# Looking for advice on AI-driven development

> Source: <https://dev.to/rfromc/looking-for-advice-on-ai-driven-development-5hb7>
> Published: 2026-08-31 07:47:58+00:00

Greetings! Hope you're all having a marvellous day!

I'm looking for some guidance, as I feel like I'm hitting a soft roadblock.

I'm not technical, but my work experience comes from working in software development companies for many years. As such, I'm familiar with many development processes, which seems to be helpful.

I've spent the last six months learning and slowly building my own app, moving from simple vibecoding to spec-driven development (using Claude Code in VS Code).

However, as my app (a cataloguing app based on React, TypeScript, and Vite) gets more and more complex, and the tasks seem to eat up more and more tokens, I keep getting a feeling I might be doing something wrong. For example, Claude prompts me to run `/code-review`

in a loop sometimes, and I'm not sure if that makes that much sense after the second or third time.

I keep reading guides and digging for new knowledge (at the moment I'm trying the whole subagents thing), but it starts to feel like I'm chasing my own tail. By the time I get to try something I learnt about, like the Task tool in Claude, the Claude docs say it is not useful in Opus 4.8 any more.

Then there is the whole MCP thing, which I haven't looked into yet, and I'm not even sure I need to.

**So, my question would be: how would you advise me to move forward?**

I'd like to release my app at some point next year, and I don't want it to be some useless slop, but the speed at which AI development is progressing overwhelms me a bit.

Let me know if I need to provide more details about the app or the way my development process is set up.

Thanks in advance!