Get Started

6sense

At 6sense, we’re so much more than an AI SDR team. 6sense AI Email Agents automate routine tasks and follow-ups, freeing up sales to focus on more strategic activities and giving marketing more control over the pipeline creation process.

Role I played:

Team Assembly & Vision

  • Built a cross-functional “AI SDR” squad with:
    • 2 NLP engineers (email generation & intent analysis)
    • 1 data scientist (lead scoring & response prediction)
    • 1 sales ops expert (CRM workflows)
    • 1 email deliverability specialist (inbox placement)
    • 2 full-stack engineers (API integrations)

Key Decision: Hired a former top-performing SDR to train our AI on high-converting email tactics

Alignment with GTM Teams

  • Worked closely with sales + marketing leadership to:
    • Identify 5 core email workflows to automate (e.g., cold outreach, meeting follow-ups)
    • Define handoff rules for when humans should take over
    • Establish shared KPIs (reply rates, meetings booked, pipeline generated)

Core Technical Execution

AI Email Engine Architecture

Key Components I Oversaw:

  • Intent & Context Modeling
    • Built a lead scoring model using:
      • Firmographic data (technographics, funding)
      • Engagement signals (website visits, content downloads)
    • Trained a BERT-based classifier to predict:
      • Best email type (e.g., cold vs. nurture)
      • Optimal send time
  • Email Generation
    • Developed a hybrid template + LLM system:
      • Templates: 200+ marketing-approved email skeletons
      • LLM Fine-Tuning: GPT-3.5 trained on:
        • Top-performing SDR emails
          • Brand voice guidelines
          • Prospect’s LinkedIn/website context
          • Added dynamic placeholders (e.g., “I saw your post about [topic]…”)
        • Reply Handling
          • Implemented a 3-tier response system:
            • Tier 1: Auto-reply to common queries (e.g., “Send me a demo”)
            • Tier 2: Flag for human SDR (e.g., pricing questions)
            • Tier 3: Escalate to AE (e.g., “We’re already using a competitor”)
        • Continuous Learning
          • Created a feedback loop where:
            • Human SDRs could rate AI suggestions
            • Email performance (opens/replies) retrained models nightly