Podcast Summary: “Farm-to-table software” — How I built a Thanksgiving party hub using Lovable

Podcast: How I AI
Host: Claire Vo
Episode Air Date: November 19, 2025
Episode Length: ~34 minutes

Episode Overview

In this special pre-Thanksgiving episode, Claire Vo walks listeners through building a highly personalized Thanksgiving party hub using "Lovable," a Vibe coding tool. Focusing on making event management software both practical and artful, Claire shares her pro tips for design upgrades (with Google Fonts and Midjourney), user customization (like dietary restrictions), and a time-saving workflow for integrating recipes with ChatGPT — all aimed at demystifying how to blend AI tools, design thinking, and code-light development in a real-world scenario.

Key Discussion Points & Insights

1. Choosing Lovable as the Party Hub Platform

Timestamp: 01:51 – 05:46

• Why Lovable? Claire chose Lovable over other Vibe coding tools because "it has a little bit nicer of a design component," feels more "human" for personal projects, and offers a "warm inviting autumn aesthetic."
• Initial Build: Prompted Lovable to create a hub for:
  • Managing invites
  • Dish coordination
  • Shared recipes
  • Photo gallery
• Claire’s Critique: The default "V1" app was “boring” with "a pretty, pretty boring stock photo," some "layout issues," and unimpressive typography.

2. Upleveling Design: Typography & Aesthetics

Timestamp: 05:47 – 14:58

• Leveraging Google Fonts: Claire shares that most Vibe coding tools support Google Fonts out of the box.
  • "One of the things that I think you can do to up level any design is get a really nice typography combination." (06:38)
  • Uses Canva’s Google Font combinations page for inspiration.
  • Chooses "Homemade Apple" for headlines and "Railway" for body text for a cozy, handwritten vibe.
• Implementation Demo: Shows how to instruct Lovable to use these exact font names, and highlights Lovable’s integration with Tailwind CSS for font configuration.
  • "Just by reading these three files, you can understand. This is how I pull in Google Fonts. This is how I configure them in Tailwind, which is very popular right now with some of these apps." (09:52)
• Result: The new typeface "is really, really, really adorable" — and looks "so cute," though Claire admits "not as readable, but… I don’t care."

3. Customization with Midjourney for Unique Visuals

Timestamp: 14:59 – 22:18

• Swapping Out Stock Images: Claire wanted a more original header image, so she turned to Midjourney.
  • References style code tips found on X (formerly Twitter), like Michael Rabohm’s "whimsy textured storybook paint" aesthetic.
  • Adjusts parameters, aspect ratios ("do a 2:1"), and demonstrates prompt tweaking.
  • Memorable moment: "Oh my gosh, this is totally giving me California Thanksgiving vibes… I really love this. I think it's super cute." (20:55)
• Personalization: Updates header copy to “Claire’s Thanksgiving feast," further personalizing the app.

4. Incremental Bug-Fixing and Honest Debugging

Timestamp: 22:19 – 26:58

• Live Demo of Vibe Coding Realities:
  • Claire encounters and narrates spacing/clipping issues, fixes icons ("replace the star with a pumpkin and leaf emojis").
  • Learns pumpkin emoji defaults to jack-o-lantern: "I forgot that the pumpkin emoji is by default a jack o lantern. We are past Halloween now..." (24:41)
  • Shares the workflow of iteratively testing fixes, taking screenshots, and prompting the app to make specific design tweaks.
  • Admits limitations: "Sometimes vibe coding is not perfect and I don't want to spend time today on optimizing this headline height."

5. Adding Custom Features: Dietary Restrictions Logic

Timestamp: 26:59 – 29:20

• Practical User-Centric Enhancements:
  • Adds dietary preference/restriction fields to guest management: "I have relatives that are vegan. I have a husband who is gluten free, dairy free. I have a me that wants to eat everything…" (27:18)
  • Chooses a multi-select approach for the most common dietary restrictions (“gluten free”, “dairy free”, “vegan”, “no restrictions”, etc.)
  • Maps dietary tags to dishes so all guests know which items meet their needs — "So we can now start to see what kind of dishes are coming, who can eat them, will they meet our guest needs?" (29:09)
  • Emphasizes the power of Vibe coding to quickly iterate on feature ideas without deep engineering work.

6. Recipe Management Workflow Using ChatGPT

Timestamp: 29:21 – 33:42

• Problem Identified: Traditional recipes online are challenging for families/kids — “you’re scrolling up and down… hands trying to figure out well, how many chocolate chips, how much butter, how much flour, because the measurements and the instructions are split up.” (30:31)
• AI to the Rescue: Claire’s workflow:
  1. Copy-paste recipe into ChatGPT (using GPT-4.5/5.1).
  2. Prompt for a new structure: “title, description, cook time, ingredients, servings and instructions… make sure both the ingredients and measurements are in line so I do not have to go back and reference the ingredients.”
  3. If model is slow or stuck, just "bail and restart" with another model.
  4. Get a step-by-step output with measurements for each step.
  • "For if you have a seven year old and you need to tell them to follow a recipe step by step, this is what we need."
• Memorable Moment: "We basically productized this flow" with a site called Runaway Pancakes (“easy recipes for their kid and adult helper helpers… includes pretty creepy but high quality mid journey generated images…”) (33:11)
• Final Integration: Shows adding AI-formatted recipes to her Lovable party hub’s recipe section for group sharing.

Notable Quotes

• On Personalization:
  "Something much more well designed, something that really feels like, you know, artisanally crafted farm to table software, which is what we want all of our Vibe coding apps to look like." (20:19)

• On the Realities of Coding with AI Tools:
  "Sometimes Vibe coding is not perfect and I don't want to spend time today on optimizing this headline height." (26:50)

• On Using ChatGPT for Recipes:
  "One of my personal favorite AI use cases, which is to bring those two things together … and put them in an order you can actually follow with your kids." (30:37)

• On Practical Prompts:
  "...With this prompting I was pretty specific that I said let's start with a multi select of the most common ones. And multi select means people can have more than one dietary preference, which they do." (28:05)

Key Timestamps for Important Segments

• 01:51 – Introduction to the Thanksgiving Party Hub concept
• 05:47 – Initial criticism and design improvement planning
• 06:38 – Discussing Google Font combinations for design upgrades
• 09:52 – Explaining integration with Tailwind and Google Fonts
• 14:59 – Using Midjourney and prompt styling for custom header images
• 20:55 – Reaction to custom Midjourney output
• 22:19 – Debugging layout and design issues live
• 24:41 – Issues with emoji choices and lighthearted typing fails
• 26:59 – Adding dietary restrictions as a feature
• 29:21 – The challenge of online recipe formats and AI-powered reformatting
• 30:37 – Using ChatGPT to make kid-friendly recipes
• 33:11 – Productizing the process (Runaway Pancakes site plug)

Final Takeaways

Claire Vo's episode is a step-by-step, honest, and fun showcase of using AI-powered, low-code tools to craft genuinely personal software solutions — from elevating the visual design of a holiday event hub, to creating sensible workflows for allergy management and collaborative recipes, to leveraging AI for family-friendly cooking. Her demonstration underlines that blending thoughtful design, prompting skills, and iterative development with AI can turn generic software into "farm-to-table" digital experiences — and even the imperfect steps are part of the creative journey.

Resources Mentioned

• Lovable Vibe Coding Tool
• Google Fonts & Canva’s Font Combos Page
• Midjourney (with references on X/Twitter)
• Runaway Pancakes (runawaypancakes.com)
• ChatGPT (GPT-4.5/5.1) for recipe formatting

If you want to explore the tools Claire used, experiment with prompt engineering, or simply are looking for ideas on making gatherings more inclusive and fun with AI, this episode is a lively, practical primer—full of candid, copyable workflows and authentic guidance.