About Flect

Lex Fridman (3+ years ago) - "one of the things that sucks with podcasts is it's hard to find stuff..."

Hi, I'm Pat Nadolny, a software engineer with a love for podcasts and an itch to solve problems. Flect was born from frustration—a simple yet persistent problem I couldn't ignore. I love listening to podcasts, but finding specific moments in long-form episodes felt like an impossible task. I wanted to recall a clip I vaguely remembered—what the hosts were discussing, the ideas they shared—but no existing tools could deliver. It felt like the pre-Google days of the internet, but for podcasts.

I started digging. I tried keyword searches, AI-powered summarization tools, and everything in between. Nothing worked the way I wanted. I didn't just want summaries or titles—I wanted the exact moment. The moment that made me think, laugh, or want to share it with friends. And so, I built Flect.

What Is Flect?

Flect lets you describe a moment from a podcast, and it returns a list of relevant YouTube clips, each linked to the precise timestamp where that moment begins. Imagine searching for that Lex Fridman quote about what sucks with podcasts, or a guest's thought-provoking take from years ago. With Flect, you'll find it. So far, I've indexed 50+ podcasts (more added hourly, send an email with requests!), over 18k episodes, and more than 50k hours of content—all searchable by context, not just keywords.

Why It Matters

Podcasts are exploding in popularity, yet we're stuck with search tools that barely scratch the surface. With Flect, you're not limited to exact keywords or episode titles. You can explore ideas and discussions across podcasts in a way that's intuitive and effective. It's built for podcast lovers by a podcast lover—for people who deserve better than the tools that currently exist.

How It Works

Flect uses a lean and efficient tech stack that I've carefully pieced together:

  • YouTube Transcripts: A rich starting point for understanding podcast content.

  • AWS Lambda, dbt, and Athena: These power the data processing and keep things scalable.

  • OpenAI Embeddings and MyScale Vector DB: To understand and search content contextually.

  • Supabase: For authentication and backend management.

  • Next.js and FastAPI: The backbone of the user experience.

You can try a few suggested searches on the landing page or dive in with your own. Beyond search, Flect lets you save and share clips, and I'm actively working on an AI chat feature that will let you converse with podcasts, summarize content, and find hidden gems.

A One-Person Operation

This is a side project, a labor of love. Flect is built entirely by me, during nights and weekends, fueled by my passion for podcasts and problem-solving. As an indie hacker, I'm not a big company—just one person trying to make a tool that works and solves a real need. If you find value in Flect, know that your support means everything. Whether you share it, use it, or just let me know what you think, it's all deeply appreciated.

Get in Touch

Flect is evolving, and your feedback is invaluable. What works? What doesn't? What features would make it even better? Let me know! You can reach out directly, share your thoughts, or just say hi. Together, we can make finding moments in podcasts as easy as it should be.

support@flectai.com