Skip to main content
Back

Anthropic Claude Office Integration , Google Chrome AI API , Nous He

Anthropic ships Claude to Office apps, Google launches on-device Chrome AI, Zyphra releases 8B reasoning model, Qwen 3.6 adds multi-token prediction

May 8, 2026 6 min read

Hey,

AI is quietly eating the software stack from the inside. Claude now lives natively across your entire Office suite, context traveling with you from Excel to Word to PowerPoint. No copy-paste, no re-explaining.

Meanwhile, research is catching up to the bigger problem: models that never stop learning. Modular memory, inspired by how our brains work, may finally fix catastrophic forgetting.

The pattern? AI isn't just a tool anymore. It's becoming the operating layer.
Together with · partner_image
Author photo
Today's Author
Lior Alexander. Founder of AlphaSignal and former ML Engineer. Previously built ML systems at Iguazio, Guesty, Enphase, Mila.
Summary

Read time: 6 min 30 sec

Top News

▸ Anthropic ships Claude across Excel, Word, PowerPoint, and Outlook

Tiger Data

▸ Your Agent Needs More Than 2 Projects

Top Repo

▸ Nous Research's Hermes agent now builds full HTML-native videos with HeyGen

Fin

▸ The future of Fin and the Customer Agent category

Top Repo

▸ Google ships Chrome's on-device AI API despite near-universal opposition from browser makers

Signals

1. Zyphra releases open-source 8B reasoning model that rivals DeepSeek and GPT-5
2. PriorLabs releases open-source foundation model that learns tabular data instantly
3. New paper argues modular memory is the missing piece for AI that never stops learning
4. Xiaomi's MiMo reasoning model gets a 6-bit MLX build for Apple Silicon
5. Qwen 3.6 ships with multi-token prediction for faster speculative decoding
Top News
Anthropic ships Claude across Excel, Word, PowerPoint, and Outlook
27,816 Likes
Anthropic ships Claude across Excel, Word, PowerPo

Claude just landed inside your Microsoft Office apps, and it's more useful than a simple chatbot plugin.

Here's the big idea: context travels with you across apps. Start a conversation in Outlook, jump to Excel, then build a PowerPoint deck, and Claude remembers everything from that whole session. No re-explaining, no copy-pasting context.

What you can actually do with it:

  • Excel: Ask about any cell, update assumptions without breaking formulas, and build models from scratch with real formula logic
  • PowerPoint: Build slides inside your existing template, keeping your fonts, layouts, and colors intact
  • Word: Edits land as tracked changes, so nothing gets overwritten without your approval
  • Outlook (beta): Triages your inbox, drafts replies, and schedules meetings. Nothing sends until you click send

To get started, grab the add-in from Microsoft AppSource and sign in with your Claude account. It runs on all paid plans, and if your org already routes traffic through Bedrock or Vertex AI, you can plug right into that.

Read more
forward →
Presented by Tiger Data
Your Agent Needs More Than 2 Projects

Most Postgres providers give you a handful of databases and call it a free tier.

Your agent burns through that before lunch.

Ghost is built for how agents actually use databases:

  • Disposable
  • Isolated
  • Burst-oriented

Fork production in 58ms. Spin up a fresh instance per task. Delete without cleanup. No project limits, no shared state, no naming conventions that nobody follows. Same SQL. Same tooling. Different lifecycle.

100 hours/month free. 1TB storage. No credit card.

Start building
partner with us →
Top Repo
Nous Research's Hermes agent now builds full HTML-native videos with HeyGen
4,301 Likes
Nous Research's Hermes agent now builds full HTML-

Your AI agent can now make videos. No editor. No timeline scrubbing. Just a prompt.

Nous Research's Hermes Agent just got an official HyperFrames skill from HeyGen. The big idea: videos are just HTML files. Since AI agents already know HTML really well, they can now write, edit, and render full videos the same way they write code.

Why does that matter? AI agents struggle to navigate complex GUI-based video editors, but they are experts at writing HTML and JavaScript. HyperFrames skips the GUI entirely.

Here's what you can build with one prompt:

  • Turn a PDF into a walkthrough video
  • Turn a repo into a launch reel
  • Animated title cards, social overlays, captioned talking-head videos, and audio-reactive visuals

Install it in one line: $ hermes skills install hyperframes

Same input always produces identical output, so it fits cleanly into automated pipelines. You need Node.js 22+ and FFmpeg, then you're good to go.

Read more
forward →
Presented by Fin
The future of Fin and the Customer Agent category

Customer expectations are at an all-time high but most support systems weren't built to keep up.

On May 14, join Intercom's founder and CEO, Eoghan McCabe, as he reveals how Customer Agents close the gap, plus a groundbreaking new product.

Save your spot
partner with us →
Top Repo
Google ships Chrome's on-device AI API despite near-universal opposition from browser makers
2,917 Likes
Google ships Chrome's on-device AI API despite nea

Google just shipped the Prompt API in Chrome, and the web standards community is not happy. This API lets any website send prompts directly to an AI model running locally on your machine, no cloud needed.

Here is the problem. Mozilla, WebKit, W3C TAG, and Microsoft all raised serious concerns or outright opposed it before it shipped. Google shipped anyway.

Here is what this means for you:

  • Any website can use the AI model on your device with no permission prompt
  • To use this API, you must agree to Google's terms of use, which is unusual for a web standard
  • The API only works with Google's Gemini Nano, so your code is tied to one company's model
  • Chrome silently downloaded a 4GB model to your machine and will re-download it if you delete it

The core concern: web standards are supposed to work across all browsers. This one does not, and it sets a precedent where a browser vendor bundles a product and calls it a standard.

Read more
forward →
Signals
1
Zyphra releases open-source 8B reasoning model that rivals DeepSeek and GPT-5
2,344 Likes
2
Use real-time data from the Brave Search API + AWS AgentCore to build a market research agent.
Presented by Brave
3
PriorLabs releases open-source foundation model that learns tabular data instantly
6,751 Stars
4
New paper argues modular memory is the missing piece for AI that never stops learning
427 Likes
5
Xiaomi's MiMo reasoning model gets a 6-bit MLX build for Apple Silicon
48 Downloads
6
Qwen 3.6 ships with multi-token prediction for faster speculative decoding
448 Likes
At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 200,000 developers stay informed and ahead. We're passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.

Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We're based in the United States. By using our site, you agree to be governed by U.S. laws.
Looking to promote your company, product, service, or event to 250,000+ AI developers?
Work with us