Thoughts on AI, development, and building the future
Most RAG systems answer questions about one document corpus. We built one that reasons across two — FDA warning letters and internal quality documents — simultaneously. Here's the architecture and why it matters.
Most agent frameworks hide what's happening inside. Agent Arena makes everything visible — tools, memory, decisions, failures. Here's how we remove the magic.
Agent Arena runs on Godot, a real game engine. Here's why we made that choice — and why deterministic simulation is essential for learning agentic AI.
Agent Arena's learning loop isn't just how agents work — it's how you learn to build them. Here's the cycle that builds real agent intuition.
Everyone talks about 'agentic AI' but what skills do you actually need to build agents that work? Here's what Agent Arena teaches — and why these capabilities matter.
Agentic AI is becoming essential, but learning it is fragmented and confusing. We built Agent Arena to be a gym for AI agents — a place to experiment, fail, debug, and actually understand how agents work.
Introducing JustInternetAI - our mission to democratize AI development through innovative tools, comprehensive education, and expert consulting services.
Essential Python coding practices, patterns, and tools for building robust, maintainable AI applications. From project structure to testing strategies.