AI Bdr

AI BDR Skill

ElevenLabs Conversational AI integration for AI-powered outbound cold calling.

What I Do

• Make outbound cold calls with AI BDR agents
• Retrieve call transcripts and audio
• Update agent prompts
• Support two call types: Platform (general Shakudo) and Copilot (Azure pro-serve)

When to Use Me

Use this skill when:

• Making AI-powered cold calls to prospects
• Retrieving call transcripts for analysis
• Updating BDR agent prompts
• Testing outbound calling workflows

Quick Start

# Load credentials
source /root/gitrepos/business-automation/agent-skills/ai-bdr/.env

# Make a Platform call (general Shakudo value prop)
curl -X POST "https://api.elevenlabs.io/v1/convai/twilio/outbound-call" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": "agent_8701kg3zx1taen79g245ft4486cy",
    "agent_phone_number_id": "phnum_6901kg3mt255ea1rdvpm8carn48q",
    "to_number": "+1XXXXXXXXXX"
  }'

Available Agents

Agent Comparison

Shared Resources

Part 1: Making Calls

Make a Platform Call (General Shakudo)

Use this for prospects interested in self-hosted AI platforms, data sovereignty, or MLOps.

source /root/gitrepos/business-automation/agent-skills/ai-bdr/.env

curl -X POST "https://api.elevenlabs.io/v1/convai/twilio/outbound-call" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": "agent_8701kg3zx1taen79g245ft4486cy",
    "agent_phone_number_id": "phnum_6901kg3mt255ea1rdvpm8carn48q",
    "to_number": "+1XXXXXXXXXX"
  }'

Key Value Props:

• Data never leaves their environment (VPC, on-prem, air-gapped)
• 170+ pre-integrated tools (MLflow, Kubeflow, Spark, Ray, etc.)
• Production-ready in weeks, not months
• SOC 2 Type II, HIPAA-compliant

Make a Copilot/Pro-Serve Call (Azure Focus)

Use this for prospects struggling with Microsoft Copilot or Foundry implementations.

source /root/gitrepos/business-automation/agent-skills/ai-bdr/.env

curl -X POST "https://api.elevenlabs.io/v1/convai/twilio/outbound-call" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": "agent_1301k3vpw7bxf0b8cqd0bepbzs48",
    "agent_phone_number_id": "phnum_6901kg3mt255ea1rdvpm8carn48q",
    "to_number": "+1XXXXXXXXXX"
  }'

Key Value Props:

• Embedded engineers for Copilot customization
• Azure Foundry implementation expertise
• Pro-serve model (hands-on help vs. platform)

Response Format

{
  "conversation_id": "conv_xxxxxxxxxxxxxxxx",
  "agent_id": "agent_xxxxxxxxxxxxxxxx",
  "status": "initiated"
}

Save the conversation_id to retrieve the transcript later.

Part 2: Getting Transcripts

Get Conversation Transcript

CONVERSATION_ID="conv_xxxxxxxxxxxxxxxx"

curl -X GET "https://api.elevenlabs.io/v1/convai/conversations/${CONVERSATION_ID}" \
  -H "xi-api-key: $ELEVENLABS_API_KEY"

Response includes:

• transcript - Full conversation with speaker labels
• duration - Call length in seconds
• status - completed, in_progress, failed
• metadata - Agent ID, phone numbers, timestamps

Download Call Audio

curl -X GET "https://api.elevenlabs.io/v1/convai/conversations/${CONVERSATION_ID}/audio" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  --output "call_${CONVERSATION_ID}.mp3"

List Recent Conversations

curl -X GET "https://api.elevenlabs.io/v1/convai/conversations?agent_id=agent_8701kg3zx1taen79g245ft4486cy&page_size=20" \
  -H "xi-api-key: $ELEVENLABS_API_KEY"

Part 3: Updating Agent Prompts

Update Platform Agent Prompt

curl -X PATCH "https://api.elevenlabs.io/v1/convai/agents/agent_8701kg3zx1taen79g245ft4486cy" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "conversation_config": {
      "agent": {
        "prompt": {
          "prompt": "YOUR_NEW_PROMPT_HERE",
          "temperature": 0.8
        }
      }
    }
  }'

Update Copilot Agent Prompt

curl -X PATCH "https://api.elevenlabs.io/v1/convai/agents/agent_1301k3vpw7bxf0b8cqd0bepbzs48" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "conversation_config": {
      "agent": {
        "prompt": {
          "prompt": "YOUR_NEW_PROMPT_HERE",
          "temperature": 0.8
        }
      }
    }
  }'

Update First Message

curl -X PATCH "https://api.elevenlabs.io/v1/convai/agents/agent_8701kg3zx1taen79g245ft4486cy" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "conversation_config": {
      "agent": {
        "first_message": "Hi, this is Kaji, a Shakudo-powered AI BDR. Do you have 30 seconds?"
      }
    }
  }'

Part 4: Agent Prompts Reference

Kaji-Platform Prompt (General Shakudo)

The Platform agent focuses on:

• Data Sovereignty - VPC, on-prem, air-gapped deployments
• 170+ Tools - MLflow, Kubeflow, Spark, Ray, vector DBs, LLM orchestration
• Fast Deployment - Weeks not months to production
• Security - SOC 2 Type II, HIPAA, RBAC, OWASP Top 10 LLM

Success Stories:

• CentralReach (Healthcare) - Months to weeks deployment, CR NoteGuardAI
• Fortune 500 Bank - Multi-tenant AI platform with compliance
• QuadReal (Real Estate) - Flexibility, no vendor lock-in

Key Opening:

"Appreciate it. Quick question - is your team trying to get AI into production right now?"

Full prompt: See /root/gitrepos/ai-bdr-agent/kaji-platform/prompt.md

Kaji-Copilot Prompt (Azure Pro-Serve)

The Copilot agent focuses on:

• Embedded Engineers - Hands-on implementation help
• Azure Expertise - Copilot customization, Foundry integration
• Pro-Serve Model - Services not just platform

Key Opening:

"Appreciate it. What's your team working on these days on the AI side?"

Full prompt: See /root/gitrepos/ai-bdr-agent/kaji-copilot/prompt.md

Part 5: Call Best Practices

When to Use Each Agent

Call Flow Guidelines

1. Opening - Be transparent about being an AI
2. Discovery - Ask ONE question at a time, wait for response
3. Dig Deeper - Don't accept surface answers, explore pain
4. Connect to Value - Only pitch AFTER understanding pain
5. Close - Always try to book a meeting, email is last resort

Banned Phrases

Never program agents to say:

• "Would you be interested in discussing..."
• "That's great to hear" (when they mention a problem)
• "Is there anything else I can help you with?"
• Any corporate/formal language
• Long monologues listing all features

Error Handling

Related Resources

• Platform prompt: /root/gitrepos/ai-bdr-agent/kaji-platform/prompt.md
• Copilot prompt: /root/gitrepos/ai-bdr-agent/kaji-copilot/prompt.md
• ElevenLabs Console: https://elevenlabs.io/app/conversational-ai
• ElevenLabs API Docs: https://elevenlabs.io/docs/conversational-ai/api-reference