Welcome to Manzia Research

Manzia is an information portal focused on Trusted Agents — the research, design patterns, deployment playbooks, and evaluation methods that make AI agents safe to delegate real work to.

What we publish

• Research synthesizing the state of agent trust.
• Design notes on guardrails and UX patterns.
• Deploy field reports from production.
• Evaluate and Benchmark writeups on measurement.
• The Agent Trust Framework itself, kept open at Manzia/agent-trust-framework.

How we publish

Authors commit markdown to this repository. CI validates frontmatter and cross-links; merging to main ships the article via S3 and an event- driven processor that updates the live site without a redeploy.