Cirrusly Clever exists because I wanted to learn by building, and by comparing.
Rather than treating AI systems as black boxes, I've been exploring large language models by putting them side by side and asking the same questions of each. Many of the projects here are built specifically to compare different providers (OpenAI, Anthropic, Google Gemini) and different classes of models within them, not just for correctness, but for tone, behavior, latency, and failure modes. Building small, focused systems is the best way for me to learn, play with, and show where meaningful differences actually exist and where they don't.
The projects on this site: Cloud Sayings, Cloud Challenge, Operation Black Swan, and Assay, are deliberately varied in form but consistent in intent. Each is an experiment in learning, system design, and human–AI interaction. Some explore how models respond to identical prompts, others examine what it's like to build software with AI as a co-developer, and others test how people reason under constraints, uncertainty, and probabilistic events. None are meant to be polished products; they are working artifacts designed to be used, inspected, and learned from.
A core principle across all of them is open engagement. Wherever possible, these projects invite public interaction, through voting, gameplay, shared artifacts, public documents, or open source code. Learning accelerates when systems are exercised by real users and when assumptions are exposed to scrutiny. The guiding philosophy is simple: Serious Learning Through Playful Experimentation. Cirrusly Clever is a home for those experiments, and a record of what they reveal over time.