Recap from last week
Like all TV show recaps, you’re free to skip this segment.
I’m building Agora to make learning philosophy more accessible by making reading the great books more social.
I’ve built multiple prototypes to validate various hypotheses.
I’m now building a proper product testing two key things:
Does social reading create retention?
Will engaged users pay for expert insights?
I have a 4-5 month roadmap to build a reader, questions at the end of each chapter and professor insights sprinkled through various books to test this (as well as user growth targets and customer interviews in that time).
What I built last week
I’ve built a screen to login and sign up.
I’ve built a dashboard where you can choose the book you want to read (I’ll be controlling the books on the platform).
The tech stack I’m using is Next.js, React, Supabase, Tailwind CSS, Vercel.
Tools and tricks
I used various forms of Claude a lot and a bit of ChatGPT. I’ve always been a ChatGPT guy but the past week I’ve found myself using Claude way more. Its output from writing to coding is significantly better (at this stage).
The most important thing I’ve been doing is going through each line of code I write and understanding how it interacts with everything.
If Claude goes off on a run (one of those times it gallops away writing hundreds lines of code when you asked it how many mls of coffee go into a triple shot latte) I tell it to start over and just solve the very first problem we were looking at - that’s it.
I’m building this app and remembering how to code at the same time here, so I’ve built a project in Claude that’s my coding tutor. My tutor has been programmed to ask me a question to challenge my understanding of each concept used. Sometimes it’s too basic but often it actually challenges me.
Every key concept I learn, I turn into an anki card and I go over these each morning.
Helping me build features has been Claude Code in the command line, and it’s been a game changer.
I’ve created various sub-agents which take the work from good to amazing. I’ve created a sub-agent to be a frontend developer, a backend developer and a brainstormer (plus quite a few others).
These are simply markdown files that I’ve added to the code I’m writing telling the agent how to act. I call on them in Claude Code as I need, i.e. when I’m discussing a specific feature with it.
I’ve added some superpowers thanks to Jesse at Massively Parallel Procrastination and Claude Code adding a plugin marketplace. Just started working with them on Friday arvo, but results seem promising so far. You can install them using these commands:
/plugin marketplace add obra/superpowers-marketplace
/plugin install superpowers@superpowers-marketplaceWhat have I read
A great JavaScript teacher: https://javascript.info/
This blog is where I’ve got a lot of my Claude Code tricks from: https://blog.fsck.com/2025/10/05/how-im-using-coding-agents-in-september-2025/
If you’ve got no understanding of Claude Code and how it differs from Claude (normal), this is a good read: https://www.producttalk.org/claude-code-what-it-is-and-how-its-different/
I actually found the next.js article on learning react very good: https://nextjs.org/learn/react-foundations
Not related to building this, other than it’s a great read to keep things in perspective: https://www.themarginalian.org/2025/10/14/annie-dillard-weasel/
Another thing not totally related, but I like reading anti-AI articles to keep balanced (it can be very easy to be swept up in the whole AI is the only thing in the world tsunami): https://www.anildash.com/2025/10/17/the-majority-ai-view/
Next up
This week and next week is all about building the reader. By the end of this week, I should be able to log on, choose a book and start reading it. Keep your thoughts with me.
Nick