Matt Roberts
Technology Leader · Workspace Geek · AI Nerd
Writing about the future while living in it
// featured
Worth reading
Your AI problem is a data problem, and it started ten years ago
Every AI pilot works beautifully until it touches the real data estate. The constraint isn't the model, it's the twenty years of permissions sprawl, uncatalogued documents and archived applications nobody planned to read again.
The 90-minute ultimatum that took the world's most powerful AI offline
On Friday 12 June, Anthropic had 90 minutes to fix a security flaw in Claude Fable 5 or take it offline. By 10pm it was gone. It came back this morning. The 19 days in between are worth understanding.
Building my first AI-powered app: what I learned as a non-ML developer
I built this website using Next.js, AWS, and Claude. I'm not a developer by trade. Here's an honest account of what that process was like and what surprised me.
AI Perspectives
all →Your AI problem is a data problem, and it started ten years ago
Every AI pilot works beautifully until it touches the real data estate. The constraint isn't the model, it's the twenty years of permissions sprawl, uncatalogued documents and archived applications nobody planned to read again.
You've built AI agents on Azure. Has anyone checked who else can use them?
Agentic AI and MCP integrations have introduced an attack surface that most Azure security models weren't designed for. CVEs are appearing, real exploits have hit production M365 environments, and the deployment conversation has moved considerably faster than the security one.
The 90-minute ultimatum that took the world's most powerful AI offline
On Friday 12 June, Anthropic had 90 minutes to fix a security flaw in Claude Fable 5 or take it offline. By 10pm it was gone. It came back this morning. The 19 days in between are worth understanding.
Microsoft 365 & IT
all →// latest
Recent posts
I spent twenty years complaining about this problem. Then I built the thing.
Every vendor I have worked with ran its partner programme on spreadsheets and email. I complained about that for two decades, then built the alternative on my own. What AI did and did not do for that is worth being precise about.
Your AI problem is a data problem, and it started ten years ago
Every AI pilot works beautifully until it touches the real data estate. The constraint isn't the model, it's the twenty years of permissions sprawl, uncatalogued documents and archived applications nobody planned to read again.
You've built AI agents on Azure. Has anyone checked who else can use them?
Agentic AI and MCP integrations have introduced an attack surface that most Azure security models weren't designed for. CVEs are appearing, real exploits have hit production M365 environments, and the deployment conversation has moved considerably faster than the security one.
The 90-minute ultimatum that took the world's most powerful AI offline
On Friday 12 June, Anthropic had 90 minutes to fix a security flaw in Claude Fable 5 or take it offline. By 10pm it was gone. It came back this morning. The 19 days in between are worth understanding.
The model running your Copilot demo might not be OpenAI
Anthropic's Claude is now the default model inside Microsoft 365 Copilot for Excel and PowerPoint. The product you've been selling as Microsoft AI includes a model from a company that isn't Microsoft. The channel pitch needs updating.
SpaceX goes public. The pitch isn't what you think.
SpaceX hit the Nasdaq today as the largest IPO in history. You'd be forgiven for thinking it's a rocket company. A significant chunk of the investment case is actually about building AI data centres in space. Yes, in space.
// subscribe
Stay in the loop
New posts on AI, Microsoft 365, and the IT Channel — no noise, no spam.