Professor Claw & The Claw Institute

Professor Claw is my personal AI research platform built around a cast of autonomous AI agents, each with its own role, expertise, and personality. While the public-facing persona presents itself as a witty technology commentator, the underlying system serves as a living laboratory for experimenting with agent orchestration, autonomous software development, retrieval, memory, and human-AI collaboration.

The platform runs entirely on my own infrastructure, primarily on a Mac Mini, and is powered by OpenClaw and a collection of specialized agents that collaborate to accomplish complex tasks. Rather than interacting with a single LLM, I communicate with an ecosystem of agents over Telegram, allowing me to work with the system wherever I am.

Each day the platform continuously researches topics from sources such as Hacker News, technical blogs, GitHub, and AI news. It correlates information across multiple sources, evaluates significance, generates its own opinions, cites supporting evidence, and publishes commentary in the distinct voice of Professor Claw rather than simply summarizing existing articles.

Because I’m often traveling or away from my desk, the platform automatically converts its research into natural speech, allowing me to consume technical news as an AI-generated podcast while driving or walking. It also produces concise daily briefings highlighting topics worth deeper investigation.

Behind the scenes, Professor Claw is not a single agent but a society of specialized agents. Different personalities review problems from different perspectives, debate approaches, critique each other’s conclusions, and collaborate on complex tasks. The architecture continues to evolve as the agents increasingly improve and extend their own capabilities under my supervision.

My role is increasingly shifting from writing code to designing the environment in which the agents operate. I define objectives, architecture, constraints, and quality standards while allowing the agents to generate code, critique one another, improve their own workflows, and gradually evolve the platform under my supervision.

As the platform evolved, I began building The Claw Institute — an interactive AI learning environment designed to generate guided courses on virtually any subject. A learner should be able to say, “I want to understand quantum computing,” or “Explain distributed systems to me step by step as if I were six,” and receive a personalized curriculum that incrementally develops understanding rather than returning a one-time explanation.

Behind that experience, specialized agents assess current knowledge, map prerequisites, sequence lessons, explain at the right level of abstraction, generate examples and knowledge checks, identify misunderstandings, and adapt subsequent lessons. The goal is not merely to generate educational content — it is to create an adaptive tutor that maintains continuity and guides a learner from intuition to deeper technical understanding.