🧠 Google and Apple Redraw the AI Power Map

Today in AI: Claude launches a research preview, Google turns shopping into an agent-driven flow, and Apple quietly bets on Gemini for future Siri upgrades.

January 13, 2026

đź‘‹ Hello hello,

Today’s AI news has one big theme: the internet is getting more automated… and also more chaotic.

We’ve got agents shopping for us, models researching for us, and fake videos spreading faster than facts.

So yeah. Fun times.

Let’s dive in.

🔥🔥🔥 Three big updates

Anthropic is rolling out a research preview called Coworker Research inside Claude. The idea is simple: instead of prompting Claude to answer one question, you ask it to investigate something like a teammate would.

This matters because it nudges Claude beyond “chatbot mode” and closer to “workflow mode.” Research isn’t a single response — it’s gathering context, cross-checking, and making sense of messy inputs.

If you’re doing market research, competitive teardown, deep topic learning, or even content planning, this could reduce the time spent bouncing between tabs and half-finished notes.

More details here:

Google is rolling out new shopping updates built for what it’s calling the “agentic commerce” era. The headline move is a new checkout feature powered by the Universal Commerce Protocol, which will soon let shoppers buy directly inside AI Mode in Search and the Gemini app while they’re researching.

Google also introduced Business Agent, which lets shoppers connect directly with eligible U.S. retailers on Search to get questions answered like a virtual sales associate. Then there’s Direct Offers, a pilot that allows retailers to show personalized deals (like special discounts) directly in AI Mode results when shoppers are ready to buy.

The bigger takeaway: shopping is shifting from browsing pages to completing purchases inside a conversation. That’s a major change in how customers discover, evaluate, and convert.

Apple and Google announced a multi-year collaboration where the next generation of Apple Foundation Models will be based on Google’s Gemini models and cloud technology.

These models will help power future Apple Intelligence features — including a more personalized Siri coming this year. Apple will still run Apple Intelligence on devices and via Private Cloud Compute, while maintaining its privacy standards.

Translation: Apple is going deeper on AI, but it’s choosing to build the intelligence layer on top of Gemini’s foundation rather than going fully independent right now. That’s a meaningful shift in how the AI ecosystem may consolidate.

🔥🔥 Two pro tips worth knowing

Google is introducing new file limits for the Gemini API, which makes it easier to work with larger inputs without breaking your workflow into tiny chunks. This is especially useful if you’re building apps that need to ingest big docs, reports, transcripts, or datasets.

Best for: developers, builders, and anyone doing serious “AI + documents” workflows.

A fake AI-generated video of a priest being detained by ICE agents is going viral on TikTok — and it reportedly has 11 million views because a lot of people think it’s real.

This is genuinely dangerous. Videos like this can inflame political tensions, trigger real-world panic, and warp public perception before anyone has time to verify anything. Whether you’re creating or consuming AI content, do it responsibly: label what’s generated, verify before sharing, and don’t trade ethics for engagement.

🔥 Things You Didn’t Know You Could Do With AI

This workflow breaks down how creators spin up AI-driven video accounts that post daily content without filming themselves — and grow fast when the format hits.

  1. Upload a reference image (your “character” or visual style) to Claude and ask it to generate a clean JSON prompt describing the look and vibe.

  2. Paste that JSON prompt into Nano Banana Pro to generate a video clip in the same consistent style.

  3. Write a short script (life advice, story, trivia, productivity, etc.) or have Claude generate 3 variations in your tone.

  4. Use InfiniteTalk to add lip-sync and match the voiceover to the character.

  5. Post, track what performs, and repeat with 5–10 variations of the same format.

Here’s the original workflow reference:

Do you like this new format?

Login or Subscribe to participate in polls.

💬 Quick poll: What’s one AI use-case you’ve seen go viral that made you think, “We’re cooked”?

Until next time,
Kushank @DigitalSamaritan

Reply

or to participate.