AI Agents -- The Latest Push from Microsoft and Salesforce

Today on the AI Daily Brief, Microsoft and Salesforce both announced new AI agents. Before that in the headlines, Gavin Newsom signed some AI bills, but says that he's concerned about SB 1047. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show. Notes welcome back to the AI Daily Brief Headlines edition. All the daily AI news you need in around five minutes. Currently in San Francisco, Dreamforce is going on, which is Salesforce's big annual event where they do a lot more than talk about just Salesforce in the main episode. Today we'll talk a little bit about some of their announcements, particularly AI agents in Slack. But one of the more interesting conversations that happened was California Governor Gavin Newsom speaking with Salesforce CEO Mark Benioff, during which he actually live signed three AI related bills focused on election ads and deepfakes on stage. Bloomberg says the bills he has signed are aimed at cracking down on the use of artificial intelligence that aims to deceive voters in political ads, as well as legislation that provides new safeguards for actors and musicians and other entertainers over their digital likenesses. Now, while SB 1047 might be the most controversial legislation, apparently it's not the only controversial legislation. Elon Musk tweeted, the Governor of California just made this parody video illegal in violation of the Constitution of the United States. Would be a shame if it went viral. This is a Kamala Harris campaign ad parody from back in July. Now Elon already has 14 and a half million views on that, so that debate is likely to heat up when it comes to SB 1047. However, Newsom said that he was concerned about the bill particularly he's worried about a potential, quote unquote, chilling effect on the development of AI. During that conversation with Marc Benioff, he said, we dominate this space and I don't want to lose that. He said that one of the issues is the weighing of risks that are demonstrable versus hypothetical, and pointed out that, quote, the the impact of signing wrong bills over the course of a few years could have a profound impact on the state's competitiveness. We haven't had much of a chance to get into this, but one of the things that's interesting about OpenAI's new Zero1 model release, which was of course previously known as Strawberry or Q internally, is that it takes a different approach to scaling than just throwing more compute at the issue, Rohit points out. So now that you can scale inference, time compute to get better answers. What's the fate of SB 1047 or the various AI acts that anchored on training flops? Basically, one of the concerns for people who are opposed to SB 1047 is that any sort of numbers or delimiters you put on model size or cost are likely to not make any sense in a very short amount of time. The fact that we almost immediately got a model that was organized and trained differently seems to be at least some evidence that that could be an issue. Newsom has until the end of the month to sign or to veto SB 1047, and I will of course keep you posted as that evolves. Interesting news from Microsoft and BlackRock. They have formed a new group to go out and raise 30 to $100 billion to invest in AI data centers. The companies are two of the leaders in what's called the Global Artificial Intelligence Infrastructure Investment Partnership, or gaiip, which also includes Global Infrastructure Partners, which is an infrastructure investor that is being acquired by BlackRock, and MGX, which is a tech investor based in the United Arab emirates. MGX has G42, the controversial UAE company that Microsoft recently made a minority investment in, as one of its founding partners. Between this and Sam Altman's plans to go raise trillions for a global AI infrastructure initiative, it's clear that this buildout just continues to ramp up. Beleaguered intel has announced its plans to deal with flagging fortunes. The company is planning to spin out its Foundry business and at the same time announced a big chip deal with aws. Now this transition to having a chip foundry division was supposed to be Intel's big comeback play. However, it hasn't worked out as well as they had hoped and so now they're transitioning Intel Foundry to become an independent subsidiary that will gain an operating board, including independent directors, even though its leadership won't change and the subsidiary will remain inside the intel company. Intel also announced that the new Foundry company would pause chip fabrication projects in Poland and Germany for two years, but that at the same time they had signed a deal with AWS to co develop an AI chip with Intel's 18A chip fabrication process. The AWS deal was described as a quote, multi year, multi billion dollar framework and overall the market liked what they heard. The combination of announcements sent intel up 6%, although it has still been a really rough year for that company. Lastly, today, 01 continues to roll out bringing better reasoning to a variety of different applications. One of the latest is Perplexity, where CEO Aravind Srinivas said, we've added a new focus on Perplexity for pro users. It will use the new OpenAI 01 Mini. There's no search integration yet, the model is slow and usage is limited because of rate limits. It's good for puzzles, math and coding. Pro searches are your best way to use multi step reasoning and web search. I think it's fair to say that we are still very early in figuring out what the better reasoning capabilities of O1 are really going to be most useful for. But with the model coming to places like Perplexity, there will be many more shots on goal to figure that out. That's going to do it for today's AI Daily Brief Headlines edition. Next up, the main episode. Today's Episode is brought to you by Venice. Venice is a private, uncensored, generative AI app. It accesses open source models to enable text, image and code generation without the fear of being spied on or having your data exploited. Discuss anything with Venice without concern about it being monitored, sold or given to advertisers and governments. Venice is different because your conversations and creations are kept securely within the browser, never stored or accessible by Venice. Unlike other AI apps, Venice won't tell you what's okay to say or not. Venice won't patronize you. It simply provides direct access to machine intelligence. No topics or off limits, no ideas are taboo. With Venice, you're in control of the AI as you should be. Pro subscriptions are available for $49 a year or $8 per month. AI Daily Brief listeners receive a 20% discount on Venice Pro. Visit Venice AI NLW and enter the discount code NLW Daily Brief. That's NLW Daily Brief, all one word. Today's Episode is brought to you by superintelligent, which is of course our platform that helps you learn how to use AI tools and perhaps even more importantly, gives you ideas on the best use cases that are actually going to help you achieve whatever it is you want to achieve. To recognize the End of Summer and Back to School, Back to work. We are running our best promotion ever when you sign up for Super Intelligent between now and the end of August using code so back your first month will be 100% free. The platform features over 600 fun, highly practical AI tutorials that get you using AI fast and with an eye to actually transforming how you get things done. We've just launched super for Teams, so if you have a group of people at your company that want to figure out how to use AI together, I highly suggest you check it out. But for those of you who are using superintelligent as an individual. Once again, if you sign up for superintelligent between now and the end of the month using Code soback, you will get your first month 100% free. Go to be super AI and check it out today. Welcome back to the AI Daily Brief. Today we are talking about one of the persistently hot topics that has lurked around ever since the beginning of this whole wave of generative AI, but which remains an elusive, if exciting, part of the future. As you'll see today, though, Big Tech is making a bet that that future is increasingly here, and we are of course talking about agents. However, to set this up, let's get the context of what Microsoft has called its Copilot Wave two. This was an announcement from earlier this week where Microsoft wrote we're launching the next wave of Microsoft 365 copilot, bringing together web and Work and Pages as a new design system for knowledge work. So there are three big parts of this Wave two announcement. First, they announced something called Copilot Pages, which they characterized as a dynamic persistent canvas designed for multiplayer AI collaboration. Second, they made a big update around Copilot in the Microsoft 365 set of apps. They've updated Copilot and PowerPoint and Outlook, and they've also now tried to add advanced data analysis in Microsoft Excel. Excel and spreadsheets have historically been sort of a little bit behind when it comes to the value of generative AI, so it's notable that they're now trying to bring those capabilities into that particular product. Finally, and most relevantly for this conversation, they announced that they're introducing Copilot agents, quote, making it easier and faster than ever to automate and execute business processes on your behalf. There's a lot that's interesting in here. Copilot Pages does seem like a potentially really interesting interface, given that teams need to collaborate on various issues and use AI together. It seems to me that the introduction of Pages reflects the same sort of impulse that we're seeing at companies like Anthropic, where we're moving into an era where it's not just about capabilities, but also about user interface and user experience. As I mentioned, I think when it comes to the 365 updates, the Excel update is the one that I'm most interested to see how it evolves. As Fortune puts it, Excel now has the ability for generative AI to automatically compose code in Python directly in Excel and use that to automatically run sophisticated data analysis tasks, including forecasting and data visualization still, obviously the big thing and where most of the conversation is focused is on this agent idea. There's not that much more info about agents that was released, the Verge writes. The agents act like virtual employees to automate tasks instead of Copilot sitting idle waiting for queries. Like a chatbot, it will be able to actively do things like monitor email inboxes and automate a series of tasks for data entry that employees normally have to do manually. What's interesting about this is more the fact that this battle is finally coming to the actual marketplace rather than just being theoretical. All of the big tech companies are of course working on agentic AI, but Microsoft was the second big company to actually put production ready agents into the market. Salesforce announced its agent platform earlier this month and at Dreamforce yesterday discussed a new set of Slack AI agents as well. Slack announced that it would support agents from Adobe, Anthropic, Cohere, Perplexity Writer and others, as well as of course Salesforce agents. Now one of the interesting implications of this is of course around the business model, bloomberg writes. Salesforce's new AI strategy acknowledges that AI will take jobs. They continue the way that they frame it. Is Salesforce unveiling a pivot quote now saying that its AI tools can handle tasks without human supervision and changing the way it charges for software? Basically, Salesforce is charging $2 per conversation held by its agents, which is a totally different approach to pricing. As Bloomberg points out, the goal is not only to have a business model that matches the actual activity, but also to, quote, protect Salesforce if AI contributes to future job losses and business customers have fewer workers to buy subscriptions to the company's software. Now. Salesforce, for its part, is really pushing the idea that agents aren't just an alternative to the assistant or co pilot era of AI, but that they are actually the answer to disillusioned business customers. In a statement, the company wrote that the AI agents in contrast to now outdated copilots and chatbots that rely on human requests and struggle with complex or multi step tasks. In his speech, Salesforce CEO Marc Benioff argued that the new AI agents will be highly accurate and secure because of how much customer data Salesforce already holds, in contrast to quote those nasty copilot. TechCrunch explored the Slack dimension of this writing. Slack is turning into an AI agent hub. Should it Basically, the piece questions whether agents are really important for Slack. However, along with agents is coming a different vision? CEO Denise Dresser said that they're trying to make Slack not simply a work messaging platform, but rather a digital workplace that brings all of your people and processes together. Now, if you've been following along with the headlines that Klarna CEO announced that they were going to drop Salesforce in Workday to build their own AI replacements that got mentioned at this event as well. Benioff said that he's very skeptical and wants to see some proof that the company is actually doing it, basically accusing them of just trying to grab headlines. Speaking of Workday, they also held a big event this week. Rough timing given that it's happening the same time. It's Dreamforce and they as well announced four new AI agents. The four agents are called Recruiter, Expenses, Succession and Optimize. From CIO.com, recruiter helps source candidates, recommend talent and automate Outreach Expenses quote unquote virtually eliminates the need for manual expense reporting. Succession automates the succession planning process by prompting managers to update succession plans. And you get the idea these are all verticalized, task or domain focused agents. Of course, it's not just big tech. You are starting to see more agentic AI actually come to market from startups as well. One that I've been watching is called Auto, which, as described by founder Sally Omar Otto, lets you use AI agents within tables to automate hours of manual research in a few minutes. The point is, after a very long time in development, the first wave of agents is actually being brought to market. The question of course is will these agents actually be able to do something useful or are they still not quite there yet? That is going to be a big exploration over the coming months and one I will be excited to share with you here. For now though, that's going to do it for today's AI Daily brief. Until next time. Peace. Sam.