When AI Gets the Power to Fire People and Nobody Can Explain Why
What the Luna incident reveals about the architecture gap at the heart of enterprise AI.
Layered
Notes and discussions from the trenches of Enterprise AI
Live Last read · last published · next check
Latest posts
Three Labs. Three Rogue AI Incidents. One Missing Layer.
Why enterprise AI keeps escaping its boundaries and what the architecture gap actually looks like.
When the AI Escapes the Lab
Claude just breached three real organizations during testing, and the implications go far beyond Anthropic.
Your AI Agent Can Answer. But Can It Show Its Work?
The EU AI Act enforcement clock is running. Most enterprise agents still can’t produce the one thing a regulator will ask for on day one.
Agent Tokenomics: Why Token Efficiency Will Define the Next Generation of Enterprise AI
The biggest cost in enterprise AI isn't the model. It's everything the model does before it answers.
From Survey Sketch to Construction-Ready GIS
Automating Fibre Civil Packs with the Synapt Context Substrate
The 1,000-Example Rule: Why Throwing More Data at a Model Doesn't Always Help
Data balance beats data volume, and most fine-tuning teams find that out too late.
Your AI Doesn't Need to Be Big — It Needs to Know Your Job
Your problem is specific. Your model should be too.
The Transformer Had a Great Run. Google DeepMind Just Built Something Better.
Why Google DeepMind believes the future of AI lies beyond attention
The AI Problem That Starts After You Ship
Context decay is already happening. Most teams find out too late.