making it work #5
Remy vs Claude Code vs Replit (hackathon life) & more tools!
This is the fifth post in a new series called making it work — where I document what I’m actually building with AI, what broke, what surprised me, and what it means if you’re a marketer who keeps hearing terms you don’t know but nodding like you know what it all means. (I was you. Recently. It’s fine.)
What I’m building
Recording from a cafe in Miami. My partner is speaking at Consensus. I’m here vibing, and there’s apparently a main road directly in front of me, so apologies for the noise in the video!
Some exciting career and project stuff is in motion. I’ll write about it properly in an actual essay soon. Also, sorry for the delay. I’ve been in full event-hosting and planning mode & interviewing pretty heavily. Here’s me making up for the lack of an update last week!
For now: the build update.
I wanted to build the same thing (my lil private skincare app) to test how different platforms would implement it. Some things worked really well; some didn’t. And, well, as usual, I want to try more things. Let’s get it!
Claude Code: deployed and mostly working, the keyword is ‘mostly’
I deployed my skincare app through Vercel this week. It’s live, I did that!
Overall, the UX is solid. Wishlist, stash, and routine are all built out. The core flows are there, but there are a few things I’m still working through:
The design is very Claude-coded. If you don’t code, that’s still impressive. But if you have a design opinion, you’ll notice it immediately. I need to do something about that.
Everything also pre-populates too much. I want the app to feel more blank and editable by default, not filled in with placeholder content out of the gate.
The AI feature doesn’t work yet. I have a mentorship/build session tonight specifically to get through that. So… deployed, functional, AI pending.
The honest take on Claude Code for non-technical builders: you can control a lot, but there are always small things to figure out with each feature. It’s not magic. Every piece requires some back-and-forth and figuring out. If you don’t have that patience, you’re probably better off with Remy.
Remy: actually worked end-to-end
Remy (by MindStudio) did the most, by a significant margin.
The AI works. I tested it live, talked to the AI skin concierge, and asked it to add items to my wishlist. It did all of that, and the wishlist save worked! Home also loaded eventually. Happy days!
The sign-in flow is different, too. It sends a link to your email instead of a password, which I had never thought of. I liked it, although I’m not sure about the safety aspect. I just know that I hate remembering passwords.
Vibe-coding across different platforms has definitely shown me what’s possible and what I like. Building some product intuition here.
The way I’d describe Remy now, after actually building on it: it’s a product agent that codes, not a coding agent that needs product guidance. And that distinction is bigger than it sounds.
Most tools in this space are coding agents that do a little bit of PM and design on the side. Remy flips that. It leads with the product (specs, planning, and approvals) and then builds on it. It feels fundamentally different. Here’s an IRL example of a plan:
Think about how a real product team works. You don’t just have engineers. You have PMs, PMMs, UX designers, people thinking about the user experience before a line of code gets written. Most vibe-coding tools skip all of that and hand you a blank prompt. Remy actually does the PM work first. It came up with specs, had me approve them, then built from there. I also just saw this roadmap:
And this pitch deck:
And color scheme, typography, visual identity, etc:
And part of the initial release notes (handy for product marketing):
Remy is useful well beyond landing pages. It’s thinking about the product, not just the interface. Try it (Remy)!
Cost note: I’ve spent about $250 total on Remy. $200 of that was credits, so effectively $50 out of pocket. Still worth flagging.
Replit: not for me
I also built the same project on Replit during the Remy/Replit hackathon. I gave it the same brief I gave Remy and Claude Code.
I didn’t like it. The design wasn’t there, and the build felt off. Same instructions, noticeably worse output. I’m not writing it off entirely, but for this project, it wasn’t the right tool.
Tools I want to try next
Lenny just dropped his list of products he’s been using. Two I’m adding to my list:
CleanShot — screen recording/capture tool, seems affordable, trying it this week.
Rocket — makes emoji shortcuts actually work. The default Mac fn shortcut is unreliable for me, which is a problem when you’re doing video-style content.
Willow — a new Wispr Flow?!
Magic patterns — another AI prototype ish / design tool?!
If you have other screen recording tools you like, send them my way too. Screen Studio sucks.
Where I am
Current ranking for non-technical builders who want something working fast:
My preference in order would be:
Remy
Claude Code
Replit (yes, I’d rather just built it myself on Claude Code)
More to come when the AI feature is actually live! Hopefully by tonight… see you next week!
Subscribe if you want this in your inbox every week — honest, mid-process, no clean answers. Or reply and tell me what you’re building. Going to be doing this every Thursday!