My life has been taken over by AI projects and news. It’s a non-stop echo chamber bubble in Silicon Valley. At work (Box), I help launched Box’s AI APIs and Aaron still shares his AI thoughts 128x a day. At home, I’ve been contemplating what kinds of experiments I can undertake with AI.
As a PM, a former founder/hacker (with an emphasis on hacker, not a real software engineer), and someone whose brain is basically an idea factory, this “vibe coding” concept with AI is proving to be a dream come true. I get to roll up my sleeves, learn the latest AI tricks, AND maybe, just maybe, turn one of these experiments into a real, live app!
So, Why Now? What’s the Big AI Deal?
I’ve had a front-row seat to AI at Box, launching AI APIs and watching things like MCP, AI Agents, and creator productivity totally transform. It feels like we’re in that sweet spot: innovation is still in exponential growth, but things are just stable enough to actually build cool stuff without it changing an hour later. AI is early enough to that new use cases are unfolding daily, but not so bleeding edge that you can’t get your footing on the technology.
My “Vibe Coding” Guiding Principles
Before diving into these AI experiments, I established a few guiding principles:
Avoid B2B SaaS Commodity Traps: Having been a B2B SaaS founder and PM for so long, and even bought/sold a SaaS business, I know that many “new” ideas are basically copies of old, but slightly better. The real differentiator often lies in marketing, positioning, and sales. For these personal projects, I’m looking elsewhere.
Focus on Passion or Problem-Solving: Time is a luxury these days. If I invest it, I want to be enthusiastic about the project or the problem it addresses.
AI-first Approach: The problem should be 10x easier to solve with GenAI compared to traditional programming. The AI should be core to the solution.
Potential for a Side Project/Hustle: The ability to turn it into a side income source is a definite plus.
The Hack Lineup
Here are the ideas I’m exploring. This post will focus on the first one:
Collaborative AI Travel Planner with Replit) — This is the focus of today’s post.
AI Basketball Coach from Video Recordings (with Firebase Studio) — Imagine AI providing tips on your basketball technique. Further imagine that it’s Michael Jordan coaching you! More on this in a future update.
Text to Calendar Events (TBD) — To simplify event scheduling. Stay tuned for this one.
Ideation Zone: Exploring ideas for transforming blog posts/docs/text into other formats (podcasts, infographics, 3-min videos, etc.). The possibilities are numerous.
Hack #1: Revisiting My Trip Planner Concept with AI & Replit
Let’s go back a bit. Back in 2009, before Pinterest gained widespread popularity and Google Docs became the standard for collaboration, I founded Duffelup.com (coverage by TechCrunch). It was a collaborative and visual trip planner, merging aspects of visual appeal similar to Pinterest with Google Docs’ collaboration features. I’ve always enjoyed planning trips, especially when friends/family also chip in ideas.
Cork board for friends & family to pin ideas
That project involved over two years of nights and weekends learning and coding Ruby on Rails. As the only programmer, I also learned about deploying to AWS (back then, only EC2 existed). This was when I gained experience as a “full-stack developer” somewhat unintentionally. To summarize, it reached over 50k MAU without any marketing, generated a whopping $10/month from affiliates, and I decided the return on investment wasn’t sufficient for the effort involved. I sold it and used the proceeds as a down payment for my VW Golf (not a house, sadly, but still a positive outcome).
Fast forward 15 years. I recently planned a trip to Spain, and the process was familiar: going through notes from friends, copying and pasting ideas into a Google Doc, and manually mapping them to see their locations. I realized that a gap in collaborative travel planning persists. So, I decided to revisit my trip planner concept, this time using Replit to “vibe code” it with AI assistance.
This post isn’t a detailed product review, but here’s a glimpse of the product and some initial learnings:
I asked for a Google Authentication button with the logo. Replit nailed the visuals perfectly on first try. The backend logic? Well, it’s basically vaporware.
I’m impressed that it automatically added a few destinations with images, instead of a blank canvas.
After creating a trip, I can add a few items on this “visual board”. I didn’t further refine the look & feel, which needs more love.
Replit even designed this nifty modal without me even asking.
Prompt #1:
Create a trip planning website that multiple users can collaborate and plan trips together. The website will allow users to create itineraries, add hotels, transportation, top visited sites, and automatically suggest restaurants around the top sites and hotels. The trip planner’s UI will look like a corkboard with drag and drop interface, almost like users are pinning ideas to a board to make trip planning fun and visual.Prompt #2:
For user authentication, also add Google oauth with a “Sign in with Google” button.Instant Gratification: The first prompt spat out basically everything you see in the screenshots. For prototyping? Insanely good.
PM Dream, Developer’s “Hmm…”: This is fantastic for Product Managers like me to quickly show, “See? This is what I mean!” But, let’s be clear, it’s a galaxy away from a production-ready app. The database isn’t set up, authentication is flaky, drag & drop doesn’t work, etc.
Deployment Hiccup: Getting it live seemed a bit tricky at first glance. Looks like I’d need to manually set up Firebase, which adds a bit of friction.
Time Spent: Just 1 Hour! This includes signing up to Replit (and I’m the type of person who skips tutorials). For a functional prototype that actually looks like something? Impressive.
That’s the summary on my first AI vibe coding experiment! It’s amazing how quickly you can get an idea off the ground these days.
Stay tuned for updates on my other AI hacks — the AI Basketball Coach is calling my name!
What about you? What AI experiments are you cooking up? Drop your thoughts and ideas in the comments below — I’d love to hear what you’re working on!