#llm

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.

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.

Your support bot does not need a PhD. Most production AI work is repetitive, narrow, and latency-sensitive, which is exactly where a 3B to 14B parameter model wins on cost, speed, and privacy. NVIDIA Research now argues the same thing in a paper, and the pricing backs it up.

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 demo is easy. streamText, a text box, tokens on the screen, and it feels done. What breaks in production is retries, cost control, silent errors, and the one user who paste-bombs your context window. Here's the checklist I wish I had before shipping.

Prompt injection is the SQL injection of the AI era, except worse. SQLi has a clean fix. Indirect prompt injection doesn't, because an LLM reads instructions and untrusted data through the same channel with no reliable way to tell them apart. OWASP ranks it the number one risk in AI applications, and the numbers back that up.

'AI engineer' isn't machine learning wearing a new hat. It's a different job that starts from a product and reaches for a model, not the other way around. Most job posts are confused about which one they're hiring for, and that confusion is where your leverage is.

Beyond basic prompting: system prompt architecture, structured outputs, RAG patterns, LLM-as-judge evaluation, and cost optimization for production AI.

GPUs get the headlines, but inference serving, networking, and observability are where AI systems actually win or lose. Here's what the stack looks like now.