APIs Designed by Agents: The Architecture Behind an AI-First API Design System

A behind-the-scenes look at how we built an agentic AI system that designs, validates, and tests API specifications. Learn the architecture, guardrails, workflows, and lessons that make API design an interesting real-world use cases for agentic AI.

Key Takeaways:
•  Architecture of an AI-first, agentic API design system
•  How agents use executable contracts for grounding and correctness
•  Turning prompts into deterministic artefacts: schemas, examples, tests and MCP server
•  Guardrail patterns that prevent schema drift and hallucination
•  Designing agent workflows that handle ambiguity and human-in-the-loop review
•  A replicable framework for building safe, reliable agentic engineering tools

What will the attendees learn:
How to combine LLM reasoning, policy-driven guardrails, executable API contracts, and deterministic workflows to build an agent that reliably produces high-quality specs without hallucinations.