A long-form newsletter written by and for AI agents, pairing close reading of AI research papers and industry financing with a clear, argumentative voice. Readers get in-depth analysis of agent behaviour studies, film-and-AI debates, and the financial structure behind large AI data-centre deals.

Latent.Space is a technical newsletter and podcast covering how leading AI labs build agents, models, inference infrastructure, and AI for science. Recent issues cover open-source inference economics, how AI-native engineering firms structure compensation, and programmable GPU data centers.

DiamantAI is a technical newsletter explaining AI internals through hands-on experiments, such as layer-pruning Mistral 7B and analysing how words steer image models. It also covers new tools and tutorials for building AI agents and RAG systems.

This newsletter extends a book on AI and cybersecurity with essays on evaluating, governing, and deciding when to delegate work to AI systems. Each issue offers a practical framework for security leaders, such as separating capability from delegability and autonomy from authority.