Microsoft Agents, First API for Grok, and Notion Email Client

PLUS HOT AI Tools & Tutorials

Oct 25, 2024

Hello and welcome to our weekly roundup!

Well, it's been a busy week. While OpenAI is having a bit of a rest, Microsoft, Notion, Stability AI, and even xAI, news about which appears quite seldom, took the stage.

Let's go through all these updates.

Perhaps the main release of the week was “computer use” by Anthropic. We made a post on this topic with a step-by-step guide on how to run Claude AI on your PC.

You'll find everything you need to know here:
Claude Can Control Your PC
Stepan Ikaev
·
Oct 24
Read full story

This Creators’ AI Edition:

Featured Materials 🎟️
News of the week 🌍
Useful tools ⚒️
Weekly Guides 📕
AI Meme of the Week 🤡
AI Tweet of the Week 🐦
(Bonus) Materials 🎁

Creators AI Recommends: AI Art & Animation
AI Art & Animation is a newsletter for creatives exploring the latest in Gen AI, including 5 quality image prompts delivered weekly. Perfect for artists, media professionals, and digital creators—check it out 👉
AI Art & Animation

Featured Material 🎟️

Microsoft Will Let You Create Your Own AI Agents

Big news from a big company. Microsoft has announced that it will greatly expand the functionality of its AI platform Copilot Studio as early as next month. Users can create their own agents, honed to perform specific business operations. The company believes this update will accelerate the integration of AI into complex industries.

Specifically, agents will be able to act on behalf of employees to automate repetitive tasks, provide analytics, and optimize operations. Copilot Studio will get several new tools that combine personal, business, and analytics data to make the process more robust. This will allow companies to create greater control, transparency, and security agents.

To convince potential customers of the platform's effectiveness, Microsoft clarified that Clifford Chance, McKinsey & Company, Pets at Home, Thomson Reuters, and many others are already building agents to increase revenue, reduce costs, and scale impact. The first results are already in.

McKinsey & Company, for example, has created an agent that speeds up the client onboarding process. A pilot project showed that turnaround time could be reduced by 90% and administrative work by 30%.

Microsoft has launched ten new autonomous agents in Dynamics 365 as an add-on. These promise to help sales, service, finance, and supply chain teams drive business value. Among them are:

Sales Qualification Agent
Supplier Communications Agent
Customer Intent & Customer Knowledge Management Agents

Next year, the company will create more agents that autonomously perform tasks in different areas.

Microsoft also cited several numbers showing how AI is helping it transform itself: the sales team increased revenue by 9.4% and deals by 20%.

I'll clarify here in case we don't have any public data to back up this information. And I'm not the only one who doubts Microsoft's ambitions. Salesforce CEO Marc Benioff did not like Copilot at all.

Keep your mailbox updated with key knowledge & news from the AI industry

Marc Benioff Against Microsoft Copilot

Okay, we're used to Twitter scandals, but they don't often involve top executives of giant corporations. Benioff had already criticized Microsoft at the Dreamforce customer conference, but he went harder on Copliot this time.

Here is what he wrote about Copilot on X:

It just doesn’t work, and it doesn’t deliver any level of accuracy. Gartner says it’s spilling data everywhere, and customers are left cleaning up the mess. To add insult to injury, customers are then told to build their own custom LLMs. I have yet to find anyone who’s had a transformational experience with Microsoft Copilot or the pursuit of training and retraining custom LLMs. Copilot is more like Clippy 2.0.

How do you feel about this opinion?

News Of The Week 🌍

Elon Musk’s AI Startup Launches an API

Two months after Elon Musk promised to make Grok available via API, it's finally here. Currently, the API offers access to a model called "grok-beta," though it's unclear if this is the latest Grok 2 model or an earlier version. The documentation mentions both Grok 2 and Grok mini, suggesting multiple models may be available in the future. Developers describe Grok as a general-purpose model capable of generating and understanding text, working with code, and analyzing images.

Pricing and Usage

Input tokens: $5 per million (approximately 750,000 words)
Output tokens: $15 per million

If you want to try the Grok API, you’ve to take the following steps:

Sign up for the service and create an API key through the xAI Console.
Authenticate requests by including the API key in the Authorization header.
Make API calls using curl or the provided SD.

Notion Reveals Its Email Client (with AI!)

Notion held its “Make with Notion” conference and made some big announcements. The main one was the Notion Mail, which is currently in the closed testing phase but will be available early next year. The company will integrate email functions into the Notion Workspace with this product. There, you will be able to manage your emails together with notes, databases, and projects.

And, of course, Notion Mail will support a host of AI features. The new app will allow you to automate the organization of your emails, and it will also allow you to outsource some tasks entirely to Notion AI. In particular, this model will be able to conduct business correspondence and schedule tasks.
We will know the quality of implementation in 2025.

Genmo Introduces Mochi 1 Open-Source Video Generator

Another AI-based video generator has been released, this time from startup Genmo. The company introduced Mochi 1, an open-source model for generating text in video. The developers recognized that many modern AIs like to fantasize too much, causing users' expectations to often diverge from the results. Therefore, Mochi 1 will strictly adhere to the given instructions.

This is the main difference between the new platform and its competitors.

Genmo has already launched a playground where you can try out Mochi 1 for free. The scales are also available on AI model hosting site Hugging Face.

Sharing is caring! Refer someone who recently started a learning journey in AI. Make them more productive and earn rewards!

Refer a friend

StabilityAI Releases Stable Diffusion 3.5

StabilityAI announced the new Stable Diffusion 3.5 family of AI models. It differs from its predecessor with a higher level of realism, prompt adherence, and text rendering compared to SD3. The family comes in three sizes: Large (8B), Large Turbo, and Medium (2.6B). All of these are already available for use under the Stability AI Community License. You can download Stable Diffusion 3.5 Large and Stable Diffusion 3.5 Large Turbo from Hugging Face and the output code on GitHub now.

As you can see from the generated photo above, Stability AI has been placing a big emphasis on realism lately. But that's not all. With the new update, users have more control over image creation. As the developers say, Stable Diffusion now supports just about any visual style.

Spirit LM: Meta’s Multimodal Model for Text and Speech Generation

Meta's Fundamental AI Research team has introduced SPIRIT LM (Or SPoken and WRitten Interleaved Transformer Language Model), an open-source model that can process both text and speech as input and output. According to the developers, this platform was created to solve some problems associated with existing AI voice systems, which often sound unemotional and robotic.

Spirit LM offers a new design with tokens for phonetics, pitch, and tones to add these expressive qualities to its speech output. It can also learn new tasks across various modalities, including automatic speech recognition, text-to-speech, and speech classification.

If you are interested, you can already test it. Meta Spirit LM is available to the research community under the non-commercial research license FAIR.

Perplexity AI vs News Corp

To finish, let's discuss legal stories related to the AI world. Earlier this week, one of the largest media holdings, News Corp, sued Perplexity for stealing content from WSJ and the New York Post. According to the lawsuit, the AI search engine exploits news from these media outlets, violating copyright law. Moreover, it says that some responses directly copy content without altering the resulting data.

It didn't take Perplexity long to respond. The AI developer answered that it doesn't believe corporations can't own a copyright on facts and that the allegations are baseless. The blog post's conclusion was rather harsh.

AI-enhanced search engines are not going away. Perplexity is not going away. We look forward to a time in the future when we can focus all of our energy and attention on offering innovative tools to customers, in collaboration with media companies.

Looks like we're about to see a trial. The only thing I’d like to mention here is the funny moment when Perplexity played the “evil corporation” card. We are, after all, used to this status being given to AI developers, not the other way around.

Useful Tools ⚒️

GOSH - Free AI Price Tracker for any product at any store

Kolena - an AI platform that turns unstructured data into actionable insights

Overlap Studio – Transform any long video into clips, blog posts, and more

Duonut – AI Conversational surveys with real-time follow-ups

Rupert AI - AI toolset for designers and marketers. It’s a platform designed to streamline AI workflows for businesses by providing access to open-source models for text, audio, video, and images. Rupert includes:

Facebook Ad Testing Tool: Evaluates and ranks pre-test images to optimize ad campaigns based on potential performance.
AI Product Photography: Generates high-quality product images for marketing, which is ideal for e-commerce and brand promotion.
AI Fashion Model Generator: Allows you to create diverse fashion looks by mixing and matching outfits on virtual models.