The Brief · by KniteSpace

AI, explained. Not sold.

A no-hype read on what's actually happening in AI — the real developments, the real sources, and what they mean. For people who'd rather understand it than be pitched it.

Issue · Updated regularly

This week in AI

Five things worth knowing, with sources you can check yourself.

Models

Microsoft unveils its own "MAI" models.

At Build, Microsoft introduced a family of seven in-house models — led by a reasoning model with a 256K-token context window — and loosened parts of its OpenAI agreement. A sign that the big cloud providers want their own model independence.

Jun 2, 2026CNBC →
Policy

The EU agrees to simplify and delay parts of the AI Act.

A provisional "Digital Omnibus" deal extends compliance deadlines for high-risk AI systems (into 2027–2028), eases obligations for smaller firms, and adds new prohibitions. The core rules stand — this is a timeline shift, not a repeal.

May 7, 2026EU Council →
Tools

Autonomous AI agents reach general availability.

Microsoft's Agent 365 and Salesforce's Agentforce orchestration moved from preview to GA this spring — AI shifting from chat assistants to agents that take multi-step actions inside business systems, raising fresh questions about oversight and governance.

May–Jun 2026Microsoft →
Adoption

Adoption is near-universal. Returns aren't.

2026 surveys converge: ~91% of businesses use AI in some form, but only around a third report significant ROI, and most still aren't using it meaningfully. The differentiator has moved from having AI to actually operationalizing it.

Labor

AI's most-exposed workers? The highly paid ones.

Anthropic's labor-market research, using real usage data, finds today's most AI-exposed workers tend to be educated, experienced, and well-compensated — with no clear unemployment spike yet, though entry-level hiring in exposed roles is the area to watch.

Context

What ties these together.

The tools are getting cheaper and more autonomous, the rules are settling, and the data keeps saying the same thing: the gap isn't access to AI anymore — it's the know-how to turn it into results. That's the whole story of this moment.

The BriefKniteSpace