We Built a Women-Centered GPT. It Flopped – and Taught Us Everything

We Built a Women-Centered GPT. It Flopped – and Taught Us Everything

From CARE Failing Forward by Emily Janoch

August 12, 2025 · 26 min · Episode 126

About this episode

This episode discusses the failure of a women-centered AI tool and the lessons learned from the experience.

What happens when you try to build an AI tool that works for women entrepreneurs – and it totally flops? In this episode of Failing Forward, CARE’s Koheun Lee and Sarah Hewitt share the story of their ambitious attempt to create a women-centered GPT trained on real-world data from women entrepreneurs. Spoiler: it didn’t go as planned. But the failure revealed a lot. In this episode, Koheun and Sarah discuss: Why CARE built a custom GPT to fight bias, and what went wrong How even well-trained AI tools can reinforce stereotypes and exclude women What we learned about prompt design, user behavior, and the limits of scrappy innovation Why most users still defaulted to mainstream AI tools Actionable tips for using AI more intentionally, and with less bias What this "failure" taught us about building better tools and better teams Tune in for a candid conversation about tech, bias, and what it really means to learn in public. To learn more and join the conversation, visit the Women’s Entrepreneurship LinkedIn Community of Practice.

People in this episode

Host: Emily Janoch

Guests: Koheun Lee, Sarah Hewitt

Topics covered

  • AI tools
  • women entrepreneurs
  • bias in technology
  • failure analysis
  • prompt design
  • user behavior

Keywords

  • women-centered GPT
  • AI bias
  • entrepreneurship
  • failure
  • innovation
  • technology

Mentioned in this episode

Organizations: CARE, Women’s Entrepreneurship LinkedIn Community of Practice

More episodes of CARE Failing Forward

Explore listener stats, chart rankings, contacts and more on the CARE Failing Forward podcast page.