Saving 10–20 Hours Per Product Discovery Project with AI

After shifting to a sales-led growth model, customer calls became a key input for product research, but reviewing them was slow and resource-intensive. I helped build a GPT-based AI assistant to extract insights faster.

My responsibilities: gather datasets, define GPT instructions, troubleshoot, and demonstrate AI capabilities to teams.

The Goals:

  • Reduce the time spent analyzing customer and sales calls.

  • Enable teams to uncover insights that inform roadmaps, feature prioritization, and business strategy faster.

  • Build an internal proof-of-concept to justify further investment in AI.

The Approach:

  • Create a GPT-powered assistant that parses call transcripts from HubSpot

  • Enable team to get answers to customer questions through natural-language queries.

  • Present live demo at company-wide Hackathon.

The Results:

  • Saved 10–30 hours per product discovery project depending on scope.

  • Enabled more frequent and parallel research initiatives.

  • Adopted by marketing for campaign planning based on real customer insights.

  • Convinced leadership to invest in engineer resources to automate transcript collection from our CRM.

Integration use cases are essential to align sales, product, and developer teams on the solutions customers are looking for.

With a simple query, marketers, product managers, and even new employees can quickly understand the challenges customers face.