Independent software development project
An independent project exploring more reliable AI-assisted software development through collaboration, technical analysis, implementation, and code review.
Project overview
Software development is increasingly a collaboration between people and AI systems. Today that collaboration is mostly improvised: a developer prompts a model, receives output, and judges by hand whether it is correct. For small, self-contained tasks this works reasonably well. For complex, long-running technical problems it becomes fragile — context is lost, results are hard to verify, and quality depends heavily on how well each individual prompt happens to be written.
Collaborative Brainpower is an independent project exploring a more deliberate version of that collaboration: one where AI assistance is organised around technical analysis, implementation, and review, and where the person directing the work stays responsible for judgement and final decisions.
The goal is not to remove people from the process. It is to make their time count where judgement matters most, while allowing AI systems to take on more of the mechanical and exploratory work in a way that remains verifiable.
Approach
The project is guided by a small set of ideas about how AI-assisted development should work in practice.
01
Exploring ways to coordinate AI-assisted development. Rather than a single prompt-and-response exchange, work is organised into stages with defined inputs and expected outputs, so contributions from different tools and models can be combined coherently.
02
Emphasising code quality, validation, and consistency. Generated work is checked against the stated requirements, reviewed for correctness and maintainability, and held to the same standard as any other contribution.
03
Keeping important decisions under human control. Scope, architecture, trade-offs, and what ultimately ships remain human decisions; AI assistance informs those decisions rather than replacing them.
Project
Collaborative Brainpower (CB) helps developers get more reliable results by having several AI models work together on one software task, with a person in charge.
01
Different AI models contribute to the same piece of work, each in a clear role, so their strengths add up instead of competing.
02
Generated code is checked by a second, independent AI before you rely on it, giving you a clearer picture of what is correct and what needs attention.
03
Changes are only applied after you approve them, and sensitive information is protected from being shared by accident.
CB is built and maintained by Doan Dai Duong, founder and lead developer, in Ninh Binh, Vietnam.
Status
Independently developed. Collaborative Brainpower is an independent software development project. It is not yet incorporated as a company, and it is not affiliated with, endorsed by, or sponsored by any AI provider.
Under active development. The technical foundation is being built and refined, and the approach described on this site is still undergoing validation.
Describing direction, not availability. Nothing here should be read as a product announcement, a release commitment, or a claim that every described idea is complete. This site describes intent and direction.
Contact
The project is open to conversations with potential partners, AI providers, and people interested in where this work is heading.
Please include a short description of your interest and the context you are writing from. This is a small independent project, so replies may take a little time.