À PROPOS DE CET ÉPISODE
An AI agent invented a job candidate and emailed the offer to his sales team. Nobody asked it to. Alex Salazar found out when someone replied — and raised $60M to make sure it cannot happen to you.
Alex Salazar is co-founder and CEO of Arcade.dev. He built developer authentication once before — Stormpath, acquired by Okta — and is now building the authorisation layer for AI agents.
His case is blunt: agents are not failing because the models are wrong. They fail because nobody can prove who was allowed to do what. Gartner expects over 40 per cent of agentic AI projects to be cancelled by the end of 2027, inadequate risk controls among the reasons.
What we cover:
- Identity is easy. Authorisation is the wall.
- The three A's, and why authorisation is rocket science
- Does the intern get to see the CEO's pay? The question that kills enterprise AI projects
- Why enterprises are quietly banning coding agents
- Guardrails are not governance
- Walling European data off from the US and China
- On the record: within 24 months, the browser is dead
Hosted by Amer Altaf — Founder & CEO of Arkava, Managing Editor of The Control Layer.
📺 Watch: https://youtu.be/jbWHWnZgdLA
📰 Companion analysis: https://thecontrollayer.arkava.ai
