#ai-agents

Most MCP tutorials stop at hello-world. Then you try to run one on the open internet with real auth, real load, and real attackers, and every assumption breaks. Here are the five design decisions that actually decide whether your server survives production.

MCP took off faster than anyone's threat model. We wired arbitrary third-party tools straight into models that can't tell an instruction from data, then acted surprised when tool poisoning hit 72.8 percent and Anthropic's own Inspector shipped a CVSS 9.4 RCE. Here's the real attack surface, and what governance is finally starting to look like.

OpenAI shipped three models on July 9 at $5, $2.50, and $1 per million input tokens. The prices are a clean scalar, which means the only question that matters is whether the top tier earns its 5x. For most of your traffic, it doesn't.

Four shapes get thrown around for multi-agent systems: pipeline, supervisor, swarm, and full autonomy. Three of them you probably don't need. Anthropic's own data says most of the win comes from spending more tokens, not from clever coordination, so pick the simplest shape that ships.

The line that clicked for me: behavior belongs in the weights, knowledge belongs in the context. Stop fine-tuning your FAQ into a model. The job that actually moves the needle now is curating the smallest set of high-signal tokens the model sees at inference time.

The cloud round-trip is the bottleneck for agents, not model quality. On-device time-to-first-token runs 15 to 80 ms against 180 to 600 ms in the cloud, and a single agent task chains 10 to 30 model calls. Here's what shipping a local agent on Apple, Qualcomm, and Google silicon actually looks like in 2026.

Your AI agent is a user with production credentials, and most teams treat it like a feature. Nearly 80% of organizations say their agents have already acted beyond their intended scope. Here's how to govern a non-human user like you'd govern a human one: scoped identity, least privilege, quotas, and an audit trail back to a human sponsor.

Claude Dispatch runs desktop tasks from your phone, as long as your computer stays awake. That catch isn’t a beta bug. The model thinks in Anthropic’s cloud; only the hands run on your laptop, and every rival already moved those hands off the desktop.