Creating Agents from Chat

Turn successful chat sessions into reusable, automated agents.

Why Convert to Agent?

If you find yourself asking the same questions repeatedly in chat, convert to an agent:

Automate with triggers
Run on schedule
Standardize process
Share with team

How to Convert

Complete Chat Session

Have a successful conversation that achieves desired outcome

Click 'Create Agent'

Button appears at end of chat session

AI Generates Prompt

Cotool analyzes conversation and generates system prompt

Review & Customize

Edit the generated prompt as needed

Add Triggers

Set up automation (Jira, Slack, cron, etc.)

What Gets Captured

From your chat conversation:

Tools used
Investigation approach
Decision-making logic
Output format

Example

Chat conversation:

You: Investigate alert SEC-123
Agent: [Checks SentinelOne, Splunk, VirusTotal]
Agent: Determined to be false positive
You: Update the ticket
Agent: [Updates Jira]

Generated agent:

You are an alert triage agent.

When given a ticket ID:
1. Fetch alert details from SentinelOne
2. Search Splunk for related activity
3. Check indicators in VirusTotal
4. Determine severity
5. Update Jira ticket with findings

[Rest of generated prompt...]

Best Practices

Use chat to prototype complex workflows
Test the generated agent in Builder before deploying
Refine the auto-generated prompt
Add error handling and edge cases

Creating Agents

Learn more about agent creation

⌘I

Getting Started

Core Concepts

Chat

Agents

Triage

Hunt

Detections

Improving Agents

Settings

Integrations

Why Convert to Agent?

How to Convert

What Gets Captured

Example

Best Practices

Creating Agents

​Why Convert to Agent?

​How to Convert

​What Gets Captured

​Example

​Best Practices

Creating Agents

Why Convert to Agent?

How to Convert

What Gets Captured

Example

Best Practices