Live Vibe Coding: Building a Production-Ready Hermes Agent v0.14.0 on Lovable
By Will Melton · May 16, 2026 · 1:18:51 read
Category: Business & Economy
Show: Live Vibe Coding With Will Melton
Today we're going live to build a Hermes Agent on Lovable — and this one has real teeth. Hermes is an open-source autonomous agent that lives on your server, remembers what it learns, and gets more ca
Today we're going live to build a Hermes Agent on Lovable — and this one has real teeth. Hermes is an open-source autonomous agent that lives on your server, remembers what it learns, and gets more capable the longer it runs. We're wiring it directly into our existing AI automation stack to handle one of the most repetitive bottlenecks in our agency workflow: turning meeting transcripts into structured build-ready work plans and dispatching them through our product pipeline without touching them manually.
This is part of my ongoing 52 Apps in 52 Weeks challenge — live, unscripted, real builds that ship. No slides. No demos. Just a builder working through problems in real time on Lovable.
Here's what we're solving today. Every meeting we run gets transcribed. That transcript has to be processed through Claude with project context, converted into a work plan, entered into Lovable's agent, assembled in our product roadmap, executed, and reviewed. Right now there are at least five manual handoff points in that chain. Today we start collapsing them into a single automated pipeline with Hermes as the routing brain.
We're also getting into how Hermes fits inside Spaceport — @Xponent21's internal operating system built on Lovable and Supabase — and how tools like Recall.ai solve the hard problem of capturing audio across Zoom, Google Meet, and Microsoft Teams without building your own notetaker from scratch. The play is a white-label meeting intelligence layer we're calling Scout: bot joins your meeting, Recall handles the audio infrastructure, Whisper transcribes, Claude processes, Hermes dispatches. Full pipeline, marginal API costs, no subscription notetaker required.
We'll also look at how Lovable is expanding its MCP connector ecosystem — Granola, Notion, Figma, Linear, Miro, n8n, Sentry, Sanity, HeyGen, and more are all available now as chat connectors — and what that means for building agentic workflows directly inside your Lovable projects without leaving the platform.
If you're building with AI, running an agency, or just trying to eliminate the manual work between your meetings and your actual product output, this stream is for you.
Get 10 free Lovable credits here: https://lovable.dev/invite/S8XSI99
Subscribe and hit the bell — we stream Saturdays at 1 PM EST, Sundays at 10 AM EST.
Learn more at https://studio.willmelton.com | https://xponent21.com
---
LINKS
Lovable (10 free credits): https://lovable.dev/invite/S8XSI99
Studio: https://studio.willmelton.com
Xponent21: https://xponent21.com
Channel RVA: https://channelrva.com