Welcome back to another edition of AI in Action. We just stepped into August 2025 and man, Zach, what a week. Lots of stuff going on before we dive into the main topic today, Zach, let’s hit some of the latest developments in the AI world. There’s been a lot going on. I think the best way to tackle this is let me tee up the headline, and then you just spend a few moments unpacking why it’s important, why we should care, and the overall, implications in the AI landscape.

Does that sound good? Yeah. I mean, there’s a lot going on, you know, billion dollar browser wars and billion dollar power moves. So, yeah, let’s dive into it. People are throwing a lot of resources into AI, and they’re very smart people that are doing it. So we really need to be keeping our eye on this. And small businesses and growing businesses all need to be keeping our eye on this.

Big news hit this week with massive implications. ChatGPT conversations are now indexable by Google. What does that mean? Yeah, I think this is a it’s a pretty big deal. It’s not it’s not that they’re now indexable. They’ve always been indexable, but only if you did one specific thing. And that’s if you were having a conversation with ChatGPT and you hit that share button to generate a public link that you could, you know, share with your friends.

We’ve probably all done that at some point because we want to we want to share the insights that we’ve been getting from, you know, a given conversation. Well, what happens when you do that is yeah, generates a link. But I think a lot of people’s expectations, mine included, is it works kind of like Google Drive, which is it’s a private link that anyone that gets the link can access.

But I don’t think we expected it to show up in the search engine results. Well, yeah, we found out that this is actually indexable by search engines. And so if you type in, if you if you search a site like chatgpt. com/conversations or I can’t remember what the exact link is, but you can take that and put a topic in and you can start going and looking through all these public conversations that have been indexed by the search engine.

So I think it’s, it’s not an incredibly surprising. I don’t think there’s any sort of expectation of privacy around these tools anymore. And I don’t think that’s necessarily right. I don’t want to argue the, you know, the right or wrong about it, but the fact of the matter is, most of this stuff is, is just don’t have any expectation of privacy, right?

And it’s publicly indexed. Even if you pay for the tool. Is there any way to turn this off? Well, I think if you’re on an enterprise planet it works a little bit differently. So like enterprise is kind of designed to, you know, generate those links and share among the company, which is like Google Drive. But if you’re on a free plan or a pro plan, yeah.

That public link becomes indexable. So, I would say if you’re using ChatGPT and you’re worried about this, you can go into your settings, go to your data privacy settings, and you can view all the public links that you created. You can remove them. Now that doesn’t mean that it’s going to just magically disappear from the internet because, you know, the internet archives and things like that exist.

So it’s already out there. But you can you can remove it from what’s already been shared. And then now you’re aware, right? So just be aware that if you’re if you’re going to create that public link that is going to be indexable at least today, maybe that changes in the future if there’s some backlash from this. But yeah, just be aware of that when you’re sharing those conversations.

Let’s move on. Chat GPT five is coming, we think, next week. This is first week of or last week of July, August 1st. So we think, next week, we’re going to see GPT five. Yeah. Yeah. I think that is the expectation. Maybe it’s next week, maybe it’s later in August. But, it’s really just the next iteration of OpenAI’s large language model.

So it’s rumored to be about a million input tokens. Which is great because that’s that’s getting up there with, Google Gemini. So Gemini has always had a pretty big context window, so it’s good to see that we’re increasing that context window here. And it’s supposed to have about 100,000 output tokens. Going to bring in some built in tools supposed to bring a code interpreter.

So, dynamic reasoning, the ability to call tools in parallel. Basically that means if I am going to, you know, call like a Google Drive or a connection to a CRM or something like that, instead of having that happen in sequence where I’m first going to go call Google Drive, get the results, and then I’m going to go call Salesforce, it could do both of those things at once so that you get a faster response back.

Which is great. So I think I think it’s going to be, you know, I mean, it’s a better model, right? We see this come out all the time. I think we’re almost a little desensitized to how quickly this stuff is coming out. It’s just a yeah. No surprise here. We got another upgrade. It is absolutely an arms race.

And the tools are taking off. So much so that anthropic is basically like hey guys back off. They’ve introduced this week a cap on cloud code usage. What does that mean. Yeah. So now Pro and max users are that are in the top 5% are going to kind of face some new caps here. So, it’s really just to curb things like account sharing and service strain.

This is probably something we’re going to see more of as these companies start to figure out pricing models and realize that certain models are going to be less profitable than others, or even issues with like the top 5% of users that are using it more than anyone. I mean, they’re they’re taking a lot of resources. And so I think it’s it’s similar to something like Salesforce, which has a shared tenant environment.

Everyone on Salesforce is going to be subject to certain limits so that you’re not overtaking a server compared to everybody else on that server. Right. This is not a shared tenant, right? It’s not the same idea from an enterprise standpoint. But, if certain people are using a huge amount of the power and the resources available to the tool, it’s going to make it less efficient.

It’s going to work less well for some of the other users on the system. And so I think this is an effort to make sure that it’s working well for everyone. But I think if you are in that sort of top five usage to your top 5% usage, you may not like the result that you get here, but yeah, I also think we’re not going to see this be the final change, that this is going to change over and over again with all these tools as, as you know, the companies start to figure out how they want to price these products.

I mean, they can’t be such a bargain forever. They’re such a massive, energy, uptake required. There’s there’s massive infrastructure required to operate these tools to grow these tools. And we know we’ve seen, you know, the investment in talent, we saw recently meta poaching Apple’s top AI talent to bring them over, you know, hundreds of millions, billions of dollars in these deals to bring these folks on board.

So it’s it’s not inexpensive to build an AI, model. And I think we’re going to see those, this business models evolve as well. Let’s move on. AI’s breaking the internet. Yeah, it really is. There’s, there’s this whole idea of this dead internet theory. I don’t know if you’ve heard this, this concept, but it’s like, you know, at some point, it might even be today, we might already be in the dead internet era, but we’re so much content is generated by AI that it loses the human element.

Right? I think we’re seeing that pretty rapidly. I mean, you look at social media, I mean, you can look at your Google search results and, some of it’s just AI slop. Some of it’s like good stuff. It doesn’t mean everything generated by AI is, you know, slop. But there’s a lot of slop out there, right? This is if you’re just using AI without doing it in a conscious way or using it to kind of enhance the work that you’re doing and you’re just allowing it to, you know, write entire research reports or anything like that.

Right? So yeah, I think I think the whole dead internet theory is interesting. You know, Google’s AI search is reducing clicks by 50%. So if you’re on an old ad model and you’re relying on, you know, pay per click and things like that, I mean, it’s it’s pretty different, right? The world is changing pretty rapidly from a marketing perspective and an SEO perspective.

So we’re seeing sites that are, you know, fighting back with paywalls and CAPTCHAs, and Cloudflare has this pay to crawl model now where you actually have to pay to, to have your site crawled by these AI tools. So that’s pretty interesting. So yeah, not now. You’re seeing basically like the AI layer where, you know, you’ve got this bot paid, ad driven instant answers optimization, and then the human layer where you get gated content subscription base is built for trust.

So I think we’ll see a, a progression into more of that material on, on the internet, which I think is, a little sad, right? Because we’ve been so used to this open internet concept and, really, you know, the past few decades have been driven by the internet being completely open and free. And I don’t know that this is making the internet less free from a regulatory perspective.

But just the idea that there’s so much out there that we can’t differentiate between what’s AI generated and what’s real and what’s not, you know, moving to more of a paywall based subscription type of model may be one of the only ways that we can, you know, ensure that we’re getting that good information. So it’ll be interesting to see where this goes in the next six months, two years, five years.

I think the landscape of the internet is, you know, at least positioned to change dramatically if things continue along this current trajectory. Yeah, I agree, there’s just I mean, if you think at this time last year, some of the some of the things that we’re talking about didn’t even exist and the atmosphere and the, the internet’s changed so drastically, over the last 12 months will definitely be interesting to watch.

One of the things I thought this week was, was cool that ChatGPT released was study mode. So instead of now just going on to ChatGPT and, you know, a student or whoever asks a question and gets, gets the answer, it has more of a mode where they, it interacts with them to challenge them to, to get them to, you know, instead of it giving them the question, you know, just the answers.

It’s questioning and giving them hints and guides and guide and guiding them through. But this quote specifically from Zuckerberg, he said that superintelligence is now in sight. And so he’s saying that meta is seeing indications of self improvement with their models. They really want to focus on empowering individuals to achieve goals with AI, as opposed to focusing just on workplace automation.

And so I think they’re really trying to put I kind of into your day to day life as opposed to, you know, just use it to, to automate, you know, parts of your, your day job. Right? So you’re seeing things like they’re smart glasses that they already have which, which are pretty cool, by the way. I mean, just the standard augmented reality Ray-Bans that they have are pretty sweet.

But yeah, those things are awesome. Yeah, yeah, I’ve got some friends that have it. And I mean, they’re pretty, pretty awesome for just recording video on the fly. But now imagine like layering in AI capabilities into that. So, having the ability to have AI be aware of what’s going on around you and make suggestions and you can ask it questions and things like that.

I think that’s kind of the vision they’re going after. So, yeah, I mean, you know, what kind of see where it goes. Ultimately, we’re still talking about large language models. So the idea of self-improvement, I’m not sure exactly what that means, but I guess we’ll we’ll find out here in the near future, kind of where Matt is going with this.

Cool. Let’s, let’s dial in on Salesforce, a couple of big, items in the news. This was as recent as yesterday was force and ServiceNow are both investing $750, $750 million each in Genesis. Please tell us the implications there. Yeah, I also have invested $750 in Genesis. That’s all I’ve gotten. Yeah, $750 million each. Pretty interesting.

So Genesis is a contact center as a service platform. So it has sort of close ties to things like Salesforce Service Cloud and the offerings that ServiceNow actually has on the market. But, both companies, Salesforce and ServiceNow have connections with Genesis. And so you can you can integrate with this platform, and it can essentially just expand the capabilities that you have.

With Service Cloud or with ServiceNow. And so I think this is this is interesting because, you know, analysts would look at this and say, okay, they’re really focusing in on sort of multi-system agent tech AI for the future, which I think may be true, but also this was already kind of a strategic investment. This isn’t the first time that these companies have invested in Genesis.

They’ve had partnerships with Genesis for a while. So I think this is kind of a natural progression. I do think that it’s a way for, these companies to have influence over where Genesis kind of develops their technologies in the future, with the preference for integrating well with Salesforce and ServiceNow, which I think is really interesting. And then, of course, yeah, the ability to continue to use AI throughout the system.

So, you know, we’re talking about Agent Force and Salesforce and you want to be able to leverage Agent Force with a connected Genesis org. You know, this could indicate that there’s going to be more capabilities being developed in that way. So we’ll just have to kind of keep our eye on and see what happens. Is interesting that they announced both investments at the same time.

It seems like that puts Genesis in an odd place since these platforms are pretty competitive with each other. Yeah, I would say they’re very competitive with each other. So, you know, Service Cloud and ServiceNow are direct competitors. Right. And so it is interesting to see this. I think you could even look at this as two of the leading service platforms investing in the future of AI.

So maybe putting their differences aside and saying, all right, it’s important for us to invest in this tool for both of our futures to make sure that we’re, you know, able to adopt the latest technology and continue innovating. So, yeah, I mean, like I said, I, you know, I don’t know where this is going to go or, you know, what the, implication of this will be for each of these platforms in the near future.

But I think it’s, a pretty interesting indicator into the importance of, you know, investing in AI forward tools and how serious these companies are. And by spending, you know, multi billions of dollars, as we’ll talk about kind of in this, this next acquisition that Salesforce is working on. Yeah, it’s an absolute arms race. I don’t I don’t really remember the last acquisition Salesforce did, but they had announced one this week a company called Bluebirds.

What is Bluebird Stadium. Yeah well last acquisition was Informatica. And so that’s you know kind of like the data pipelines and plumbing making sure that you’ve got the right data for AI. And now Bluebirds they are an AI prospecting platform a AI prospecting platform. So, you know, ways to identify high intent leads and reach out to those leads.

And Salesforce already has some tools for this, right? Like they have an SDR agent which has the ability to go in and reach out to leads contacts and initiate that outreach, manage that conversation and try to book a meeting. So I think this is a direct investment in trying to improve that tool and innovate on top of the capabilities they’ve already built.

So, this is a, deal for an acquisition now that hasn’t gone through yet, but they’ve signed the definitive agreement, so there’s still some time, you know, for this to actually close and be completed. But assuming that it does, I think we can expect the capabilities of bluebirds to start being woven into the Salesforce ecosystem. And I think specifically, it’s going to look like improvements to the SDR agent, or maybe even building out sort of an entire AI prospecting suite to complement what Salesforce already has on offer.

Since we’re talking about Salesforce, let’s just stay here for a minute. They’ve got some upcoming enhancements to their AI tools. Let’s talk first about Agent Force three and the Command Center. What’s next? With the evolution of that. Yeah. So in June, they announced Agent Force three, kind of the end of June. So it’s been just over a month since they announced this.

But command Center is the combination of what they’ve formerly called Interaction Explorer, which is a robust sort of dashboard of everything going on with your AI agents. So if you want to know, all right, what are some of the keywords? What are some of the main topics that people are discussing with these agents, and how successful all these conversations and where are those conversations falling off?

How effective is our reflection, you know, things like that? Reporting on that has been kind of difficult, up to now. And this is really an effort to make that much easier. So Command Center is going to be your central sort of location inside of Salesforce to look at everything going on with your AI agents. So analyzing performance, trying to identify bottlenecks, fine tuning that behavior because you’ll be able to see how those different conversations are progressing and identifying where they’re falling off.

So it’s really a way to empower users of agent force to monitor agent adoption, troubleshoot issues, and ultimately just drive better outcomes. I’m excited for this. I’ve been waiting for Interaction Explorer to come out for a while because, you know, we do a number of these implementations and one of the sort of number one requests that we get from our clients is I really like to see better reporting.

So this is directly addressing that. And that should be coming out at some point this month in August. I don’t know if that’s going to be, you know, next week or, or the end of August, but we should be seeing that really soon. And you can see why that’s important. As companies start to embrace agents and use agents for more and more of their activities and more and augmenting their workforce, this thing’s going to get gigantic.

You’re in need some sort of 10,000ft view of, hey, how are these guys doing? It’s almost like a performance management tool for your agents. Yeah. I mean, if you think about how Salesforce does, you know, service, cloud and omnichannel and you can be a supervisor and omnichannel and see kind of everything that’s going on with all of your customer service agents.

It’s kind of a similar idea, except now you’re monitoring digital labor, right? So you’re saying, all right, how are all these, you know, individual AI agents doing? And how can I get a, an understanding of their overall performance where I can see, you know, issues that might need improvement, things like that. So, yeah, I think it’s going to be a big, a big upgrade to anyone who’s already on Agent Force.

Heard you. You mentioned something in an earlier meeting about native MXGp support. What is that in law? Yes, that’s another part of Agent Force three, and that was supposed to come out in July. I haven’t necessarily seen that capability. So that may be something that’s coming out here in August. But the idea here is Agent Force is going to have native MXGp support.

MXGp is the model context protocol that anthropic came up with. It’s basically a shared agreement on how to connect large language models to things like APIs and different systems. And so it’s it’s a framework to allow people to use a shared language to create these connections so that a large language model, let’s say, has connections to, you know, Google Drive and Salesforce and HubSpot.

Well, I can just start asking, hey, I need you to go and, you know, tell me about how this opportunity’s been progressing and it can go and leverage all of those tools, assuming you’ve given it access to it, to go and figure that out. And so instead of creating, let’s say like one point connection to Salesforce, I can create sort of multiple tools and skills around Salesforce, around Google Drive, around HubSpot, whatever that could be.

And it’s able to essentially intelligently route those actions based on what’s being asked. So now we’re seeing that come to Agent Force, where providers will be able to build these MXGp servers, these actions that you’ll be able to call from Agent Force. And so if you’re in that agent force builder, you know, if you’ve worked with topics and actions, that’s things like creating flows or apex for that agent to go and perform those actions.

Well. Now that can also be an MCP server to another tool. So an example that they used was PayPal and it was the ability for that agent to just reach directly out to PayPal and create an invoice and bring that invoice back into the conversation and show that information. So if you were to do that today without MCP, I mean, you’d have to build some sort of connection out to PayPal.

You’d have to be able to handle all the logic and the responses, and then the agent would have to call that. And so this is this is really kind of a much cleaner way to connect. And I think it’s important that, that they’re leveraging MCP because that is it’s really kind of the way to connect to other systems.

Now. It’s a new protocol. So there’s, you know, a lot of sort of security requirements around it. You got to make sure you’re building it in the right way. But, I think it’s important that Salesforce is doing that in order to stay up to speed with the competition, because everyone’s really trying to figure out how to implement MCP, and hub hubs has already done it.

Yeah, I was about to say HubSpot beat him to the punch on this one. Even recently announcing MCP integration with Cod, as well as open AI. So they’re starting to look at some of these other tools. Speaking of open AI, that’s kind of been Salesforce’s go to Lem today, but there are more options on the horizon.

Yeah. So with Agent Force, it’s a little bit complicated because you kind of have two different points in time where you hit a large language model. So if you’re running a prompt template and doing some natural language processing, you’ve always been able to select from multiple models. And so that today supports open AI, supports Gemini Sports Llamas Sports Claw.

There’s there’s a bunch of sort of different things you can select there. But the reasoning when you actually are using Agent Force to make a decision for how it is supposed to take action, that’s always been run through open AI. So that Atlas reasoning engine, where you’re, you know, determining the next best action. That’s always been through open AI.

Well, now with Agent Force three, we will have the ability to use clod for that reasoning engine. And then in the future, Gemini will be available for that as well. So we’re getting more model choice, not just in the prompts that we’re using, not just in that natural language processing, but also now for the reasoning engine. And so if we want to use something other than open AI, those possibilities are coming to the platform that we’re excited to talk about our next series of episodes with our audience, talking specifically about what businesses need to do, what how they need to be preparing, how they need to be thinking about AI and what they need

to be doing now to get ready. This is a fast moving landscape. It can feel like businesses are in a tornado. They’re out there running their businesses, so they may not even have time to think about this. What we want to do is provide some really structured guidance around how they need to be thinking about AI and what they need to be thinking about.

So this AI readiness series, we’re going to unpack this over the next several weeks. Tell us about the five pillars of AI. And then give us a little bit of a preview of our next episode. Yeah, absolutely. Yeah. So the framework that we’re kind of working with here is, is a combination of some things that we’ve seen as well as, some stuff that Salesforce has come up with.

So, their framework is people process and technology. We agree with all of those items. And then we’re also adding strategy and data. So those five pillars are strategy people process technology and data. So it really speaks to kind of each one of those areas in terms of hey, how can you start to prepare to leverage AI at your company?

And each pillar is really pretty important because you’ve got to have a strategy, you’ve got to have the right adoption, you’ve got to have agreement on how to use these tools at your company. You’ve got to have data to support it. You’ve got to have the processes that you can articulate to AI to actually go out and do this job.

And so yeah, well, we’ll be talking about each of these pillars in the coming weeks. I’m excited. Next week we’re going to have our CEO, John Burdette, on the podcast. And we’re going to talk through our strategy for some motion and what that looks like, as well as how you can think through adopting an AI strategy and creating a strategy around AI at your company.

That’s been fun. As always, we’re going to let everybody go. If you are serious about putting AI to work in your business, we would love to have a conversation with you. Fast slow motion. com/ai connect with our team. We’d love to help you out.

Yeah.