Chanl

VAPI

Give your VAPI voice agents tools, knowledge, and memory — via a single MCP connection

VAPI Integration

Your VAPI agent handles the conversation. Chanl handles everything behind it — knowledge base search, persistent memory, custom tool execution, and quality scoring. One MCP connection, and your VAPI agent goes from a voice interface to a fully-equipped backend.

What Your VAPI Agent Gets

Knowledge Base

Your agent searches your docs, FAQs, and policies during calls — grounded answers, no hallucinations

Persistent Memory

Remembers customers across every conversation — preferences, history, past commitments

Custom Tools

Calls your APIs during live calls — CRM lookups, order checks, ticket creation

Plus automatic call transcription, scorecard evaluation, and analytics on every conversation.

Prerequisites

  • A VAPI account with at least one assistant
  • A Chanl workspace — sign up free
  • A Chanl API key (Settings → API Keys, with tools:read and tools:execute scopes)

Setup

Your workspace's MCP endpoint follows this pattern:

https://{workspace-slug}.chanl.dev/{toolset-slug}

For example, if your workspace slug is acme and you want the built-in admin tools:

https://acme.chanl.dev/admin-tools

Find your workspace slug in Settings → Workspace in the Chanl dashboard.

In your VAPI dashboard, go to your assistant's configuration and add an MCP server:

{
  "type": "streamablehttp",
  "url": "https://acme.chanl.dev/admin-tools",
  "headers": {
    "X-API-Key": "chanl_key_abc123..."
  }
}

Replace acme with your workspace slug and the API key with your actual key.

VAPI supports MCP natively — no custom integration code needed. Your VAPI assistant automatically discovers all tools available in the Chanl toolset.

Run a test call in VAPI. Ask something that requires knowledge base search or a tool call:

"What's your refund policy?"

If the knowledge base is configured, your agent will call kb_search, retrieve the relevant policy text, and answer accurately.

Check the call transcript in VAPI — you'll see Chanl tool calls logged alongside the conversation.

Using Custom Toolsets

The admin-tools toolset includes system tools (knowledge search, memory, workspace management). For production, create a focused toolset with only the tools your VAPI agent needs:

# Create a support-focused toolset
chanl toolsets create --name "Support" --slug "support"

# Add specific tools
chanl toolsets add-tool --toolset support --tool lookup_order
chanl toolsets add-tool --toolset support --tool create_ticket

Then point your VAPI assistant to the custom endpoint:

https://acme.chanl.dev/support

Your agent gets lookup_order and create_ticket — nothing else. Clean, scoped, production-ready.

What Happens During a Call

I need to check my order status memory_search(customerId) Past context: "Premium customer, 2 previous calls" lookup_order(orderId: "12345") {status: "shipped", tracking: "1Z999..."} Your order shipped yesterday, tracking number is 1Z999... memory_add("Asked about order 12345, shipped") Caller VAPI Agent Chanl MCP

Every tool call is logged. Every conversation is transcribed and scored. You get full visibility in the Chanl dashboard without any additional setup.

Monitoring VAPI Calls

Once connected, VAPI calls flow into Chanl's monitoring automatically:

  • Transcripts — full conversation text with speaker turns and timestamps
  • Scorecards — auto-grade every call against your quality criteria
  • Analytics — call volume, sentiment, resolution rates across your VAPI agents
  • Alerts — get notified when an agent goes off-script or sentiment drops

View everything in Observe → Analytics in the Chanl dashboard.

Troubleshooting

What's Next

Knowledge Base

Upload docs for your agent to search during VAPI calls

Scenarios

Test your VAPI agent with AI-generated callers before going live

Scorecards

Auto-grade every VAPI call against your quality standards

Tools

Create custom API tools for your VAPI agent

On this page