Plurai is launching on Product Hunt today, introducing the first vibe-training platform to build real-time, tailored evals for your AI agents, with high accuracy, at a fraction of the cost.

I had the opportunity to collaborate with their team on this first launch after months in stealth modeI, no pressure, and wanted to share with you some insights on how we prepped it.

We ve been seeing a consistent pattern across agent systems:

GPT-5 works well as a judge on average cases
but breaks down on edge cases and policy boundaries.

That s exactly where reliability matters.

In our recent work, we took a different approach:

From the very first version, we ran it on itself. The MCP kept our code quality high from day one, which meant the agent always had clean, well-structured code to build on. The feedback loop was tight enough that we could ship feature-complete in a fraction of the time we'd expected.

Then we hit a wall. Performance wasn't where it needed to be, and it was baked into our choice of language (Python). There was no patching our way out of it.

Your favorite agentic terminal @Warp is now open source!

@zach_lloyd explains the rationale here.

• Repo

• Contributing guide

• Roadmap

34 days ago, VibeTalent had 0 builders.

Today: 103 active builders. 54 shipped projects. A marketplace that's actually moving.

No paid ads. No big network. Just one bet every dev marketplace rewards storytelling. We'd reward shipping.

The premise

Hello everyone,

I ve spent the last few weeks building Pika Review, a CLI tool designed to shorten the feedback loop for code audits.

While cloud-integrated review bots are great, they trigger after code is pushed. I wanted to shift that intelligence to the local terminal, acting as a gatekeeper for staged and unstaged changes before they ever reach a remote repository.

Key Capabilities

If you want the exact pitch deck that got us in let me know.

Otherwise, here's the overview:

For us, it started from something frustrating: creating content felt very annoying and time-consuming. We tried the classic way: scripting, memorizing, filming, editing. But none of it felt authentic. And honestly, it was eating time we needed to focus on other things.

At the same time, we kept reading the same advice everywhere:
"founders should build in public and create content consistently". Easy to say but harder to do in reality. So instead of forcing ourselves to create content from scratch, we tried something simple: recording our own calls and using those moments as content.

AI is incredibly good, I d even say almost perfect.

And for many people, that uniformity of perfect templates is starting to feel annoying.

Here's an MCP server that lets you talk to Claude Code from anywhere you can negotiate a WebRTC connection (or make a phone call):

https://github.com/pipecat-ai/pi...

Here's an MCP server that lets you talk to Claude Code from anywhere you can negotiate a WebRTC connection (or make a phone call):

https://github.com/pipecat-ai/pi...