Apple, Anthropic, and OpenAI all made the same move last week.

Apple, Anthropic, and OpenAI all made the same move last week.

Apple rebuilt Siri around your phone (messages, photos, calendar, email). Not to make Siri smarter. To make Siri useful by giving it access to the context in your life.

Anthropic launched Claude in Slack. Your team tags it into a channel, gives it access to selected tools and data, and it works through tasks in the thread.

OpenAI published a study showing that inside their own company, Codex became the go-to tool for AI-assisted work. Not because it was the smartest model. Because employees trusted it with their actual files.

Three different companies. Three different products. One identical bet: the smartest model doesn't win. The one closest to your data does.

That's not an AI strategy. That's a land grab.

The pitch sounds great. "Let us come to where you already work." But the moment one of these tools starts building context from your Slack history, your codebase, your customer conversations, your switching costs go from zero to painful. Fast.

If Claude has six months of your team's Slack context baked in, moving to a competitor means starting from scratch. If Siri knows your calendar, your contacts, your habits (and it already does), you were locked in before the AI part even started.

This is the oldest play in enterprise software. Make yourself useful, then make yourself hard to leave. CRM vendors did it. Cloud providers did it. Now AI labs are doing it.

The difference? They're not just storing your data. They're learning from it and becoming more useful the longer you stay. That makes every previous lock-in look gentle.

So before you let an AI deeper into your Slack, your files, or your phone (more than you already have), ask one question:

Can I take my context with me when I leave?

If the answer is no (or worse, if you don't know), you're not adopting a tool. You're signing a lease. And the rent's going to go up.