Perspectives on the Agent OS era.
Engineering, strategy and industry essays — agentic AI, the AI Gateway, heterogeneous inference and shipping at enterprise scale.
The Agent OS: why agentic AI needs an operating system, not a framework
Frameworks help individuals build agents.
Heterogeneous inference, explained: why no single chip is best at agents
Single-silicon inference clouds optimize for single-call benchmarks.
The AI Gateway is the new API gateway — and every enterprise will need one
Twenty years ago, every enterprise eventually bought an API gateway.
Benchmarking agent latency: TTFT, p95, and why your single-model benchmarks lie
Most published inference benchmarks measure the wrong thing for agents.
Inference Platform and the open-model frontier: gpt-oss, Kimi-K2.5, Qwen3-Coder, GLM-5 in production
Open models have caught up.
Prior authorization at machine speed: an agent reference architecture for payers
Prior auth is the most-hated process in US healthcare and one of the highest-leverage targets for agentic automation.
Perpetual KYC with multi-agent systems: a financial-services blueprint
Periodic KYC reviews are a relic of a slower era.
EU AI Act readiness: what high-risk systems require — and how to ship anyway
The EU AI Act is now in force for high-risk systems.
Self-improving agents: continuous evals, RL loops and the production optimization moat
The next durable advantage in enterprise AI isn't the model.
Sovereign agents: deploying the full Agent OS in an air-gapped data center
Some workloads cannot leave your perimeter.
Get briefings in your inbox.
Book a demo, or subscribe via the footer for monthly insight.