Here’s what we’re reading and thinking about in the news this week:

AI Stacks Making Waves 🌊 

Everyone's asking "what AI tools should I use?" But the real question is: what kind of AI stack matches your team?

Google recently baked Gemini into every Workspace plan. No add-on. No opt-in. AI is just the default now for 3 billion users. They even started calling reusable automations "skills." Sound familiar?

But here's the thing. Google made one decision for a lot of people. And it's surfacing a question that every team is quietly trying to answer on their own.

3 AI stacks are emerging

We're watching this play out across every company we work with.

Three patterns, three very different bets.

Bet #1: The enterprise default. 

Your team already lives in Google or Microsoft. AI just showed up inside the tools you were already using. For a lot of teams, this is honestly fine. The learning curve is almost zero because the interface didn't change. The AI just... appeared.

Bet #2: The modern-native stack. 

Startups and leaner teams going all-in on tools like Notion or ClickUp that have been aggressively built up with AI baked in. These have tighter feedback loops and more flexibility. But you're betting on that platform's roadmap. Worth considering.

Bet #3: The roll-your-own crowd. 

This one's new. Non-technical operators are vibe coding their own AI operations systems with tools like Claude Code and Cursor. A year ago this was a developer-only move. It's not anymore. Regular business operators are building custom automations that fit their exact process.

Our experience: we tried option 3. Here's what happened.

Early on, we built our own AI operations stack from scratch.

Custom workflows, custom integrations, everything tailored to exactly how we work. And honestly? It gave us a real advantage. We were moving faster than teams using off-the-shelf tools because our AI did things those tools couldn't do yet.

Then the market caught up.

Features we'd spent weeks tuning started showing up as default capabilities inside the platforms everyone already used. The edge disappeared.

That's not a reason to avoid building your own. But it is a reason to build with the expectation that you'll probably switch.

What we believe even more now is that the stack is temporary.

The playbook is the asset (shocker, we know!)

Here's what actually survived our platform switches: the documented processes.

The playbooks that described how work should get done, what good output looks like, and what steps to follow.

Those transferred perfectly from our custom stack to the next tool. Because they lived in documents we owned, not inside a platform we rented.

Just another example of the teams that will win this aren't the ones who pick the "right" AI stack. They're the ones who document their playbooks and build a system in a format that outlives whatever tool they're using today.

3 questions to ask your team this week

1. Where does your team already live? If everyone's in Google all day, fighting that to adopt a shiny new tool creates friction that kills adoption. Match the tool to the existing habits.

2. What's your team's appetite for learning new things? Be honest. A powerful tool nobody uses is worse than a simple tool everyone uses. The lowest learning curve wins.

3. Is your AI ops system documented somewhere you control? If your entire AI workflow lives inside one platform and nowhere else, you don't own it. Write it down in a doc. Make it portable.

There's no wrong answer on which stack to pick. But there is a wrong way to pick one: chasing what's trending instead of matching what fits.

Which camp are you in: enterprise default, modern native, or rolling your own? Hit reply and tell us. We're genuinely curious how this is shaking out.

LINKS

For your reading list 📚

• Anthropic and OpenAI both launched billion-dollar joint ventures with Wall Street firms this week. Their goal? Get AI embedded inside portfolio companies before IPO season. AI labs are becoming consulting firms now.

• OpenAI is reportedly building a phone where AI agents replace apps entirely. No home screen. No app grid. You just tell it what you want. Mass production in 2028, allegedly.

• The Pentagon signed AI deals with 8 tech companies and blacklisted Anthropic as a "supply chain risk" for refusing to drop safety guardrails on weapons. But recently… a federal judge blocked it. So this story is far from over.

• Nearly 80,000 tech workers were laid off in Q1 2026, with almost half of cuts explicitly citing AI. Meanwhile, researchers are warning this could be an "automation trap" that hurts the companies doing the cutting.

That's all!

We'll see you again soon. Thoughts, feedback and questions are much appreciated - respond here or shoot us a note at [email protected]

Cheers,

🪄 The AMP Team (formerly: the AI Exchange Team)