anupsingh123

We are obsessed over speed. First version of agents were taking 10+ minutes to get a preview, now it takes less than 1 minute. We started with over engineering, kept experimenting and simplifying. Now we have final version of AI Agent that takes an idea and convert into working product in less than 1 minute. May be fastest right now? We should have an eval for speed testing AI App Builders.

I've been building an AI coding agent platform and kept running into this annoying problem: agents would get halfway through a task, hit an error or get interrupted, and completely lose context. They'd either start over from scratch or create duplicate files because they forgot what they already did.

Tried the usual approaches - longer context windows, better prompts, chat history - but none of it worked reliably. The fundamental issue is that LLM agents are stateless by nature.

So I built a state machine that persists everything to DynamoDB: - Which development phase we're in (requirements → frontend → backend → etc) - Granular todos within each phase - What's been completed vs what's pending - Sandbox state (E2B sandboxes can die/restart) - S3 code sync status

Now when something goes wrong (and it always does), the agent just resumes from the exact todo it was working on. No context loss, no duplicate work.

As a founder, I’m constantly tempted to build new developer infrastructure (“Layer 2”), but at JustCopy.ai we’re laser-focused on making agent orchestration (Layer 1) truly reliable. Instead of reinventing containers/cloud, our agents integrate with AWS, Vercel, etc.—and hand off work like a mesh network, not a linear pipeline.

This post shares honest lessons on agent reliability, context engineering, and avoiding “buzzword” distraction. Posting for feedback and real debate—are we missing something by not building Layer 2, or is this the right call?

btw this was DeepSeek-V3.2. If I'd been using Claude Sonnet 4.5, we'd be looking at a $2000 bill instead.

That's one approach. For me, the agent setup cut what used to be a full day of manual work down to minutes - even with the $200 learning tax, that's still a net win. But I get the skepticism.

I think there's some confusion about what justcopy does - it's for cloning YOUR OWN projects, not scraping other people's websites. Built it out of frustration when I tried to fork one of my projects for a different idea and it took a full day even with Claude Code and Cursor. Lots of manual config updates, dependency changes, renaming stuff, etc. The $200 mistake was about agent orchestration, not the ethics of the product. But appreciate the feedback - clearly need to communicate the use case better.

Yes, I built 7 AI agents for copying any website. The goal is to create production-quality copies. You can sign up and give it a try! I'm still refining these agents, but first I'm trying to add restrictions so they don't burn through my wallet lol

I started this project out of frustration. When I tried to clone other projects using Claude Code and customize them a bit—simple Next.js, ECS, CDK, and Express server setups—it took several hours just to get everything working. I realized that while vibe coding is great, it's still time-consuming to build a production-ready, functioning product.

How would that help with supervising agent runs for each user on justcopy.ai?

Classic "I'll be right back" moment that cost me real money.

Building justcopy.ai - lets you clone, customize and ship any website. Built 7 AI agents to handle the dev workflow automatically.

Kicked them off to test something. Went to grab coffee.

Came back to a $100 spike on my OpenRouter bill. First thought: "holy shit we have users!"

We did not have users.

Added logging. The agent was still running. Making calls. Spending money. Just... going. Completely autonomous in the worst possible way. Final damage: $200.

The fix was embarrassingly simple: - Check for interrupts before every API call - Add hard budget limits per session - Set timeouts on literally everything - Log everything so you're not flying blind

Basically: autonomous ≠ unsupervised. These things will happily burn your money until you tell them to stop.

Has this happened to anyone else? What safety mechanisms are you using?

I hate doing the same thing twice. And I was setting up React/Next.js/Container based projects constantly. New idea? New side project. Experimenting with something? You guessed it - new project. I Got Tired of Setting Up React/NodeJs Projects, So I Built AI Agents to Do It For Me. right now its highly opinionated and next I am making agents customizable because I know most developers are highly opinionated and they want agents to perform in their preferred way

What was the product?

I built it for my self as managing a large team and dealing with real estates, relationships, society is exhausting some time. I just dump my thoughts and live care free. I launched it recently and seeing a 100% WoW growth. Yesterday I started an fb ad just to see if this growth can be 10x. I am surprised that it's so simple and lot of players in the market, still people are actively using it. I have no big plan, it's just a hobby project.

My team is building something cool for running scheduled job. Let me know if you are interested

I hate setting up development environment. The more senior I became, the more context switch happens at work. What do you hate most at the work?