Here Comes the Internet of Agents
22:34 Share

Today on the AI Daily Brief, Microsoft and the Internet of Agents. And before that, in the headlines, all about the startup that wants to spend a million dollars on agents. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. Thanks to today's sponsors, KPMG, Blitzy.com, and Super Intelligent. And to get an ad-free version of the show, go to patreon.com slash ai daily brief.

Welcome back to the AI Daily Brief Headlines Edition, all the daily AI news you need in around five minutes. And in fact, we're going to try to be extra fast today, given that it's a long conversation with Microsoft Build, and I'm kind of thinking that there's going to be a lot of densely packed news this week. Kicking off though, an interesting story that's been getting a lot of buzz around the internet, a Y Combinator startup is in the market to spend a million dollars on agents.

Web crawling company Firecrawl has posted three new ads on YC's jobs board, but humans need not apply. Firecrawl wants a content creation agent, a junior software engineer agent, and a customer support agent with a $25,000 monthly salary listed for each.

Now, they had actually done something similar to this back in February, but didn't find any AI applicants that were up to the task. However, founder Caleb Pfeffer said that more than 50 applicants have applied in the week since the ad went live this time. He writes, The future belongs to those who can build and control armies of AI agents.

We're hiring hundreds of them and their creators across our business. And indeed, this is an important detail. Although they want these agents to be autonomous, doing things like deciding what to create, creating it, posting it, and measuring the impact and growing from that feedback, they also want to hire the human creators behind the agents.

Now, obviously, this is a publicity stunt, at least in part, but it also is a very first interesting glance at what agent hiring might look like in the future. Pepper said, AI can't replace humans today. The future, what we see, is a world where the next 10x engineers are operating armies of agents, AI systems that they're building, maintaining, and monitoring. What we want to do is work with people that want to be one of those agent operators. This, my friends, is a theme that you will hear lots about when we get into Microsoft Build.

Next up, however, another story on the theme of digital teammates. Salesforce has announced something called AgentForce in Slack that allows companies to build and deploy task-specific AI agents that can interact and have access to Slack, which is, of course, where many employees are already spending their time. Salesforce Chief Product Officer Rob Seaman said, "'Similar to how employees with specialized skills and knowledge collaborate to solve problems, our customers need AI agents that collaborate to solve problems for both their own customers and their employees.'"

Now, VentureBeat sees this as a shift from trying to have general purpose agent to more specialized agents. But it seems to me like if each of the agents is more limited in scope, their capabilities have increased. John Cusera, the SVP of Salesforce AI said, with these digital teammates in AgentForce and Slack, AI agents can reason and take multi-step actions like orchestrating workflows, interacting with multiple systems across the business, and adapting their approach to the specific task at hand. TLDR digital teammates are here.

Next up, a follow-up to a story from last week. XAI has followed through on its promise and revealed Grok's system prompts, which show that the chatbot is supposed to be a hardwired skeptic. Among the instructions, Grok is directed to be balanced and neutral. It's also told, quote, you are extremely skeptical. You do not blindly defer to mainstream authority or media. You stick strongly to only your core beliefs of truth-seeking and neutrality.

Now, it's important to note that we didn't get access to the system prompt that went haywire last week, but this system prompt certainly falls in line with Elon Musk's public personality. I think outside of the specifics, it gives us more of a look into how AI chatbots are programmed. In a separate prompt for the tweet analysis tool, for example, Grok is instructed to, quote, "...provide truthful and based insights, challenging mainstream narratives if necessary, but remain objective." So if you're wondering why Grok seems more edgy than its rivals, that's probably your answer.

Overall, people are excited about this, but want to see it taken to the next level. Daniel Kokotajlo writes, "...publishing system prompts for the public to see? Good! Thank you! I encourage you to extend this to the spec more generally, i.e. publish and update a live document detailing what goals, principles, values, instructions, etc. you're trying to give Grok. The equivalent of OpenAI's model spec in Anthropix Constitution. Otherwise, you are reserving to yourself the option of putting secret agendas or instructions in the post-training. System prompt is only part of the picture."

XAI engineering leader Igor Babushkin responded to that writing, good point. So is this a turning point for AI transparency? Certainly can't hurt.

Lastly today, some more information from inside the guts of a high growth AI startup. A piece in the information has shared a number of details about Perplexity's financial inner workings, including the fact that last year the company made $34 million in revenue, but burned about $65 million. Still, because of its very frequent fundraising, it ended 2024 with around $850 million in the bank.

Their biggest source of cost was web services for R&D, including AWS, at $48 million. That's more than double, for example, what they spent on payroll.

Still, while they're losing money, it feels like this is a pretty good base to build from. The question then remains, just can perplexity continue to prove that there remains a massive opportunity in what used to be pejoratively called AI wrappers? A question to continue pondering, but for now, that is 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 KPMG. In today's fiercely competitive market, unlocking AI's potential could help give you a competitive edge, foster growth, and drive new value. But here's the key. You don't need an AI strategy. You need to embed AI into your overall business strategy to truly power it up.

KPMG can show you how to integrate AI and AI agents into your business strategy in a way that truly works and is built on trusted AI principles and platforms. Check out real stories from KPMG to hear how AI is driving success with its clients at www.kpmg.us slash AI. Again, that's www.kpmg.us slash AI.

Today's episode is brought to you by Blitzy, the enterprise autonomous software development platform with infinite code context. Which, if you don't know exactly what that means yet, do not worry, we're going to explain, and it's awesome. So Blitzy is used alongside your favorite coding copilot as your batch software development platform for the enterprise, and it's meant for those who are seeking dramatic development acceleration on large-scale codebases. Traditional copilots help developers with line-by-line completions and snippets,

But Blitze works ahead of the IDE, first documenting your entire codebase, then deploying more than 3,000 coordinated AI agents working in parallel to batch build millions of lines of high-quality code for large-scale software projects. So then whether it's codebase refactors, modernizations, or bulk development of your product roadmap, the whole idea of Blitze is to provide enterprises dramatic velocity improvement.

To put it in simpler terms, for every line of code eventually provided to the human engineering team, Blitze will have written it hundreds of times, validating the output with different agents to get the highest quality code to the enterprise in batch. Projects then that would normally require dozens of developers working for months can now be completed with a fraction of the team in weeks, empowering organizations to dramatically shorten development cycles and bring products to market faster than ever.

If your enterprise is looking to accelerate software development, whether it's large-scale modernization, refactoring, or just increasing the rate of your STLC, contact Blitzy at blitzy.com, that's B-L-I-T-Z-Y dot com, to book a custom demo, or just press get started and start using the product right away. Today's episode is brought to you by Super Intelligent, and more specifically, Super's Agent Readiness Audits.

If you've been listening for a while, you have probably heard me talk about this, but basically the idea of the Agent Readiness Audit is that this is a system that we've created to help you benchmark and map opportunities in your organizations where agents could specifically help you solve your problems, create new opportunities in a way that, again, is completely customized to you. When you do one of these audits, what you're going to do is a voice-based agent interview where we work with some number of your leadership and employees and

to map what's going on inside the organization and to figure out where you are in your agent journey. That's going to produce an agent readiness score that comes with a deep set of explanations, strength, weaknesses, key findings, and of course, a set of very specific recommendations that then we have the ability to help you go find the right partners to actually fulfill.

So if you are looking for a way to jumpstart your agent strategy, send us an email at agent at bsuper.ai and let's get you plugged into the agentic era. Welcome back to the AI Daily Brief. As we discussed yesterday, this is a big week for big tech conferences, specifically AI Frontier Lab conferences, and Microsoft got us kicked off with the first day of Microsoft Build.

And in case you had any doubts what this was going to be about, I would simply point you to this anchor post by Microsoft Chief Communications Officer Frank X. Shaw, Microsoft Build 2025, The Age of AI Agents and Building the Open Agentic Web.

This was not a day of announcing some big new model. There was no announcement of any weird competition with OpenAI or anything like that. This was a day where Microsoft's assertion that 2025 is the year of the frontier firm, where intelligence can be bought on tap, where human agent teams upend the org chart, and where every employee becomes an agent boss is not some far away in the future thing, but right here, right now.

In fact, if you were watching this presentation and you were in enterprise, that was anything other than up to your eyeballs in agents. Goodness gracious, would you have felt behind. We've been talking for a while now about how we've reached an inflection point where all of our cute little conversations about pilots and explorations and the future are just quaint compared to the here and now assertive reality of change. And the fingerprints of that change are all over this event.

Let's start by just looking at this blog post. Microsoft, again via their chief communications officer, writes, We've entered the era of AI agents. Thanks to groundbreaking advancements in reasoning and memory, AI models are now more capable and efficient. We envision a world in which agents operate across individual organizational team and end-to-end business contexts.

This emerging vision of the internet is an open agentic web, where AI agents make decisions and perform tasks on behalf of users or organizations. At Microsoft, we're showing the steps we're taking to make this vision a reality through our platforms, products, and infrastructure. We're putting new models and coding agents in the hands of developers, introducing enterprise-grade agents, making our platforms the best places to build, embracing open protocols, and accelerating scientific discovery with AI, all so that developers and organizations can go invent the next big thing.

And it's worth taking a pause here and zooming out for a second. Yesterday, I talked about the five vectors of AI competition, two of them being agents and the enterprise. And these are obviously very interlinked in some ways, if for no other reason that the enterprise is about to be completely transformed by agents. The gist of what Microsoft is saying in that Work Trend Index report is that pretty much all the stuff that we do now, the actual doing of the work, will be done by agents in the future.

The people who are doing that work now will be managing the agents in the future. Your experience and the skills you've built up to this point will be redeployed in service of managing teams of these digital employees who are doing the work that you used to do.

Now, what's interesting, I think, about the ubiquity of agents and Open Agentic Web and all of these things all across this Microsoft presentation is that in some ways, it actually makes sense that Microsoft, as the default leader in enterprise software worldwide, would be the deepest in and making the deepest bet in and around agents. They have a front row seat to the changes that are coming.

And those changes are, frankly, whether they have the best software or not, in many if not most cases, coming through Microsoft Rails. CEO Satya Nadella reinforced this message with his summary thread on Twitter about this. Today, he writes, at Build, we showed you how we are building the open agentic web. It's reshaping every layer of the stack, and our goal is to help every dev build apps and agents that empower people and orgs everywhere. Satya then goes on to share the five big things that were announced today.

So let's go through those announcements, starting with their coding agent. Nadella writes, We're taking GitHub Copilot from being a pair programmer to a peer programmer. You now have a full coding agent built right into GitHub. You can assign it issues, whether it's bug fixes, new features, or ongoing code maintenance. And it will complete these tasks autonomously.

We'll go back to that graphic I had about the five vectors of AI competition outside of agents and enterprise. Another one was, of course, coding, and we expected to hear something about this. Now, I'm not sure that this is a particularly overwhelming announcement. In fact, many people have wondered how GitHub, which had such a lead with Copilot, allowed so many of these other platforms to basically lap it with developer affiliation.

At the same time, though, GitHub Copilot has the benefit of installs inside enterprises, something that Cursor and Windsurf just can't match. Now, the details of the coding agent itself were pretty de rigueur. GitHub says the agent incorporates context from related issue or PR discussions and follows any custom repository instructions, blah, blah, blah, blah, blah. The point is, the coding agent kind of does what you would expect it to do. In fact, the part of the announcement which was a little bit more buried, but which was way more exciting to developers, at least on X, was

was that Nadella and Microsoft also announced that VS Code, of which both Cursor and Windsurf are forks, would go completely open source.

Next on Nadella's list, co-pilot tuning. He writes, co-pilot can now learn your company's unique tone and language. It's all about taking that expertise you have as a firm and further amplifying it so everyone has access. And to me, what was actually interesting about this is that if you go check out the blog post that announced this particular piece, it's an even clearer representation of just how much Microsoft is actually pushing its enterprise customers into the future.

Now, co-pilot tuning is something that is likely to be extremely successful with organizations. It reminds me of what Glean has been doing, offering basically organizations the easy ability without a ton of technical capability to tune AI models using their own data. That being embedded in the Microsoft ecosystem will be extremely appealing. And more than that shows that we are getting to the advanced levels of using AI inside companies, not just hovering around the edges with our cute little co-pilot pilots, putting

Putting another explanation point on that idea, they also announced multi-agent orchestration to, as they put it, enable agents to work together as a team with human oversight and direction. As part of that announcement, they said that already more than 230,000 organizations, including 90% of the Fortune 500, are using Copilot Studio to create and customize agents.

The fact that Microsoft has now gotten into the agent orchestration game shows how quickly agents are finding their way across the entire organization. Again, the message is clear. Microsoft is saying basically that even if you don't have enough agents yet that you're going to need orchestration, you will really soon.

Next up is Agent Factory. Nadella writes, Foundry is the complete app platform for building apps and agents. We're adding more support for models from Grok, Hugging Face, Meta, Mistral, and more, plus agentic retrieval in Azure AI Search, Foundry agent service integration with Copilot Studio, and more. And we are ensuring the tools that you already use for identity, management, and security will now all extend to agents too. So two big parts of this announcement, obviously, agent, agent, agent, agent, agent, and specifically

and specifically enterprise-grade agents. And second, wait, did he say Grok? Yes, one of the things that people jumped on was that OpenAI and Sam Altman antagonist Elon Musk was invited to be a part of the conversation. Now, it's not surprising at all, if you've been watching Microsoft's strategy, that they would integrate Grok. They want Azure to be a neutral place where you can get access to all the different models, even if they have a preferential relationship with some of the companies that have their models on there, like OpenAI.

People will, I'm sure, be looking for this to be a gotcha moment, proving yet again that Microsoft and OpenAI are breaking up, but this feels completely inevitable. What's more interesting is what the information picked up on, where, as they put it, Musk said Grok was proving useful in customer service at both SpaceX and Tesla, and he hopes to sell that service to other companies. Quote, in other words, Musk wants to sell Grok to businesses just as other AI firms want to do.

Question, of course, is who will Grok's enterprise customers be? And how seriously does Elon want to pursue the enterprise opportunity? By the way, there were some other guest appearances at the event. Sam Altman was there virtually as well. Nadella and Altman talked about how the role of developers is changing. The 2021, one of the very first things we did together in GitHub. And we've been talking about how someday we'd get to like

a real agentic coding experience. And it's kind of wild to me that it's finally here. I think this is one of the biggest changes to programming that I've ever seen. But this idea that you now have a real virtual teammate that you can assign work to, that you can say, hey, go off and do some of the stuff you were just doing and increasingly more advanced things. You know, at some point say like, I got a big idea, go off and work for a couple of days and do it. And that you can issue many requests in parallel, that you can be fixing bugs, implementing new features, answering questions about the code.

This is like true software engineering task delegation. And I think it'll only get better from here, but this is just a tremendously exciting moment. It integrates very deeply with GitHub. You can give it access to a repo in an environment and you can get some pretty amazing stuff done.

So again, hitting on those big themes of coding and agents. And then there was this announcement of NLWeb. Nadella said, "This is a new open project that lets you use natural language to interact with any website. Think of it like HTML for the agentic web."

Now, I think this is interesting for more than just having three great initials at the start of the name. In their overarching post, they continue, "NLWeb makes it easier for websites to provide a conversational interface for their users with the model of their choice and their own data, allowing users to interact directly with web content in a rich semantic manner." The Verge extends the explanation, writing, "With a few lines of NLWeb code, your choice of an AI model, and whatever data you supply to the model, you can have a custom chatbot up and running in a few minutes."

Basically, NLWeb handles all the logistics of turning a question into an answer, and all you do is supply the data.

Their goal is very ambitious. They put it as NL Web becoming the fastest and easiest way to effectively turn your website into an AI app, allowing users to query the contents of the site by directly using natural language, just like with an AI assistant or co-pilot. They continue, just like the introduction of HTML made it easy for almost anyone to create a website, we want NL Web to make it easy for any web publisher to create an intelligent natural language experience for their site. As the agentic web and economy continue to grow,

NL Web will empower web publishers to participate on their terms, ensuring their website is ready to interact, transact, and be discovered by other agents if they choose. TechCrunch kind of reductively called it Microsoft's project to bring more chatbots to web pages, but I think the implications are just the broader AI-ification of every web experience.

Now, the last thing, which honestly could be a whole show on its own, was what Nadella called Microsoft Discovery. He writes, we're bringing together the full tech stack to help speed up science itself. Discovery uses agents to generate ideas, simulate results, and learn. So basically, this is Agents for Science, which, like I said, is a whole separate show. Bringing it back to Microsoft, though, one of the questions that people had going into this event was what the implications of these recent layoffs were.

Did they suggest that AI behind the scenes wasn't going that well? The Information wrote a really interesting piece called Behind Microsoft Layoffs, Automation Efforts Boom. And the TLDR of it is Microsoft dogfooding its own idea of becoming a frontier firm. They write, while the layoffs only affected 3% of staff, and Microsoft publicly said the move was intended to reduce layers of management, six current and laid off employees said executives have foreshadowed it for months as they talked up the imperative of boosting automation.

Senior leaders across the company have been repeating a refrain to subordinates. Use AI to increase productivity in sales, engineering, and customer support, among other roles. That's in line with what Microsoft has been telling its enterprise customers, that Microsoft will be customer zero for the same AI it's pitching to them as a way to spend less money on employees.

Now, a Microsoft spokesperson has argued vociferously that the company is not replacing roles with AI, but it's very clear that a transformation is underway. If you were surprised at the statistic that somewhere between 20 and 30% of Microsoft's code is now produced by AI, one Microsoft VP in charge of about 400 software engineers recently told them that their goal was to have a full half of the code they write coming from AI.

Oh, and by the way, the one other thing that I didn't even get a chance to get to is that Microsoft is super into the model context protocol. This is part of their larger open agentic web push. In that core announcement post, they write, Microsoft is delivering broad first party support for MCP across its agent platforms and frameworks, spanning GitHub, Copilot, Studio, etc.

In addition, Microsoft and GitHub have joined the MCP steering committee to help advance secure at-scale adoption of the open protocol and announce two new contributions to the MCP ecosystem. In case we had any doubts that MCP was being firmly enshrined in this emerging agent developer ecosystem, this should clear that up entirely.

And so if you look at this, on first glance, it doesn't seem like there's any one big announcement. There's nothing that's necessarily going to grab headlines. That's why, in fact, most of the articles that have been written have focused either on the Elon Grok thing or on a pro-Palestine demonstration that broke out right at the beginning of the event. But for our purposes, if you're sticking around here and listening to this show, the sum total is extremely significant. The TLDR is we are past pilots and we are on into the agentic frontier.

Microsoft is investing in agent building frameworks. They're launching agent orchestration tools. They are screaming forward. And so I would suggest that this is a good moment for most enterprises to gut check their agent strategy and ask whether they're keeping pace with a field that's evolving incredibly fast. For now, that's going to do it for today's AI Daily Brief. Appreciate you listening as always. And until next time, peace.