Hey AI enthusiasts!

The next place you meet an AI assistant might be the document already open on your screen. Today, we look at Claude’s new Google Workspace beta, give Sesame a small planning job, and unpack what an invisible text watermark can actually tell you.

In today’s edition:

• Claude comes to Docs, Sheets, and Slides

• A practical voice task for Sesame

• Four useful tools to explore

• ChatGPT’s planned EU text watermarks

• Four quick updates worth your attention

NEWS

🔵 Claude moves into your Google files

Claude can now sit beside the Google file you’re working on. Anthropic’s October 6 launch brings a beta sidebar to Docs, Sheets, and Slides for paid Pro, Max, Team, and Enterprise subscribers.

• In Docs: ask it to rewrite a section or apply formatting directly.

• In Sheets: work on formulas, pivot tables, and native charts.

• In Slides: revise a deck while using its existing layouts and theme.

Install the Claude add-on from Google Workspace Marketplace, then open it from the Extensions menu. Your organization’s administrator may need to allow it first.

The default “Ask before edits” mode lets you review proposed changes. These are direct edits to the file, so keep Google’s version history in mind when reviewing the result.

Try a bounded first job: select a long paragraph and ask for a shorter version that preserves every date and commitment. Compare the result with the original before accepting it. That gives you a useful test without handing over an entire document.

Read more here

TOGETHER WITH ALIGNED

Stop losing deals in between meetings.

Aligned is changing how B2B teams run their most important deals.

Most reps guess where a deal really stands. They forecast on a gut feeling and the buyer’s word vs what's actually happening. Then they realize the deal slipped through only after it's too late. Aligned surfaces what buyers are actually doing inside every deal: the risks, the openings, and the exact next step to keep things moving.

Start increasing your close rates and stand up your first deal room in minutes.

Start Now For Free

AI TOOL OF THE DAY

🎙️ Give Sesame one small planning job

Sesame’s October 5 getting-started guide shows how its voice assistants can move beyond conversation: connect apps, browse, and build reusable instructions called Skills. The mobile app is available on iOS and Android in supported countries, including Italy, the UK, and the US. It’s for adults, with English the officially supported language.

Try a calendar planning exercise:

1. Choose an assistant and connect Google Calendar through Apps in Agent Hub. Check the permissions you’re granting.

2. Say or type this original prompt:

3. Check the dates and gaps yourself. If the format helps, ask Sesame to turn these instructions into a reusable Skill for your morning planning.

The useful output: a proposed schedule you can review, with the final calendar decisions still yours.

Read more here

4 TRENDING AI TOOLS

📸 Snapdown: turn a screen selection into Markdown, including headings and tables. Runs on Apple Silicon Macs with macOS 14 or later; ten free captures let you try the workflow.

📝 Oats: record meetings and extract notes and action items without a bot joining the call. This open-source Mac app offers a local model option; check which backend you select.

🗣️ Willow: dictate into the app where your cursor is and get formatted text. Available for Mac, Windows, and iPhone. Read the result before sending anything important.

💻 HackerRank Chakra: rehearse an AI-assisted technical interview through its “Try Interview” flow. Practise doing the work and explaining the decisions behind it. Use a desktop browser.

NEWS

🔎 ChatGPT’s EU text watermark has limits

OpenAI announced textGrain on October 5: a watermark built into patterns of word choice, rather than a visible label attached to a paragraph.

The company plans to introduce it to eligible ChatGPT and Codex text output in the EU across plans in the coming weeks. This is a planned rollout, not a feature already switched on everywhere. Developers can opt in globally for selected API models; it is off by default there.

Detection initially goes to approved researchers and expert organizations, rather than a public checker. Short passages, constrained formats, and later edits can make the signal harder to detect. False positives and false negatives remain possible.

Why it matters: a watermark is a clue about how text was generated. It doesn’t establish that a claim is true, and a missing signal doesn’t prove a human wrote the passage. If you review other people’s work, treat provenance and factual accuracy as separate questions.

Read more here

QUICK AND FRESH

🔍 Google expands on-device search models. EmbeddingGemma 2, released October 6, supports semantic search across text, images, audio, and video. It’s a building block for developers, not an automatic upgrade to your photo library.

➗ OpenAI shares mathematics research. Its October 6 update publishes results and proof work from an internal, unreleased model. Some proofs use Lean verification; this is research progress, not a newly available ChatGPT mode.

🖼️ A visual ad test is coming to ChatGPT. Announced October 5, the US test is planned for later this month on Free and Go during image generation. Ads will be labeled and separate from the generated image.

🧩 Mistral Large 4 enters public preview. The October 6 model release gives developers another option for text-and-image tasks through the API. Public preview is the availability to watch here.

PARTNER WITH US

Building something useful with AI? Reach readers looking for practical tools and ideas. Explore sponsorship opportunities.

YOUR OPINION MATTERS

Which task would you try first: editing a document or planning tomorrow? Tell us what you’d like us to explore next.

Find us at @alltrendsai.

Until next time, stay informed!