"Farm to table software": How I built a Thanksgiving party hub using Lovable

In today’s pre-Thanksgiving episode, I walk you through how I vibe coded my very own “Thanksgiving party hub” using Lovable—and how I transformed it from AI-generated slop into something warm, personal, and genuinely useful. I show you exactly how I upleveled the typography, visuals, and structure using Google Fonts and Midjourney style references, and then I share one of my favorite real-life AI hacks: how to turn any messy online recipe into a clean, step-by-step, kid-friendly version that’s actually usable while you’re cooking. This is a cozy, practical walkthrough of my real design process—the little tricks I use to make AI-built apps feel handcrafted instead of generic. *What you’ll learn:* 1. How to build a fully functional Thanksgiving party hub in Lovable—guests, dishes, recipes, and photos 2. How I uplevel AI-generated designs using Google Fonts and Tailwind 3. How to use Midjourney style references to create custom images that match your aesthetic 4. How to add custom features to vibe-coded apps, like dietary preferences and allergen tags 5. How to iterate on layouts inside Lovable using screenshots and small, targeted prompts 6. How I use ChatGPT to restructure recipes so the measurements are embedded directly in each step 7. How to make recipes kid-friendly and easier to follow using a simple formatting prompt *Brought to you by:* WorkOS—Make your app enterprise-ready today: https://workos.com?utm_source=lennys_howiai&utm_medium=podcast&utm_campaign=q22025 *Where to find Claire Vo:* ChatPRD: https://www.chatprd.ai/ Website: https://clairevo.com/ LinkedIn: https://www.linkedin.com/in/clairevo/ X: https://x.com/clairevo *In this episode, we cover:* (00:00) Introduction to the Thanksgiving party hub concept (02:20) Starting a project in Lovable and initial design assessment (04:59) Upleveling typography with Google Font combinations (08:36) Creating custom header images with Midjourney (11:39) Adjusting aspect ratios for Midjourney images (14:22) Fixing design issues incrementally (18:52) Adding dietary-restriction functionality (23:36) AI recipe reformatting for easier cooking (26:02) Thoughts on ChatGPT 5.1 (30:51) Final implementation and recipe sharing *Tools referenced:* • Lovable: https://lovable.dev/ • Midjourney: https://www.midjourney.com/ • Google Fonts: https://fonts.google.com/ • ChatGPT: https://chat.openai.com/ • Canva Font Combinations: https://www.canva.com/font-combinations/ *Other references:* • Polenta and Sausage Stuffing Recipe: https://www.epicurious.com/recipes/food/views/polenta-and-sausage-stuffing-233030 • Runaway Pancakes (kid-friendly recipe site): https://runawaypancakes.com/ _Production and marketing by https://penname.co/._ _For inquiries about sponsoring the podcast, email jordan@penname.co._

Claire Vohost

Nov 19, 202534mWatch on YouTube ↗

WHAT IT’S REALLY ABOUT

Building a Thanksgiving party hub with Lovable, fonts, images, recipes

Claire builds a Thanksgiving “party hub” in Lovable to manage invites, potluck dishes, recipe sharing, and a photo gallery, using it as a live demo of vibe-coding workflows.
She shows how small design upgrades—especially intentional typography via Google Font pairings—can dramatically improve the feel of an AI-generated app while also teaching you where the changes live in code (e.g., Tailwind config).
She replaces generic hero imagery with a custom Midjourney header using style references and aspect-ratio tuning, then iterates through incremental UI fixes (with candid examples where vibe coding fails to resolve styling issues).
On the functionality side, she adds dietary-restriction multi-selects and dish allergen tags, and shares a favorite ChatGPT prompt pattern to reformat recipes so measurements appear inline with each step—making cooking (especially with kids) easier.

IDEAS WORTH REMEMBERING

5 ideas

Start with a clear V1 prompt, then iterate in small design chunks.

Claire begins with a straightforward Lovable prompt (invites, dishes, recipes, photos) and improves the app step-by-step. She argues incremental, targeted prompts avoid the model “getting lost” compared to dumping a large list of issues at once.

Typography is the fastest lever to make vibe-coded apps feel intentional.

By using a curated Google Font combination (Homemade Apple for headlines, Raleway for body), the UI instantly feels more “cozy” and custom. She highlights that many vibe-coding tools already integrate smoothly with Google Fonts.

Use vibe coding as a way to learn real implementation details.

Claire points viewers to the actual file changes—Google Fonts import, Tailwind config updates, and default CSS—so the exercise teaches both design taste and how modern frontend stacks wire typography.

Midjourney “style references” plus correct aspect ratios beat generic hero images.

She finds a whimsical storybook-style reference string on X, reuses the prompt, and fixes the mismatch by removing the hard-coded aspect ratio and regenerating at 2:1 for a website header. This yields a more on-brand, better-fitting hero background.

Screenshots and specific bug descriptions help, but vibe-coding debugging is imperfect.

She tries screenshot-based feedback for clipped headline text and navbar spacing; some fixes don’t stick. The practical takeaway: expect to occasionally inspect and edit code yourself when the model doesn’t resolve styling reliably.

WORDS WORTH SAVING

5 quotes

One of the things that I think you can do to uplevel any design is get a really nice typography combination in your application.

— Claire Vo

These are just a couple really simple tips and tricks that are gonna get you from what feels like a little bit like vibe coding slop to something much more beautiful… ‘artisanally crafted, farm-to-table software.’

— Claire Vo

I think if you gave a big list of, ‘I don’t like the typography, and I don’t like the header…’ it can get lost… but this incremental step-by-step is really helpful.

— Claire Vo

This is just proof that sometimes vibe coding is not perfect.

— Claire Vo

Fully reformatted… with measurements embedded directly into each step so you never have to jump back to the ingredients.

— ChatGPT (as read/quoted by Claire Vo)

Lovable project prompt and V1 feature setIncremental design iteration vs. big-bang promptingGoogle Fonts pairing for headline/body typographyTailwind configuration and learning-by-reading generated codeMidjourney style references and parameter stringsAspect-ratio tuning for header/hero imagesDebugging and limitations of vibe-coded styling fixesDietary-restrictions data model and multi-select UXDish allergen/dietary tagging and coordination logicChatGPT recipe reformatting with measurements embedded in stepsThoughts on ChatGPT 5.1 behavior (personality, tool-calling, speed)Sharing/publishing and extending into product ideas (custom GPT, API calls)

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.