BLOG
Notes on building AI agents that hold up in production.
Engineering decisions, design tradeoffs, and the occasional opinion — from the team building Cortez AI.
From Chat to Live URL: Letting an AI Agent Build and Deploy a Website
Describing a landing page and getting back a live link, in one conversation. Here's how that actually works, and where it's genuinely useful versus where it isn't.
Scoped connector access: why read and send are separate permissions
Giving an agent your Gmail inbox shouldn't mean giving it the power to send from it. We split every connector into read and send permissions, granted per agent.
Using an AI Agent for Customer Support: What Actually Works
Not every support conversation should go to an AI. Here's a realistic breakdown of what to automate, what to escalate, and how to set expectations with your customers.
Why Your AI Agent Needs a Knowledge Base (and How RAG Actually Works)
The single biggest reason AI agents give wrong answers about your business isn't the model — it's that nobody told it anything about your business. Here's the fix.
Retrieval that cites its source, chunk by chunk
An agent that can't tell you where an answer came from isn't trustworthy, it's convincing. Here's how Cortez AI's RAG pipeline keeps every reply traceable back to a real document.
Connectors, Explained: How Your AI Agent Talks to Gmail, Slack, and Everything Else
An agent is only as useful as what it's plugged into. Here's how connectors work, what read vs. send access actually means, and how to think about which ones you need.
What Is an AI Agent, Really? A Plain-English Explanation
Not another chatbot. Here's what actually separates an AI agent from the assistant window you're used to — and why the difference matters for your business.
Why we built Cortez AI around four parts, not one prompt
Most "AI agent" products are a chatbot with a system prompt. We think an agent is a model, a memory, a set of tools, and a channel — and treating it that way changes what you can build.