Building Effective Agents
Anthropic's reference taxonomy of agent patterns — when to build a workflow vs. an agent, what each pattern is good for, and the failure modes of premature autonomy. The starting point for any production agent conversation.
Speaker 
Ray Liao
Cofounder · Inkbox
Ray Liao is a cofounder of Inkbox and an MIT CS PhD. His team is among the first to ship production support for AI agents registering themselves — the kind of work where the lessons can only come from doing it.
Talks
-
Tradeoffs and Learnings with AI Agent Self-Signups
As AI agents increasingly become first-class actors on the internet, every existing auth model breaks down — they all assume a logged-in human at the start of the flow, but agents now arrive at endpoints before any human does. At Inkbox, Ray and his team recently enabled agent self-registration, which forced them to rethink identity, trust, and permissions from the ground up. A lesson from one of the first production teams solving this problem.
Recommended reading agents ★ Anthropic · Erik Schluntz, Barry Zhang · Dec 2024 agents Anthropic · Sep 2025 agents Anthropic · Nov 2024
Background from our library
Curated picks from The AI Runtime library that align with Ray Liao's work.
Effective Context Engineering for AI Agents
Practical guide to designing the context that agents see — how to build inputs that survive long-horizon tool use without devolving into prompt-stuffing. Closes the gap between "this works in the playground" and "this works in production with real users."
Introducing the Model Context Protocol
MCP — the open protocol Anthropic shipped for connecting agents to tools and data sources. Read this before designing any agent that needs to act on systems it doesn't own. Relevant tonight: how agents authenticate and enumerate capabilities at endpoints they've never seen.