I'm a non-coder who ships apps with AI.
My biggest fear: the AI quietly building the wrong thing — and me not catching it, because I can't read the code.
So I spent ~5 weeks building Window Memory. A supervisor for your AI.
Here's the story 🧵
I subscribe to and use all three tools, building one ongoing project with WM.
Managing individual MD files or entering prompts like “review the project and...” wastes time and tokens. Instead, I handle detailed architecture with Claude Code, brainstorming and frontend building via parallel agents in Cursor, and verification with Codec—all without re-explaining anything to separate AI tools. I simply open the same project folder and continue building as usual.
At the start of each section, the AI automatically pulls the latest decisions and events, then suggests the build order and tasks first.
If the work deviates from project goals, a red flag appears on the dashboard.
I subscribe to and use all three tools, building one ongoing project with WM.
Managing individual MD files or entering prompts like “review the project and...” wastes time and tokens. Instead, I handle detailed architecture with Claude Code, brainstorming and frontend building via parallel agents in Cursor, and verification with Codec—all without re-explaining anything to separate AI tools. I simply open the same project folder and continue building as usual.
At the start of each section, the AI automatically pulls the latest decisions and events, then suggests the build order and tasks first.
If the work deviates from project goals, a red flag appears on the dashboard.
@gem_dolphin Window Memory helps AI coding tools remember why your project was built a certain way. It works across Claude Code, Cursor, and Codex. No manual memory files. Just install and keep building.!! Please check my WM !
Hello @X
I want to connect with builders, creators, and thinkers who actually release products.
If you are interested in this topic:
• AI & Agent 🤖
• Web development 🌐
• Open source 💻
• Data & Visualization 📊
• Startup & Startup 🚀
• Content production & growth 📈
• No code / Automation ⚡
• Crypto & Blockchain 🔗
Please follow me or say "Hello" in the comments👇
Let's learn, make, and grow together.
#BuildInPublic #AI #Startups #WebDev #OpenSource #Entrepreneurship #Crypto #NoCode #GrowthHacking
Thank you so much for the wonderful compliment! 🙏
While building WM, I created it using WM itself, and I'm also working on another marketing automation tool.
Honestly, what I liked most is that even when using Codex, Claude, and Cursor simultaneously on one project, they share everything through a single event card.
AI tools are smarter than me, so I can run them in parallel without conflicts while comparing them!
I know other memory services exist, but this seems especially helpful for non-developers since you can see it directly.
@BuildWithxAI I'm very interested in exercise and work out up to three times a day to stay consistent with my routine and keep my initial motivation!
I'll take a closer look!!
Hey @gregisenberg,
I’d love to join your podcast.
I’m a former restaurant owner from South Korea with zero coding background. I used to run a Western-style restaurant in Korea and had never written a single line of code. But I got completely hooked on AI coding tools (Cursor, Claude Code, etc.) and now I’m all in.
I noticed how many non-technical people like me are jumping into AI coding, but they struggle with context loss, token limits, and the AI forgetting “why” they made certain decisions. So I built Window Memory (https://t.co/yBb8Ltum6q) — a tool that helps non-developers keep full context and reasoning across sessions in Cursor, Claude Code, and Codex.
I’d love to share my story: how a complete beginner went from running a restaurant kitchen to building AI tools, and what I’ve learned along the way to help other non-technical founders get started with AI coding.
Hey @iraftopo,
Just adding a bit more — I’m personally using Claude Code subscription tokens to build, and I switch between Cursor, Claude Code, and Codex all the time. Window Memory helps me keep the full context and reasoning across all of them without burning extra tokens re-explaining everything.
Since you’re also hitting token limits hard, I thought it might help you too.
Ah, and I’m a father of two as well! Let’s keep pushing together! In Korea, we call older guys “hyung” (형) 😂 Hope this helps you out, hyung!
Hey @iraftopo,
I saw you're running out of tokens again with Cursor/Claude. I’ve been there.
I built Window Memory (@windowmemory) exactly for this problem. It lets you switch between Cursor, Claude Code, and Codex while keeping full context and the reasoning behind every decision across sessions.
No more re-explaining the whole project every time you start a new chat or hit a reset. It runs locally, catches drift, and saves a ton of tokens.
If you want to try it: npx -y window-memory init
Check it out here → https://t.co/e6QR8YG0gr
Would love to hear if it helps with your workflow!
Like Elon, I want to be someone who helps advance humanity.
Ever since I was a kid, my dream has been to see with my own eyes the futuristic scenes from those sci-fi movies.
That’s why I hope AI advances even faster, and I’m committed to building things that help make that happen. 🚀