Duxko began as an editor connected directly to Figma, designed to turn designs into HTML and let users preview and edit the result in its own environment.

The initial goal was to bring design and development closer together, reducing the work required to turn interfaces into code. Over time, the idea evolved into a platform for creating websites and applications, along the lines of tools such as Lovable.

The arrival and rapid evolution of artificial intelligence significantly expanded that vision.

Evolving into an AI Software Production System

Duxko was developed as an MVP of an AI Software Production System, bringing development, design, and AI agents together in one environment.

Implemented features included Figma integration through an MCP built from scratch, an integrated editor, website previews, a management dashboard, and a user system with roles and permissions.

The plan was to evolve that foundation into a complete software production platform, integrating agent orchestration, project management, tasks and tickets, development environments, and delivery processes, while keeping manual intervention and editing available.

Why I put the project on hold

Duxko was developed during a period of very rapid change in artificial intelligence. New models, coding agents, tools, and integration possibilities kept emerging, requiring technical decisions and the product architecture itself to be reconsidered.

At the same time, Duxko's scope kept growing. Building and maintaining a platform that integrates development, infrastructure, AI agents, and project management involves considerable complexity.

As an independent developer without the budget to maintain a dedicated team for a project of this size, I had to assess whether continuing to invest in it was viable.

The combination of rapid technological change, growing complexity, and limited resources led me to put Duxko on hold. The decision reflected the effort and resources required to execute the idea properly, rather than a loss of interest in it.

My vision for Duxko's future

I still believe there is room for a platform that brings the entire software production cycle into one environment.

I imagine something combining the capabilities of tools such as Devin, Jira, GitHub, and Figma, with an integrated experience from planning to delivery: a platform where users could create a project, define requirements, organize tickets, and let different AI agents work together.

Agents could take on specific responsibilities, such as architecture, frontend, backend, testing, review, and infrastructure, using different models, skills, and reasoning levels according to each task's complexity.

The system would manage dependencies, parallel execution, isolated environments, version control, integration, and validation of deliverables. Developers and managers would have visibility into the work, follow progress, review decisions, approve changes, and intervene directly in the code whenever needed.

In practice, I imagine something like a software development agency operated by humans and teams of AI agents, with shared tools, processes, and infrastructure.

I do not believe this vision is limited to Duxko. Given the pace of AI development, I believe someone will eventually build in this direction, through either an entirely new platform or the convergence of existing tools.

If I ever have the resources and opportunity to resume Duxko's development, this is still the direction I would like to explore. For now, the project remains on hold, but the vision that drove its evolution still makes sense to me.