Cowork

Open Cowork settings → MCP Servers → Add.

Hosted (recommended)

{
  "mcpServers": {
    "talonic": {
      "url": "https://mcp.talonic.com/mcp",
      "headers": {
        "Authorization": "Bearer tlnc_your_key_here"
      }
    }
  }
}

Local (npx)

{
  "mcpServers": {
    "talonic": {
      "command": "npx",
      "args": ["-y", "@talonic/mcp@latest"],
      "env": {
        "TALONIC_API_KEY": "tlnc_your_key_here"
      }
    }
  }
}

Cowork is a chat-style MCP client that supports drag-and-drop file uploads and resource browsing. After adding the Talonic config, the eight tools appear in the tool picker and the talonic://schemas and talonic://webhooks/reference resources are browseable in the resources panel.

Like Claude Desktop, Cowork renders MCP resources in its UI. This means your saved schemas are visible without making a tool call, which helps when you want to browse available schemas before starting an extraction workflow.

After adding the config, restart Cowork to load the new MCP server. If you are using the hosted option, Cowork connects directly via HTTPS — no local process is spawned. For the local option, Cowork launches the npx command as a child process and communicates over stdio.

Example: browse schemas then extract

// Step 1: User browses talonic://schemas in the resources panel
// → Sees "Standard Invoice" (SCH-A1B2C3D4) with 6 fields

// Step 2: User drops invoice.pdf into the chat
// Agent calls talonic_extract:
{
  "file_data": "JVBERi0xLjQKJ...",
  "filename": "invoice.pdf",
  "schema_id": "SCH-A1B2C3D4"
}

// Step 3: Agent presents extracted data:
{
  "data": {
    "vendor_name": "Meridian Energy AG",
    "total_amount": 1500.00,
    "due_date": "2026-06-15"
  },
  "confidence": { "overall": 0.97 }
}

Cowork's resource browsing creates a natural workflow for Talonic users. Instead of asking the agent to list schemas via a tool call, you can visually browse the talonic://schemas resource in the sidebar, identify the right schema, and then ask the agent to extract using that schema by name or ID. This visual discovery step reduces back-and-forth in the conversation and helps users who are not familiar with their workspace's schema inventory.

For teams onboarding new members, Cowork's visual schema browser is especially useful. A new team member can see all available extraction templates at a glance, understand what fields each schema extracts, and start processing documents immediately without needing to learn schema IDs or ask colleagues which templates exist. Combined with drag-and-drop file uploads, the barrier to entry is as low as it gets.