Tag:ai
All the articles with the tag "ai".
What I Got Wrong About Letting an AI Agent Write SQL
Published: at 11:00 AM (13 min read)Five days after I published a design for giving an agent governed SQL access, I threw most of it away. What made me stop, what the first end-to-end test of the replacement showed, and the rules I would follow next time.
Your AI Agent Can Write SQL. That Was the Easy Part.
Published: at 11:00 AM (11 min read)Letting an agent query a multi-tenant ERP raises two questions that look like one. "May it see this row?" belongs to the database. "Does this number mean what the user thinks?" belongs to something else. Both fail silently.
When AI Ships Faster Than Humans Can Review
Published: at 02:45 AM (7 min read)AI moved our bottleneck from producing code to understanding what happens after it ships. Here is the feedback control loop we built in Ezzo to close that gap.
AWS Is Not Your Agent's Shell
Published: at 11:00 AM (14 min read)Coding agents make change cheap. On AWS, the real platform challenge is making every change scoped, reviewable, observable, and recoverable.
Archmesh: A Trust Layer for Multi-Agent Architecture Reviews
Published: at 12:00 PM (10 min read)Everyone is building multi-agent systems. Archmesh is the trust layer that makes architecture reviews predictable, auditable, and safe by default.