Vibe coding aims to replicate Adobe software through generative AI

A single developer is using AI to replicate the entire Adobe Creative Suite through a process described as "vibe coding," according to @kirkyuhnke on TikTok. This project aims to provide open source alternatives for Photoshop, Premiere, After Effects, and Illustrator that run locally on Windows, Linux, and macOS. The developer intends to reach near-total feature parity within a few months by leveraging generative AI to assist in the software construction process.

How it works

The source does not specify the exact technical stack or the underlying architecture used to build these applications. It focuses on the development methodology rather than the software engineering specifics. According to @kirkyuhnke on TikTok, the developer uses generative AI to facilitate a process called "vibe coding" to replicate existing software functions. This approach relies on AI to assist in writing the code required to mirror the features of Adobe products. The resulting binaries are distributed as open source software for desktop environments.

1. Generative AIThe model processes natural language or high-level intent to generate code segments.
2. Vibe coding processThe developer uses AI outputs to construct the application logic and feature sets.
3. Open source repositoryThe code is made available for public inspection and modification.
4. Desktop installerThe compiled software runs locally on Windows, Linux, or macOS systems.

What it takes to run

Deploying these applications within a professional production environment requires more than just installing the binaries. Because the source describes these as local desktop applications, they must integrate into existing asset pipelines to be useful. A studio cannot simply swap tools without considering how the data flows between them. You need to account for file format compatibility, shared storage access, and the personnel required to manage a non-standard software stack.

• Centralised storage or NAS systems with high-speed access for large media files.
• Standardised file formats that ensure interoperability between these new tools and existing creative assets.
• A dedicated DevOps or IT workflow to manage updates and security patches for open source software.
• Hardware with sufficient GPU and CPU resources to handle the heavy rendering tasks typical of video and motion design.
• Version control systems to manage project files and assets across a distributed team.

Where it breaks

The primary risk lies in the gap between functional parity and professional-grade stability. Replicating a user interface is not the same as replicating the underlying processing engines. While the developer aims for high feature parity, the source does not explain how these tools handle complex, non-destructive workflows or heavy compute loads. In a production environment, software must be predictable under stress, not just visually similar to the original.

• Lack of documentation regarding memory management and GPU acceleration for real-time playback.
• Potential for proprietary file format corruption when attempting to round-trip assets between Adobe and these clones.
• Uncertainty regarding the security of the build pipeline for open-source binaries distributed across multiple operating systems.
• Absence of verified plugin architectures or third-party API support for extended functionality.

How we would build it

If we were tasked with implementing a custom, open-source creative stack for a client, we would avoid a wholesale replacement. Instead, we would build a hybrid environment where these tools act as specialised nodes within a controlled infrastructure. This ensures that a single point of failure in an unproven codebase does not halt an entire production cycle.

1. Containerised deploymentWe would package the applications in containers to ensure consistent runtime environments across Windows, Linux, and macOS.
2. Centralised asset managementWe would implement a unified file server to manage the movement of raw data between different toolsets.
3. Automated regression testingWe would run scheduled scripts to verify that new updates do not break existing project files or render outputs.
4. Version control integrationWe would link all project metadata to a versioned system to track changes across the distributed toolchain.

Within a year, LLMs will generate more coherent glue code, but they will not build reliable processing engines. For vibe coding to succeed, models must move beyond syntax to understand the strict state management required for professional-grade data pipelines.