DEV Community: Omar Zahouri

I Built an AI Development Tool Because I Was Tired of AI Subscriptions

Omar Zahouri — Sat, 07 Nov 2026 00:50:12 +0000

I like using AI for software development.

It can help with debugging, architecture, refactoring, documentation, and even building entire features.

But there was one thing that kept bothering me:

Why does using AI for development have to mean subscribing to another platform?

I wanted something different.

I wanted to choose the AI model myself.

I wanted to use my own API key.

I wanted to keep my projects under my control.

And I didn't want another mandatory account, token system, or subscription standing between me and my code.

So I decided to build it.

The idea

I started building NullCry as an AI development environment focused on developer control.

The concept is simple:

You bring your own API key.

Instead of forcing developers to purchase AI credits from another platform, NullCry can work with the developer's own API access.

That means the developer can choose the provider and model that fits their needs.

No NullCry tokens.

No mandatory NullCry subscription.

No requirement to use one specific AI provider.

But the difficult part wasn't the API

Connecting an AI model to a web interface is relatively straightforward.

The harder problem is building something that can actually work with a real software project.

A real project isn't one text prompt.

It can contain:

Multiple files
Existing code
Dependencies
Previous changes
Bugs
Large files
Different programming languages
Changes that affect other parts of the project

So I focused heavily on context and project continuity.

The goal was to make the AI work with the project rather than treating every request as an isolated conversation.

What NullCry is trying to solve

The philosophy behind NullCry is:

«Your AI. Your Projects. Your Freedom.»

Developers should be able to choose their AI provider instead of being locked into a single ecosystem.

They should also be able to experiment with different models without rebuilding their entire workflow around a subscription.

That's the direction I'm trying to take with NullCry.

I'm not claiming it's perfect

This is still a project I'm developing and improving.

There are things I want to make better.

There are probably things developers will notice that I haven't considered yet.

And that's exactly why I'm sharing it with the developer community.

I want honest feedback.

If you were designing an AI development environment around BYOK, what would you add?

Would you prefer this model over a traditional AI coding subscription?

Or do subscriptions provide enough convenience to make the trade-off worthwhile?

I'd genuinely like to hear how other developers see it.

NullCry: https://nullcry.com

AI won't make software developers obsolete. But developers who refuse to adapt may find themselves competing with those who know how to use it. The real skill is no longer just writing code. It's understanding systems, solving problems, making decisions,

Omar Zahouri — Fri, 06 Nov 2026 14:23:42 +0000

Beyond Code Generation: Building Complete Software Projects with AI

Omar Zahouri — Sat, 31 Oct 2026 14:57:41 +0000

Building AI coding workspaces around complete projects, not isolated snippets

AI coding tools are becoming increasingly common, but there's an important distinction between generating a piece of code and working with an entire software project.

A real software project is more than a single file. It contains interconnected files, dependencies, configurations, and existing logic that must remain consistent.

One interesting approach to this problem is designing an AI workspace around three principles:

Understand the existing project

Instead of starting from scratch every time, an AI coding workspace should be able to receive an existing project through a folder, ZIP archive, or pasted source code.

Change only what's necessary

When a developer requests a modification, the system should identify the relevant files and avoid unnecessarily rewriting unrelated parts of the project.

Keep the complete project accessible

Developers should be able to inspect the resulting files, review changes, and continue working on the project.

These principles are part of the design direction I've been exploring while building an AI development workspace.

I believe the future of AI-assisted development should focus not only on generating code, but also on understanding and maintaining complete software projects.

I'd be interested in hearing how other developers approach this problem.

What do you think is the biggest challenge facing AI coding tools today?

Disclosure: This article was prepared with AI assistance.