Copilot & AI at Work

Copilot & AI at Work

Home
Nederlands
Podcast
Archive
About

The Invisible Conversation: System Prompts, User Prompts, and Copilot’s Guardrails

Exploring how Microsoft 365 Copilot orchestrates invisible logic through system prompts, functions, and GenAI-based risk classification.
READ THE LATEST

Recent posts

View all
Microsoft Frontier: Your way into Copilot's early access program
From confusion to clarity (well, mostly)
 • 
Hakim van der Maas
The Scrutiny of Copilot’s Guardrails: Intent Detection, Prompt Shields, and risk classification
Hypotheses and insights into how Microsoft 365 Copilot classifies the user prompt on the risk and enforces security & safety controls, without relying…
 • 
Robbert Berghuis
Microsoft's Copilot Fall Release: Collaboration, Mico and long-term memory
We tested the new features early and watched them evolve from buggy previews to polished announcements. Here's what actually matters.
 • 
Danny de Vries
, 
Hakim van der Maas
, and 
Robbert Berghuis
When your AI vendor asks you to click "I agree" on someone else's terms
The Anthropic integration raised a question we weren't expecting: who exactly are we doing business with?
 • 
Danny de Vries
From prompt to response: the Copilot architecture
How we used Copilot to uncover its own architecture, test hypotheses, and iterate toward a >90% suitable code base. Did you know Copilot can hand-off to…
 • 
Robbert Berghuis
Exposing the inner-workings of M365 Copilot Memory
How does it really decide what to store? How does that work? What are the function's instructions? We'll dive into how this works and provide you with…
 • 
Robbert Berghuis
Top posts
Podcasts
Logo for Podcast
Podcast
Coming soon
© 2025 Danny de Vries, Hakim van der Maas, and Robbert Berghuis
Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture