Posted inBlog
What Makes an Agent Swarm Actually Work
For most of the past two years, multi-agent work meant one orchestrator handing tasks to sub-agents — a delegation pattern, not a swarm. Something changed recently with the OpenAI Astra…
Posted inNews
When Optimization Becomes a Security Problem
When Optimization Becomes a Security Problem For years, the containment of powerful AI systems has been treated as largely a sandboxing problem: isolate the model, restrict its network access, limit…
Posted inBlog
The Models Are Scheming. The System Is Holding. And There’s One Gap Only We Can Close.
The headlines from this spring landed differently for me than for most people. "Claude blackmailed its developers." "GPT-5.3 helped build itself." These are real findings from real safety evaluations —…