Archos LabsThe Translation Layer
Where AI programs break.
Researched by Metis — an AI research agent I built — and reviewed by me before anything ships, against 25 years of breaking these programs from the inside. Most AI newsletters are prompted. This one is engineered.
Data as a Decision Infrastructure
More Data Made You Worse At This
Executives who receive more data without learning to question it make worse decisions. Here's the judgment framework that actually closes the gap.
3 min read · Published
Human-Centered Transformation
AI Training That Measures The Wrong Thing
Most AI training tracks completion, not capability. Here's how L&D leaders can measure workflow adoption and decision quality instead.
3 min read · Published
AI as Strategy
AI Risk Taxonomy Every Board Needs Now
Generative AI creates 12 risk categories your ERM register was never built to catch. Here is the practical AI risk taxonomy boards need to govern model and…
4 min read · Published
AI as Strategy
AI Agent Costs At Startup Scale
AI agent deployment at startup scale costs more than vendors show. Get the full cost breakdown covering build fees, recurring costs, and data prep before you…
3 min read · Published
Human-Centered Transformation
Why Your Team Avoids Your AI Agent
When your team works around your AI agent instead of with it, that's diagnostic data. Here's how to use it before the workarounds become permanent.
3 min read · Published
The Execution Layer
Why Your Data Model Is Breaking Your AI Agent
Engineers built most AI agent data layers for dashboards, not agents. This guide covers entity resolution and event tables for solo engineers.
4 min read · Published
Data as a Decision Infrastructure
Your AI Agent Is Lying Because Your Data Is Broken
Before you build an AI agent, audit the data it will actually use. Four failure modes founders miss and the checks that catch them before the rebuild.
3 min read · Published
AI as Strategy
Your AI Tool Isn't Broken. Your Data Is
Switching AI platforms won't fix bad output if your data is the problem. Here's how founders can tell the difference between model failure and data failure.
3 min read · Published
AI as Strategy
Why AI Tools Don't Build Capability On Their Own
Most organisations own AI tools they can't use well. We name the 4 non-technology investments that build durable AI capability and the order to make them.
3 min read · Published
The Execution Layer
MIT AI research: Vendor vs Internal Builds (2025)
MIT NANDA found vendor-led AI projects succeed 67% of the time vs 33% for internal builds. Here's when that gap is real and when it isn't.
4 min read · Published
