Vibe Coding Live with Will Melton: Come Build With Me

By Will Melton · January 24, 2026 · 1:03:56 read

Category: Business & Economy

Show: Live Vibe Coding With Will Melton

In today’s vibe coding session we’re working through a real product build: a writing/workroom experience for authors that connects planning, collaboration, and AI-generated audio—without letting UX or costs spiral out of control. We start with quick project housekeeping (including Su Chef.ai notes a

In today’s vibe coding session we’re working through a real product build: a writing/workroom experience for authors that connects planning, collaboration, and AI-generated audio—without letting UX or costs spiral out of control. We start with quick project housekeeping (including Su Chef.ai notes and dark mode polish), then tackle a livestream UX snag: when Will goes live, certain social icons disappear (Twitch shows up, but YouTube and LinkedIn don’t), so we call out the missing behavior and what should be consistent across platforms.

Chapters

From there we zoom out into the broader build approach with Lovable. The conversation is practical: what it’s like shipping tools with a React + TypeScript stack, what’s been stable, where the pain shows up, and how you keep momentum when you’re iterating sprint by sprint. A big theme is roadmap hygiene—reducing repetitive admin and using the sprint dashboard in a way that keeps work moving even when you’re not manually checking every single item. We also lean on AI chat to tee up the next sprint and keep the plan from getting derailed by one-off requests.

Midstream we shift into the core product: a book project with a large manuscript that needs re-organization and de-duplication after early AI-assisted assembly. We walk through why this tool exists, how stakeholders (author, coach, collaborators) can use it, and where features like a pitch wizard fit in—especially when balancing self-publishing one book while preparing a clearer second book and exploring publisher interest.

The second half becomes a deep architecture + business discussion about an audiobook feature. The goal is simple to say and hard to ship: convert chapters to voice, let users listen and edit efficiently, and keep it sustainable. We talk about saving generated audio files so you don’t regenerate the same content, chunking audio at the paragraph level, tracking whether text changed since audio was generated, and enabling progressive playback so users can start listening sooner. We also discuss UI patterns like a long seek bar for scanning, indicators for what’s already generated, and click-to-play at the paragraph level.

Then we look at the money side. Credits can burn fast (thousands per chapter), and large plans still require real user volume to justify. The conclusion is clear: “Play” should only play existing audio, and “Convert to voice” should be a deliberate action—chapter-level or paragraph-level—so users stay in control and understand when they’re spending.

This session is especially relevant if you’re building in public, shipping with vibe coding, and thinking about AI SEO and AI search optimization—because sustainable AI features are as much product design as they are engineering.

What you’ll learn / takeaways

* Shipping with Lovable: constraints, stability, and iteration cadence

* Roadmap + sprint flow: how to reduce repetitive status-checking

* Audio generation UX: progressive playback, seekability, paragraph targeting

* Cost controls: caching, chunking, and change detection to prevent rework

* Monetization guardrails: credits, premium features, quality vs faster models

* Scaling reality: persistent audio increases storage and operational overhead

What changed today

* Updated the plan toward paragraph-level audio chunks (not full-chapter re-renders)

* Clarified “Play” vs “Convert to voice” as separate actions

* Flagged credit burn + storage costs as gating constraints for launch

* Positioned audiobook/voice as a likely premium feature with explicit user control

If you’re new here, everything we’re building lives at [https://studio.willmelton.com](https://studio.willmelton.com). The work is made possible by @xponent21, and the learning community is @discoveraio (helping people build visibility inside large language models). Want to get involved, become a referral partner, or follow the projects as they ship? Start there—and subscribe on the Will Melton Vibes YouTube channel for more live sessions.

Live schedule (Eastern Time)

Saturday at 1

Sunday at 10

Monday at

Links

https://studio.willmelton.com

Easy ways to support me:

Subscribe to my YouTube Channel: @willmeltonvibes

Use my Eleven Labs Affiliate Link for AI Voice Generation: https://try.elevenlabs.io/z3ni0qf4lydc