Let's say I have a great product idea. I know exactly which problem it solves, what kind of experience it should offer, and how it should look. But the moment I try to explain it to the AI, the words fall apart, and I find myself staring at the blank terminal screen.
Should I start with the design? The features? The technical requirements? The idea that was crystal clear in my mind just moments ago starts to blur as soon as I try to put it into words for an AI coding assistant.
Sound familiar?
If you've ever used tools like Claude, Cursor, v0, Bolt, or Lovable to build something, you probably know this feeling. You have the vision, but translating it into something the AI can actually work with? That's a whole different challenge.
So I decided to fix it. Not just for myself, but for every designer and developer who's been through the same frustrating experience.
The Problem Nobody Talks About
The "vibe coding" era is here. You can describe an app idea and watch AI build it in minutes. It's exciting. But there's a gap that most people don't talk about: the quality of what AI builds depends entirely on what you feed it.
And most of us aren't giving it nearly enough to work with. Not because we're lazy, but because explaining a product idea comprehensively is actually really hard. You think about the UI but forget to mention the tech stack. You describe the features but skip the user flows. You nail the color palette but never mention how the navigation should work.
I kept running into this hurdle as a product designer. Every time I started a new project with an AI assistant, I'd spend 20 minutes writing a prompt, hit enter, and then realize I forgot to mention half of what mattered. So I'd rewrite it. And rewrite it again. And again.
There had to be a better way.
A Quick Word About PLAIN
Before I introduce Plainify, let me quickly explain the format behind it.
When I first started using AI to generate code, I went down a rabbit hole of workflow research. It didn't take long to notice the obvious: the quality of AI's output is directly tied to the quality of your input. A well-defined, structured spec file made a huge difference. Not just in the initial output, but also in reducing the back-and-forth with AI on both design and code after the project got off the ground.
So I decided to create a static file format where you could lay out everything a product might need, in a plain and structured way. I gave it a fancy name: PLAIN, which stands for Product Language for AI Notation. I kept refining it throughout my projects, updating sections based on what worked and what didn't, and it eventually took its current shape.
It's an open-source project, so it's very much alive and open to growth. If you have ideas, suggestions, or want to contribute, you're more than welcome. Instead of scattering your specs across Figma files, Notion pages, Slack messages, and that one Google Doc nobody can find, PLAIN puts everything in a single markdown file.
It covers everything from project overview and value proposition to design direction, tech stack, data models, and component inventory. Each section captures a different dimension of your product. Skip one, and the AI might make assumptions you didn't want. Fill them all in, and the AI has everything it needs.
PLAIN is open source, MIT licensed, and you can check out the full spec on GitHub. But here's the thing: filling out PLAIN from scratch can also feel overwhelming.
How Plainify Was Born
I realized this while preparing a vibe coding workshop for my design team. We were going to create specs together, and I already knew the pain point: remembering what each section needed would slow everyone down. What we really needed was something with ready-made options, dropdowns, checkboxes, simple form inputs that would let us fly through the whole spec and get a markdown file at the end without overthinking every detail.
So while putting the workshop together (which we ended up calling "Vibe Designing"), I thought: maybe I can actually build a quick prototype that does this. I managed to get it ready just in time for the first session. We tried it, it worked, and the feedback from my teammates was really positive. That's when I knew this was worth building properly.
That's how Plainify was born.
So What is Plainify?
Think of PLAIN as the knight and Plainify as its squire. The knight does the heavy lifting, the squire makes sure it shows up fully prepared.
Plainify is a free web tool that guides you through creating PLAIN specifications without the blank-page anxiety. You don't need to know the format by heart or worry about missing a section. Instead of staring at an empty markdown file wondering where to start, you get a structured, almost conversational experience that walks you through everything step by step. You just answer the questions, make your selections, and Plainify takes care of the rest.
Two Ways to Use It
Relax Mode is for when you just want to get moving. Describe your product idea in a sentence or two, something like "A smart recipe app named Culinara that suggests meals based on the ingredients you have at home." Pick your AI model, lean back, and you're done. In seconds, you get a complete specification covering all 14 sections, ready to feed to any AI coding assistant.
Relax Mode: describe your idea, attach visual references if you have any, and hit generate. That's it.
Custom Mode is for the control freaks, myself included. It walks you through each section with curated options, examples, and context-aware suggestions. You choose your design style, color palette, tech stack, UI library, and everything else. When you hit "Generate," one of Claude's Haiku models (3.5 for speed, 4.5 for depth) processes your selections and creates a comprehensive markdown file.
Custom Mode: walk through each section at your own pace, fill in the details, and generate your spec when you're ready.
Both modes produce a {project-name}-plain.md file that you can preview, edit, download, and drop straight into your project directory. From there, you just point your AI assistant to it:
Read the project-name-plain.md file and build the initial structure of the application based on the specifications inside.
That's it. No more writing long prompts from scratch.
Preview, edit, copy, or download your generated spec. What you see is what your AI gets.
Want to see the real output? Check out the culinara-plain.md file!
And for any section where you're not sure what to choose, there's a "Let AI decide for me" option built right into the dropdowns and checkboxes. For free-text inputs, you can simply type the same thing. Either way, the AI will pick modern, compatible options that fit with the rest of your spec.
Privacy is Not Optional
This one matters to me a lot. Plainify runs entirely in your browser. No accounts. No sign-ups. No server-side storage. Your product ideas never leave your device.
In a world where every tool wants to collect your data and train their models on your ideas, I wanted to build something different. Your specifications live in your browser's local storage. That's it. The only analytics I use is Google Analytics for understanding general usage patterns. I never see or store your actual content.
You can build genuinely useful tools while respecting people's privacy. Plainify is my proof of that.
Under the Hood
For the technically curious: Plainify is powered by Anthropic's Claude Haiku models. You can choose between Haiku 3.5 for speed or Haiku 4.5 for more detailed outputs.
There's also visual reference support. Upload screenshots, mockups, or wireframes, and the AI will analyze the colors, layout patterns, and component structures to reflect them directly in your specification. It's super useful if you already have a design direction but need to translate it into a structured spec.
The output is always a clean markdown file. Why markdown? Because it's the one format that works everywhere. AI can read it, humans can read it, Git can track it, and any text editor can open it. No proprietary formats, no lock-in.
The Person Behind Plainify
I should have probably introduced myself by now. I'm Halil, a product designer from Istanbul. I've been designing and building digital products for years, and when AI coding assistants started becoming genuinely capable, I got really excited. Going from a design concept to a working product without waiting for a development sprint? That's a dream for any designer.
My dream turned into PLAIN and when I realized the template itself was still intimidating for people, I built Plainify to make it accessible to everyone.
The whole thing was built with Claude Code, and I have to say: it's hands down the best AI coding assistant I've ever used. I've tried pretty much everything out there, Cursor, Copilot, you name it. Nothing comes close. Claude Code understands context like no other tool, picks up on what you're trying to do, and writes code that actually makes sense. As a designer who codes, that matters a lot. I don't want to debug for hours, I want to build. Claude Code lets me do that.
As someone who builds open-source tools, there's nothing more rewarding than creating something that helps fellow designers and developers do their best work.
What's Next?
Plainify is free and will stay that way. I'm actively working on new features based on user feedback, and I'm building everything in public. And while Plainify itself isn't open-source yet, its backbone, the PLAIN format, is fully open-source and open to your contributions. That's where the real magic lives, and I'd love for the community to help shape it.
If you've ever struggled to explain your product idea to an AI, give it a try. And if you have feedback, good or bad, I genuinely want to hear it. Every great tool starts as a simple idea, and with your help, this one will keep getting better.
See you around. 👋
As always thanks to Selen Yanmaz for her priceless editorial support.
Plainify: Turn Your Product Ideas Into AI-Ready Specs in Minutes was originally published in Plainify on Medium, where people are continuing the conversation by highlighting and responding to this story.