PracticalyAI

Wed, October 7, 2026
3-minute read

3 AI UPDATES  ·  2 THINGS YOU DIDN'T KNOW YOU CAN DO WITH AI  ·  1 KEEPER

👋 Good morning.

Claude just pulled a chair into Google Docs, Sheets and Slides: a sidebar that reads the open file and your selection, then edits in place if you approve. Public beta, every paid plan, per Anthropic. Gemini has had that building to itself. Claude took the desk by the window.

IN 3 MINUTES, GET 1% BETTER THAN EVERYONE YOU KNOW:

  • Claude now edits inside Docs, Sheets and Slides

  • Fourteen reference photos, one banana

  • Free Gemini drops to Flash-Lite on Thursday

  • A train window and a desktop rug, both vibe-coded

  • 226 motion prompts, credited and free to steal

3 AI updates

Claude for Google Workspace, public beta. Source: Anthropic / X.

Claude for Google Workspace is in public beta on all paid Claude plans. A sidebar opens next to Docs, Sheets or Slides, reads the file you have open plus your selection, and edits in place. The default is ask-before-edits, with each change shown as an approval card, and you can switch to accept-all. In Slides it builds from your deck's layouts and flags overlap, overflow and hard-to-read text. Separate Docs, Sheets and Slides connectors let Claude edit a Google file from a pasted link, per Anthropic and 9to5Google. On Team and Enterprise an owner has to enable those first.

Do this: If your briefs already live in Docs and your model preference is Claude, install it from the Workspace Marketplace and keep ask-before-edits on until you trust the pass.

Gemini still has the home-court advantage. Claude just took a desk in the same building.

Which AI gets to edit your Google Docs?

Claude's Workspace sidebar is in public beta on every paid plan, ask-before-edits by default.

Login or Subscribe to participate

Nano Banana 2.1 announcement

Nano Banana 2.1. Source: Google / X.

Google shipped Nano Banana 2.1 on October 6 with better visual design, mask-based editing and subject consistency, plus sharper text and infographics. Multi-image fusion takes up to 14 reference images and holds up to 4 characters and 10 objects steady across them, per Android Authority. It is rolling out in the Gemini app on the paid AI plans, with free staying on Nano Banana, and across AI Mode, AI Studio, Flow, Stitch and Google Ads. The DeepMind model card says it is built on Gemini 3.6 Flash.

Do this: If you generate product shots or deck graphics in Gemini on a paid plan, run one multi-reference pass with your real product photos before you rebuild the brand kit by hand.

Fourteen references. One banana. The fruit bowl is getting competitive.

Gemini model access changes. Source: 9to5Google.

Starting Thursday, October 9, Gemini app users without a Google AI subscription drop to Flash-Lite only, so free loses Flash and Pro, per Google's support doc. AI Plus keeps Flash-Lite and Flash. AI Pro and Ultra keep the full ladder, and 9to5Google reports AI Pro also gains Deep Think. New low, medium and high effort levels apply to every available model, and the higher you go, the more of your limit each answer uses.

Do this: If free Gemini Flash still sits in your daily stack, decide before Thursday whether Flash-Lite is enough or a paid tier is the floor.

The free tier just got a calendar invite. Attendance is mandatory.

2  things you didn't know you can do with AI

Sami catalogued over 500 hours of YouTube train-window footage into a site that lets you stare out of a train and, somehow, control the weather. It is a mood tool, not a dashboard: open it when you want a moving horizon behind the work. The catch: the link lives in the video post, and it is footage, not a focus feature. Nothing here will finish your deck.

Train-window weather site. Source: @hellosami / X.

Terkel made a rug for your desktop so you can sweep your mess under it. Icons slide under the rug in the demo, a visual joke that doubles as a tidy ritual for a junk-drawer desktop. The catch: it hides the mess, it does not sort it. The files are still there on Monday.

Desktop rug that sweeps mess under it. Source: @terkelg / X.

1  Keeper

Prompt Motion gallery. Source: @p4nthera_ / prompt-motion.com.

p4n built Prompt Motion, a gallery of motion graphics made with Claude Opus 5.5, each shown beside the prompt or skill that produced it. There are 226 so far, pulled from posts on X and credited to their creators, and the site is free. Browse it when you want a motion reference and the exact wording behind it, not another thread of prompt tips.

The catch: the prompts are the creators' originals, not a tested recipe. Steal one for a brief or a Claude session, then expect to tune it for your own footage.

Don't forget to rate today's post

This helps us put better content for you

Login or Subscribe to participate

🎤 Join Kushank Live on Friday!

How to edit videos with Claude

Build in real time and bring your questions.

📅 Friday, October 9, at 10AM Pacific / 1PM Eastern

Reply to us if you wanna be a part of this!

Hope to see you there!

That's your 1% for today. Hit reply and tell me which Google file you'd hand Claude first.

Kushank, AI teammates, and the human team
PracticalyAI

Recommended for you

View all
caret-right