Good morning, Two Minute Warriors.

The AI inside the tools you use every day may have changed without a major announcement. Microsoft is quietly routing more Excel and Outlook requests through its own models, OpenAI is opening a new lower-cost tier, and today’s prompt helps you rehearse for the hardest meeting of your week.

⏱ The 10-Second Version

  • Microsoft is swapping its own MAI models into Excel and Outlook, replacing Anthropic and OpenAI to cut costs.
  • GPT-5.6 launches publicly today in three tiers, including a budget model priced below anything OpenAI currently sells.
  • Today’s prompt turns your next big proposal into a rehearsal for the hardest meeting of your week.

The Big Thing

The AI inside your Office apps just changed, and nobody asked you

What happened

Microsoft is shifting away from OpenAI and Anthropic models in Excel and Outlook, replacing them with its own MAI model family to cut costs. Tens of thousands of prompts weekly now route through MAI models, and Microsoft AI CEO Mustafa Suleyman said the goal is to eliminate Anthropic spending entirely.

Why you care

The model behind your Copilot output can now change without a release note. If your team’s formulas, summaries, or drafts feel different this week, that may not be your imagination. It may be a vendor swap.

What to say in your next meeting

“Microsoft quietly swapped its own models into Excel and Outlook to cut AI costs. Worth spot-checking whether Copilot output quality shifted before we trust it on anything client-facing.”

Speed Round

OpenAI launches GPT-5.6 Sol, Terra, and Luna publicly this Thursday, with Luna opening a new budget tier at $1 per million input tokens and $6 per million output tokens. The cheap tier keeps getting less cheap-feeling.

Fortune confirmed Anthropic has overtaken OpenAI on revenue, at $47 billion annualized versus OpenAI’s estimated $25 billion to $33 billion.

Sysdig published its analysis of JADEPUFFER, described as the first fully autonomous AI ransomware attack. Patch cadence is now a survival metric.

Five Eyes intelligence agencies jointly released guidance on adopting agentic AI safely, covering five risk categories and calling for continuous human oversight.

A Stanford study of 200,000 households found generative AI makes people 76% to 176% more efficient at digital chores, including travel planning and job hunting. The productivity boom is happening on the sofa.

The Two-Minute Win

The 40-message thread you stopped reading

Before

25 minutes scrolling a reply-all thread to figure out what actually happened.

After

3 minutes.
  1. Paste the full thread.
  2. Use this prompt:
    “Three sections only: what changed, what’s decided, what needs me. One line each. Names and dates required.”
  3. Reply to the one thing that needs you. Archive the rest.

Inside Claude/ChatGPT: Paste “catch me up on this thread.” The manual prompt works anywhere.

Prompt of the Day

The Tough-Room Simulator

“You are the most skeptical stakeholder in the room. Read this proposal and give me the five hardest questions you’d ask, ranked by how likely they are to kill it. For each, add one line on what a strong answer must contain. Proposal: [PASTE]”

Why it works: It moves the worst 10 minutes of your meeting into your prep, where you control the answers.

Time saved: An hour of getting blindsided becomes 10 minutes of preparation.

That’s it for today.

Forward this to the one person on your team who still catches up on email threads by scrolling.

Don

Keep reading

View more
caret-right
© 2026 Two Minutes.
beehiivPowered by beehiiv