What’s New at Kilo

We are excited to launch Multi-Agent Systems, a powerful new way to build an AI workforce that works together. You can now go beyond single-purpose bots and create specialized AI team members that collaborate, share context, and coordinate tasks to solve complex problems.

What’s New:

• Collaborative Intelligence: Agents can now trigger one another's workflows and share data, allowing for sophisticated handoffs just like a human team.

• Specialized Roles: Instead of one giant agent doing everything, build distinct specialists—like "Leo" for lead qualification or "Athena" for 24/7 support—that handle their specific domains expertly.

• Shared Context: When one agent hands off a task to another, the full conversation history and data payload are preserved, ensuring a seamless experience.

How it works: Imagine a workflow where Leo qualifies a lead and creates a Salesforce record. If the lead asks a technical question, Leo automatically loops in Athena to resolve the issue, while Nexus simultaneously creates a task in Asana for your sales team.

// Example: Multi-Agent Handoff Logic
if (lead_score > 80) {
  // Leo passes the lead to the Sales Closer Agent
  await Agent.handoff({
    to: "agent_sales_closer",
    context: {
      lead_id: "001",
      qualification_summary: "High intent, budget approved",
      crm_record: salesforce_id
    }
  });
} else {
  // Leo passes the lead to the Nurture Agent
  await Agent.handoff({
    to: "agent_nurture",
    reason: "Budget constraints"
  });
}

While Kilo integrates with 500+ popular tools, we know you have custom internal tools. Today we are introducing generic Webhook support, allowing you to trigger any external system.

Features:

• Outgoing Webhooks: Configure your agent to send a JSON payload to any URL when a specific action occurs.

• Incoming Triggers: Trigger an agent to start a workflow via an external API call.

• Secure Headers: Add authentication tokens to your webhook requests for secure internal communication.

Configuration Example:

We’ve overhauled our Developer Tools category to help technical teams automate their sprint workflows. Kilo agents can now interact deeply with your code repositories and issue trackers.

New Capabilities:

• GitHub: Agents can now create issues from bug reports, comment on pull requests, and manage repository tasks.

• Jira & Linear: Automate ticket creation, status updates, and sprint tracking based on Slack messages or emails.

• VS Code Extension: (Beta) Manage development tasks and automate code reviews directly from your editor.

Smart Automation: You can now build an agent that listens to #engineering-support in Slack and automatically drafts a Jira ticket if the sentiment is "Urgent."

Why settle for one assistant when you can have a specialized team? We have rolled out Multi-Agent Systems, allowing specialized agents to collaborate and solve complex problems together.

The Highlights:

• Agent Handoffs: Create specialized roles (e.g., "Leo" for Lead Gen and "Athena" for Support) and let them pass tasks to one another.

• Shared Context: Agents now share conversation history and data payload, ensuring no information is lost when a task is transferred.

• Orchestration Layer: A new "Manager" node allows you to define which agent handles specific triggers.

Use Case Example:

A customer asks a technical question during sales qualification. "Leo" (Sales Agent) recognizes the intent and seamlessly loops in "Athena" (Support Agent) to answer, then takes back control to book the meeting.

We are thrilled to launch the Visual Workflow Editor, a new way to map out complex logic with a drag-and-drop interface. While our Natural Language Builder is perfect for instant prototyping, the Visual Editor gives you precision control for production-grade agents.

What’s New:

• Drag-and-Drop Interface: visually connect nodes, actions, and integrations on an infinite canvas.

• Hybrid Building: Start with natural language to build the skeleton, then switch to the Visual Editor to fine-tune the details.

• Live Debugging: Watch data flow through your agent in real-time to identify bottlenecks or logic errors.

How to use it: Simply click the "Open in Canvas" button on any existing agent.

// Example: Logic node configuration
{
  "node_type": "condition",
  "input": "lead_score",
  "logic": "greater_than",
  "value": 80,
  "true_path": "connect_salesforce",
  "false_path": "nurture_sequence"
}