Key Takeaways
- Google's AI Mode now connects to Instacart, Canva, and YouTube — three apps, two years after ChatGPT plugins launched
- The move is defensive: Google is playing catch-up to OpenAI and Anthropic on app integrations
- Google's real leverage isn't third-party apps — it's the data empire it already owns (Gmail, Photos, Maps, YouTube)
- "Personal Intelligence" plus app actions could make AI Mode the first assistant that actually knows you, if Google stops hedging
Google announced Thursday that AI Mode can now link to Instacart, Canva, and YouTube. Three apps. That is the headline. The subtext is more revealing: Google, the company that organized the world's information, is still building the plumbing to act on it.
AI Mode launched in early 2025 as a conversational search layer. It answered questions. Now it can add groceries to an Instacart cart, surface Canva templates, save a playlist to YouTube Music. Useful? Yes. Groundbreaking? Only if you ignored the past two years. ChatGPT plugins arrived in March 2023. Claude followed with tool use. Google's "expansion beyond answering questions" is a polite way of admitting it fell behind on the obvious next step.
The examples Google chose tell the story. A barbecue grocery list. A flyer design. A party playlist. These are consumer toys — low stakes, high demo value. They avoid the messy reality that most productivity happens inside enterprise SaaS: Salesforce, Jira, Notion, Slack. Google mentioned "working with a range of partners" and "more apps soon." Vague promises are the standard currency of platform keynotes.
What makes this potentially different from ChatGPT's plugin graveyard is the data moat Google already occupies. "Personal Intelligence" — rolled out earlier this year — lets AI Mode read your Gmail and Google Photos. That means it knows your flight confirmations, your receipts, your kids' birthdays, the restaurant you liked in Lisbon three years ago. Pair that with the ability to check local store inventory (added recently) and explore the web side-by-side (also recent), and you get something neither OpenAI nor Anthropic can offer: an assistant with context that doesn't require you to paste it in every session.
But Google hedges. It keeps AI Mode separate from the Gemini app, which also connects to third-party apps — Canva, OpenTable, Spark, Instacart — via a capability launched at I/O. Two surfaces, overlapping integrations, fragmented strategy. The left hand doesn't know what the right hand is shipping. That confusion costs developers and users alike.
The competitive pressure is real. OpenAI's ChatGPT now has a desktop app that sees your screen. Anthropic's Claude can control your computer. Google's counter is search distribution — AI Mode lives where billions already start their queries. But distribution without depth is a trap. If AI Mode becomes a better "I'm feeling lucky" button that occasionally orders groceries, it wins the demo but loses the war.
The skepticism is warranted because Google has abandoned ambitious assistant plays before. Google Now. Google Assistant's "actions." Duplex. Each promised a layer of agency atop knowledge. Each retreated to voice timers and weather cards. AI Mode feels like the latest attempt to climb the same hill.
What would prove this time is different? Depth. Not three launch partners — fifty. Not consumer toys — enterprise workflows. Not "rolling out in the U.S." — global parity. And a single, coherent surface instead of the Gemini/AI Mode split that forces users to guess which Google brain holds which capability.
The pieces are there. The knowledge graph. The personal corpus. The search intent signal. The Android foothold. The YouTube empire. The Maps location intelligence. No rival holds that full stack. But a stack unexposed is a stack wasted. Thursday's announcement is a toe in the water. The question is whether Google finally decides to swim.