Let them eat Pi

gm legends and welcome back to the Leaderboard. Today we're diving into Pi, the product, not the number, a context-aware coding assistant, and a Letterboxd style app for music. Let's dive in.

Pi is an ML & data science toolkit built for software engineers. Instead of piecing together different optimization tools, Pi brings 30+ AI techniques into one place—automating prompt tuning, search ranking, reinforcement learning, and more. It starts by building a custom scoring system based on your app’s requirements, then runs a suite of optimizations to improve performance.

🔥 Our take: Most AI tools give you raw power and expect you to figure out the fine-tuning. Pi actually builds a scoring system first, making optimization feel less like guesswork and more like engineering. And launching a product named Pi on Pi Day? That’s the kind of nerdy attention to detail we respect.

Zencoder is an AI coding assistant that actually understands your workflow. It runs inside VS Code and JetBrains, integrates with 20+ dev tools, and automates everything from refactoring to boilerplate so you can focus on real problem-solving instead of grunt work.

🔥 Our take: AI coding tools are great until they start hallucinating function names and rewriting working code. Zencoder promises deep context awareness so it doesn’t just generate code, but actually fits into how you build. If it can really cut through the noise and feel like a teammate instead of a fancy autocomplete, it might actually be worth keeping around.

Factory.fm is Letterboxd for music—a place to rate, review, and track everything you’re listening to. Discover new music through other obsessives, build your own profile, and even bring back the MySpace Top 8 energy by curating your favorite albums.

🔥 Our take: Music nerds finally have a proper home. Streaming services give you stats, but Factory.fm lets you actually talk about music—argue about rankings, obsess over deep cuts, and prove that you knew an artist before they blew up. If you still miss early internet music forums, this might be your new spot.

You can rack up a 100-day streak on Duolingo, but that doesn’t mean you’re ready to hold a conversation. Willem van den Eijkel, founder of Ellie, started a discussion about what’s missing from language-learning apps, and the general consensus was clear: they make you feel like you’re learning, but they don’t prepare you for how people actually talk.

Some takes from the discussion:

💬 Brad Harris: Most apps focus on memorization instead of real-world use. They teach you words and grammar rules, but when it comes to understanding native speakers in context, you’re on your own.

💬 Jodyl Gonsalves: There’s no real way to practice speaking. A lot of apps keep you locked into drills and exercises, but without actual conversations, it’s hard to build confidence.

💬 Ken Miller: Apps can tell you you’re at an intermediate level, but throw you into a real conversation, and it’s a different story. Native speakers talk fast, use slang, and break grammar rules—none of which these apps really prepare you for.

So is the real problem with the apps, or is there just no shortcut to fluency?