Voice Message Recording

This drill produces a batch of personalized voice messages -- one per prospect -- ready for delivery via ringless voicemail or LinkedIn voice note. It covers both manual recording (Smoke level) and AI-generated voice cloning (Baseline+).

Input

• Prospect list in Clay or Attio with enrichment data (name, company, role, trigger signal)
• ICP definition with pain points and value propositions
• For AI generation: founder voice sample (1-3 minute audio clip)
• For manual recording: phone or laptop with microphone

Steps

1. Prepare prospect-specific scripts

Open your Clay table with the target prospect batch. Each prospect needs these columns populated:

• First name
• Company name
• Trigger signal (e.g., "just raised Series A", "hired 3 engineers this month", "switched to {competitor}")
• Value prop hook (one sentence connecting their signal to your solution)

If columns are missing, use the clay-enrichment-waterfall fundamental to fill gaps. Add a Clay AI formula column that generates a complete voice script per prospect:

Template:
"Hey {first_name}, this is {founder_name} from {company}. I noticed {company_name} {signal}. {value_prop_hook}. Would love to grab 15 minutes -- I'll send my calendar link in a follow-up message. Talk soon."

Target length: 50-80 words (20-35 seconds of audio).

2a. Manual recording (Smoke level)

Record each message yourself using your phone's voice memo app, Loom audio-only, or your laptop microphone.

Workflow per prospect:

1. Open the prospect's script in Clay
2. Read it aloud naturally -- do NOT sound like you're reading
3. If you stumble, re-record immediately (faster than editing)
4. Save as MP3 or M4A
5. Name the file: vm-{company-slug}-{firstname}.mp3

Target pace: 2-3 minutes per message including setup. A 20-prospect batch takes ~50 minutes.

Tips for natural delivery:

• Smile while recording -- it changes your tone
• Stand up -- it adds energy to your voice
• Pause briefly after the prospect's name -- it sounds more personal
• Vary your intonation -- monotone = instant delete

2b. AI voice generation (Baseline+ level)

Use the ai-voice-generation fundamental:

1. One-time setup: Upload your founder voice sample to ElevenLabs (or Play.ht / Resemble.AI). Store the returned voice_id.

2. Batch generation: Build an n8n workflow or script that:

   • Pulls prospect scripts from Clay (the AI formula column from Step 1)
   • Calls the ElevenLabs TTS API for each script with your cloned voice
   • Saves each audio file with the naming convention vm-{company-slug}-{firstname}.mp3
   • Uploads the audio URL back to a Clay column for delivery mapping

3. Quality check: Listen to a random sample of 5 messages from each batch. Check for:

   • Pronunciation of company names and prospect names
   • Natural intonation (not robotic or flat)
   • Correct pacing (not too fast, not too slow)
   • No audio artifacts or glitches

   If names are mispronounced, add phonetic spelling to the script: "Hey Sarah at Ack-me Corp" instead of "Hey Sarah at Acme Corp" (if the AI mispronounces it).

3. Build the delivery mapping

Create a mapping table (in Clay or CSV) connecting each prospect to their audio file:

| Prospect Email | First Name | Company | Phone | LinkedIn URL | Audio File URL | Audio Duration | |----------------|------------|---------|-------|-------------|----------------|----------------|

This table feeds into the voice-message-delivery drill for both ringless voicemail and LinkedIn voice note channels.

4. Log in CRM

Using the attio-contacts fundamental, update each prospect record in Attio with:

• voice_message_generated: true
• voice_message_url: {audio_file_url}
• voice_message_campaign: {campaign_slug}
• voice_message_script: {script_text}

This creates an audit trail and prevents duplicate messages to the same prospect.

Output

• One personalized audio file per prospect (MP3, 20-35 seconds)
• Delivery mapping table linking prospects to their audio files
• CRM records updated with voice message metadata

Triggers

Run for each new prospect batch entering the voice message play. At Smoke: 1 batch of 20-40 prospects. At Baseline: 2-3 batches per week of 25-50. At Scalable: daily batches of 50+ via automated pipeline.