Back to sessions
AI For Sales

Claude Code for Cold Email: Skills, List Building, and Autonomous Campaign Research

Eric Nowoslawski

YouTube creator

Watch session
Duration
12 min

This session walks through four production use cases of Claude Code inside a cold email agency sending eight million emails per month. You will see how Skills files replace repetitive SOP training, how voice commands drive API-connected list builds through Prospeo, how Sonnet sub-agents perform ICP classification at no extra token cost, and how an autonomous research loop continuously rewrites copy and adjusts list filters based on live campaign results. The session is grounded in real agency workflows and requires no prior coding experience to follow or replicate.

Learning objectives

  • Understand how Claude Code differs from browser-based chatbots by accessing local files and making live API calls
  • Set up and use Skills files to encode repeatable lead generation SOPs that persist across sessions
  • Build and enrich a prospect list using voice commands and a Prospeo API skill without manually configuring filters
  • Use Sonnet sub-agents inside Claude Code to classify ICP fit and write personalized lines without incurring separate API costs
  • Configure an auto-research repository to run autonomous campaign experiments and learn from reply-rate outcomes over time

Key takeaways

  • Claude Code breaks the chatbot sandbox by connecting to local files, databases, and external APIs directly
  • Skills files let teams share and reuse complex workflows without retraining the model each session
  • Voice-driven list building via WhisperFlow and a Prospeo skill can produce an enriched, filtered CSV in minutes
  • Sonnet sub-agents handle ICP filtering and personalization inside the Claude Code plan, eliminating extra OpenAI API spend
  • The auto-research loop pulls campaign results, reads a context file, adjusts copy and list filters, and relaunches sends completely autonomously
  • Dedicating even ten percent of send volume to auto-research experiments can produce results that outperform human-written campaigns
  • No coding background is required to install or operate Claude Code for outbound lead generation workflows

Free plan, no credit card. Watching opens this session in the Triple Session app.

How Triple Session works

One coaching loop, running every week

Coaching fails when it is an event. Triple Session turns it into a loop: measure the gap, train against it, and check whether the next call moved.

  1. 01

    Identify the gap

    Every call is recorded and scored against your own playbook, so the distance between what your team says and what the playbook asks for stops being a guess.

  2. 02

    Surface the insights

    Patterns roll up across reps, deals, and objections. You see which behavior is costing pipeline, not just which rep is behind.

  3. 03

    Train the people

    Training is assigned against that specific gap: short, expert-led sessions tied to the behavior you just measured.

  4. 04

    Deliver the feedback

    Managers coach from evidence instead of memory. A scorecard, the moment in the transcript, and the one thing to practice next.

Your team's next call is already on the calendar.

Start with the free plan and run the loop on your own playbook. No credit card, no procurement conversation.