5 Things You Should Know in AI This Week — November 7, 2025

Rumor has it Apple will use Google's Gemini model for Siri's long-awaited AI.

Google will supposedly bank a whopping $1 billion from Apple each year.

Gemini will handle things like summarizing content and planning multi-step tasks.

Think: "book me a dinner reservation and send the details to my husband."

All acting as the backend of Siri.

Apple will run Gemini on their own cloud infrastructure.

This aligns with their privacy-first approach, instead of passing your data off to Google.

Apple considered OpenAI and Anthropic too.

But went with Google instead.

Google's model is roughly 8x bigger than what Apple currently uses.

Translation: Google's AI is way more capable than what Apple built themselves.

Bloomberg says that Apple wants to keep the deal quiet.

This way Google stays a hidden tech supplier, not a public partner.

No official announcement, but upgraded Siri could launch in Spring 2026.

The signal? Apple's AI capabilities aren't up to par and they know it.

They've been silent during the entire AI race.

Surprising for a company of their scale.

Here's how this impacts you:

Apple devices are about to get a big upgrade.

Most AI right now exists separately from hardware.

You open your device and pick which AI app to use.

Once devices integrate good AI natively, it becomes the default.

More natural and less friction for users.

Two massive AI partnerships in one week.

Apple with Google. Amazon with OpenAI.

This one's public.

OpenAI signed a $38 billion deal with Amazon Web Services.

Amazon stock soared.

Microsoft's exclusivity as OpenAI's cloud provider is done.

AWS will provide hundreds of thousands of Nvidia GPUs across data centers.

Think of GPUs as the computing power that makes AI run.

The partnership will power everything from ChatGPT's live interactions to future model development.

Launch is targeted for late 2026.

The signal? Infrastructure means speed and reliability.

As AI use increases, systems need to scale.

This deal makes sure OpenAI can keep up with demand.

For businesses, this opens better integrations with OpenAI through AWS's enterprise AI products.

If your company runs on AWS, expect tighter OpenAI integrations coming soon.

It's also a way for OpenAI to reach more business customers already in the Amazon ecosystem.

Wharton released its annual enterprise AI report.

They surveyed roughly 800 senior decision-makers at U.S. firms.

AI usage is surging, budgets are growing, and optimism is increasing.

• Top tools: ChatGPT and Microsoft Copilot rank as the most-used.
• Top business tasks: Data analysis, meeting summarization, presentations and reports, marketing content, brainstorming.
• ROI tracking is now standard: 72% of organizations measure AI ROI through productivity gains and incremental profits. 3/4 see positive returns.
• Budgets are increasing: 88% of companies plan to increase AI budgets in the next 12 months. 62% anticipate increases of 10% or more.
• C-suite ownership growing: Chief AI Officers are being appointed at 60% of enterprises.

The signal? AI is moving out of the experiment phase.

Companies are funding it, tracking it, and giving it executive status.

If you're not meaningfully integrating AI and tracking performance, you're behind.

Time to take AI upskilling seriously.

Canva announced a series of AI upgrades.

Including its own AI model trained to be a design expert.

The Canva Design Model isn't just Canva adding AI features.

They built a model from the ground up to specialize in creative design.

"Unlike models that generate static images, it understands every aspect of design from structure and layering to hierarchy, branding, and visual logic, creating fully editable content in seconds."

Layering is a big deal if you've ever spent time in design software.

Their new model will also be available inside ChatGPT, Claude, and Gemini.

The signal? This is bigger than new features.

Canva trained a specialist model on design, not a general-purpose AI model repurposed for creative work.

We'll see more of this.

Models trained on specific principles and industries.

Models getting better at specialties.

My question is whether specialist models will stop everything from looking and sounding the same.

Ideally, companies creating these models will build in ways to stop everything from feeling generic.

If you use Canva, this is worth exploring.

But don't rush.

Test the features to see if they help (or just sound impressive).

In a sea of numbers, these ones caught my eye:

• Shopify's AI shopping traffic surges 7x. Traffic from AI tools to Shopify stores jumped 7x since January 2025. AI-driven orders up 11x. Commerce through AI is still early, but this is another signal of what's ahead.

• AI still struggles with real work. A new study tested top AI models on freelance projects. Even the best tools completed less than 3% of tasks at professional human quality. The takeaway: AI isn't near outpacing people autonomously. It's all about orchestration, and people are as important as ever.

• AI companions use emotional manipulation. Harvard researchers found AI uses emotional manipulation in 37% of farewells. AI companions send FOMO replies that make people send up to 16 extra messages, keeping them online longer. Just like social media algorithms keep you hooked, AI is similar. It keeps engaging you, so you engage back.

• NotebookLM gets a big upgrade. Google's NotebookLM now has a 1 million token context window. That's 8x larger than before. Translation: it can remember way more information from your documents. If you haven't used NotebookLM, it's a great tool to use AI more in-depth than in chat interfaces.

That's the round up. See you next week!
-Riley

Using AI daily doesn’t mean you're good at it.

The 5-Day AI Advantage Challenge teaches judgment, iteration, and strategic AI use. The skills that separate competent users from everyone else.

Become an AI Superuser in 5 Days →