permissions should an agent receive?
Give agents only the integrations and actions required for their role. If the agent only drafts or classifies, it may not need write access to external systems.
How do agents connect to channels?
Channels define where an agent can communicate. Use WhatsApp, Instagram, Messenger, or other channels with explicit routing, approved templates, and handoff rules.
How should teams monitor an agent after launch?
Review conversations, analytics, workflow logs, escalation quality, and Max Activity. Monitoring should show both AI quality and the operational outcome the agent supports.