Claude Code & Vibe Coding My First App

Borrowed

"If you're looking for the fuel to get started, there's no fuel more powerful than your own curiosity. Following your curiosity can be scary, but if you give yourself space and time, eventually curiosities converge. Then you become untouchable."

From "The Heart to Start" by David Kadavy [Book]

Learned

Lately, I’ve had a hard time keeping my laptop closed after the workday ends. Not because of work pulling me back in, but because I’ve been in build mode. All thanks to Claude Code. Last week, I started creating a web app, and all I keep thinking about is my early days learning code.

When I first started writing and performing music in my early teens, I spent a lot of time promoting on MySpace. That’s where I met a girl from NYC who said she could make me a website. I showed her Rooney’s website, my favorite band at the time, and she replicated it using graphics I’d made in a torrented version of Photoshop (and maybe MS Paint). I hosted it on a free .tk domain. I’d soon learn that “tk” was short for Tokelau, a small territory in the South Pacific, where periodic storms would take down my site for hours at a time.

That’s what kicked off my HTML/CSS journey. I loved it. I’d get lost for hours figuring out how to make features, change layouts, and add pages. I continued tinkering in college, then, years later, at Barrel, I’d go much deeper, building interactive prototypes to show clients how the website I’d designed would move (long before Figma existed). I’d create Codepen after Codepen so developers could understand my vision for a design’s interactions. At one point, I considered changing tracks and becoming a developer.

Eventually, I found Webflow. Everything was easier, but it gave me the same energy. I jumped in headfirst, building things for clients, my own site, and many iterations of the Barrel site, including what we have today.

Claude Code is a whole new world, though. Within 24 hours of starting my idea, I had a working prototype of a web app that could have easily taken months, if not years, to build.

What I'm building

Every week, I update then review a series of spreadsheets on my own, with the team, and with Barrel founders, Peter and Wook. They give us a look into the business financials, from where things stand today to how we’re tracking toward our goals. We also have a system for reviewing the health of our client engagements with the Client Services team that lives in a Notion database.

For a while now, I’ve felt this tension: the manual effort was good for staying close to the numbers, but I spent more time managing it than looking at what it meant. There was also a disconnect between these conversations and their financial implications. We’d review the pipeline without considering how each opportunity affects our trajectory. Or we’d discuss the state of an account without an easy way to see when a contract ends or what services we’re currently offering.

This is what I set out to solve. First, for existing clients, I wanted an easy way to tie our weekly account review to the financial picture. And sure enough, that simple idea snowballed into a full-on dashboard I’m calling Barrel Agency Pulse.

It’s currently pulling from two data sources. One is manually updated by our finance team and shows how revenue will be accrued for every project. The second shows our pipeline across new and existing accounts, with the source of truth coming from HubSpot.

Instead of spending time assembling views manually each week, everything updates automatically. I can see committed revenue by month against our targets, pipeline deals weighted by likelihood, each Account Manager’s book of business, and detailed client pages that show what we’re delivering and when contracts end. These will eventually replace the Notion database. What used to live across three or four different places now lives in one. It’s made it easy to explore new features like looking back to when our oldest client joined us to see changes in revenue, number of clients and projects, and client retention.

I’m still very much in build mode, which is what’s so exciting. I know I’m just scratching the surface of what’s possible and already starting to see the business differently. The next big piece on the roadmap is integrating our call recordings/transcripts to centralize information across prospects and current clients. We built a solution last year, but it’s not as seamless as what I’m aiming for here.

There are many other ideas on my mind, like tracking profitability and debrief takeaways, but once the call integration is in place, I plan to pause on new features, start layering in AI for insights, and see how it all works in practice. For now, I’ll be using the tool on upcoming calls with the team and hope to have them as users in the coming weeks.

Shoutout to Peter for the recent inspiration with what's he's built using Claude Code.

& Thought

Where in my work am I spending more time assembling information than learning from it?