El fin de la era de las aplicaciones: ¿qué viene ahora?

[0:00] What if Door Dash was the clue to how software was sold and bought in 2026 and 2027? I think it kind of might be because on September 30th, Door Dash announced a way to let you text your agent at Door Dash and have your agent at Door Dash get you what you want. Your favorite pizza, your favorite sandwich, whatever it is. This is the future of software. And there are twists and wrinkles in business software that Door Dash doesn't have to face when you're getting a sandwich. And now I want you to take that into work with me. Just
[0:30] walk with me for a second. A new customer is going to sign up for whatever you're selling at work. And your agent is going to read the contract. Your agent is going to look at sales notes. Your agent's going to read the email thread. Agent, agent, agent, right? We get tired of it. And then your agent's going to write the kickoff plan for you. You stop opening the tool that you used to use all the time. That's what agent disruption looks like. The company holding your customer records is fine. They still get paid. But over a few months, you teach the agent everything that nobody has really
[1:01] written down at work. Like this customer needs security approval before kickoff, but that's kind of a different process. And we use it this way. And so in this world that we're all barreling toward, three things are happening at once. First, the extra tool that used to put that plan together, the software, that got less sticky, that got easier to leave. That's the risk for SAS right now. The agent you train can be harder to replace if all of those lessons learned live in your custom harness,
[1:31] maybe on your individual machine or in the memory associated with your individual agent at work. And if your CTO, your finance team, or other teams you work with want you to switch AI vendors, that's about to get really, really tough because you may all be counting very different costs. They may not see the way you're using your agent. This is the battle that is coming to teams in 2027. But first, what does this actually look like? What are the kinds of stories I'm seeing? I'm going to give you an example. It's rooted in a real
[2:02] story. I've anonymized it lightly. Let's say every week you open up a recruiting tool. You look for candidates. You copy your notes into a spreadsheet and you write up a short list for the hiring manager. That's part of your role. Then your AI agent starts doing it and you think, "I'm more productive. This is great. I'm doing my job." It pulls candidate profiles from a data provider. It drafts a short list. Now, you review it, of course. But when you give the okay, it then goes to the hiring manager. You stop opening the recruiting tool over time and the company's selling the candidate data. What happens to
[2:32] that? They still get paid because they're still selling data. You're still using the data. You're now paying for different parts of the work on top of that data. And so, you start liking how the agent is working. You stop thinking about the recruiting tool a lot. Your agent gets really good. It gets full of context and you've taught it a lot about how you personally screen for candidates that are successful for your roles, how you screen for candidates that don't really fit the company and what a great
[3:02] candidate looks like. It's also starting to learn what particular hiring managers want. It even knows who you actually hire because you feed that information back in and that seems to be helpful in helping it to pick better candidates and reduce the time you need to spend manually reviewing over time. and you're responsible and you're ethical and so you still manually review, but it's helping you a ton. Now, that story can be hard to reconstruct if you haven't been in the trenches, but I have seen versions of that. Obviously, not just in
[3:32] recruiting, across product, across engineering, across sales. Everywhere I look, you have old tools that were singlepurpose, like the recruiting tool I described, that are getting easier to leave. and the agent you trained on is getting much harder to leave. And the moment leadership wants you on a different AI vendor, that's often because of a cost conversation. They want more optionality. That suddenly turns into a really tough conversation across the company that leadership isn't
[4:02] ready for because everyone at the table has really different priorities. So, I'm going to show you how to get through that conversation as an employee, how to lead it as a leader, and really importantly, you're going to walk away with a sense of where software is going from here. How do we think about software in the age of agents? What parts of the software stack are actually more valuable. Spoiler alert, I have a lot of confidence in software going forward. It's just going to look different. So, we're going to get into what that looks like and how we think about the combined and messy world of
[4:34] software and agents. And if you're like, "Oh, this is going to be hard." It won't be hard. It'll be fun. We'll have real stories. We're going to have fun with it. And you're going to walk away with real tools. Okay, let's get started with what you are still buying when the agent is doing the kind of work that I just described. When the agent is doing a lot of what I would call the information processing part of a role and you are managing that agent, which is happening in every role in 2026, when you open an application, what are you still there for? Information you can't get elsewhere. something you need carried
[5:05] out correctly. When do you actually open that application? I'm serious. Take a second and think about it for your role. Customer service, engineering, product, I don't care. Think about it for a second. Do you open that in and I mean actually directly open the application? I don't mean ask your agent too. This is a different thing, right? We'll get into the distinction. If you the human are finding or arranging or acting on information, when and where do you do it in your role? That's an important basis point. A lot of SAS, the software businesses that we're talking about here, ends up selling a very muddy value
[5:35] proposition that covers finding and arranging and acting on information and in some cases storing information. And so an agent getting better at just a part of that process ends up exposing that entire SAS business relationship because they were never clear on their value proposition in the age of agents. And that's a lot of how I'm reading the complicated conversation around SAS companies in the age of agents. They're reasoning from the data and distribution they already have into the agent position they think that they can hold.
[6:05] And it's not like these companies aren't aware of agents. Every single one of them is leaning in on AI. They're probably talking about agents in vendor presentations. You're as a buyer not short of everyone saying they're AI native now. It's like required. And so if you want to understand what's actually happening, you have to reason back as a buyer from the data and distribution that SAS vendor has and think through how they approach their agent position from that point. If that sounds like complicated theory of mind
[6:35] and strategic thinking stuff, I've got news for you. Buying software in 2026 that actually works is a more complicated process now because you have to get it right for the future and agents keep getting better. So, this is part of why it gets hard in the trenches, right? Because people are making these decisions and they don't realize that they didn't think through the scaling power of agents in their own systems before they did so. And now employees are stuck. And what's interesting is if you are in this conversation, if you're thinking about agents, your competitive landscape
[7:06] between agent and software is really complicated right now. You will see in a second, I'm going to start naming specific well-known names and how they're all playing in this space. And you're going to realize the competitive set got broader. And then at the end of the conversation, I'm going to share a little bit about what it means for what I would call point solution SASes, less known names that are less big that are selling individual solutions and how they approach things. But let's start with the names that we're all going to be familiar with. And yes, the first one is a SAS business effectively, even if
[7:36] it's not being marketed one. And that's part of why this is such an interesting topic right now is because we're all blurring those lines together. So Meta, yes, in the SAS business, even though you wouldn't think about it, Meta has started with consumers with the Muse product, right? Muse is a personal agent, but Muse for small business launched within a couple of weeks because Meta saw how people were actually using a personal assistant product as a business tool. That's going
[8:06] to be a huge trend in 2027. So, Muse for small business adds connections to business accounts. This effectively allows you to plug into a data layer. Remember, I talked about this idea that how you position your data layer is going to matter profoundly in 2027 because that is something that if you are selling a data layer, the agent needs it desperately. So, Muse is like, we're going to add data. That's their first instinct. And so they actually add yes other tools some of which are SAS tools right Shopify is a tool for
[8:37] connecting to your shops. QuickBooks is a tool for connecting to accounting. My read here is that Meta wants to extend the relationship from just helping you with your business as a personal assistant might to helping you find customers and helping you operate. They see the muse that you have created as business HQ for a small business owner. If you're a consumer using the agent and now you go to work and you are also using QuickBooks, you just want a simple
[9:08] solution and that makes a ton of sense. And so the path from personal to business use at that scale is really short. So let's go from Meta, which no one would think of in the SAS category traditionally, to the company that invented SAS. I'm talking about Salesforce. Salesforce is the OG SAS company. They have tremendous distribution at Dreamforce. I told the Salesforce team that startups would kill for their distribution and most companies don't want to reinvent customer management software. All great things for them. So what happens to a
[9:38] company like that in the world of agents? What's interesting here is that they are in a strong strategic position regardless. I've argued that in the past, but there's a new wrinkle as agents get better and better because better models give Salesforce more chances to apply the knowledge that they already have in the system to more of the customer's work. You can think of knowledge as having latent possibilities or latent value, dead value to the business and more intelligence, more
[10:10] available can unlock more of that. And you can kind of see that ambition and where they went with Dreamforce with their KOA announcement, which is a reasoning model based on the Neotron stack that Salesforce tuned to CRM and announced on September 15th. COA is in pilot, so it's not widely available yet. Uh, and Salesforce was very careful to train it on synthetic data. They're not training on real customer data, but they understand how real customer data flows work because they've been in the business for like 20 years, right? So they understand how to build that synthetic data set well. And so Salesforce is kind of doubling up here,
[10:40] right? Like if you think about it, we've talked about the value of data in the age of agents. Salesforce has that, right? We've talked about the value of distribution in past videos. Salesforce has that. But one of the things that they need to lean into is the agent side to make sure that the agents that are used on the data are agents that are part of the brand experience, part of the Salesforce value proposition and promise. And that is really like think of Koa as the first toe in the water on
[11:10] that one. In my view at least, I think that the opportunity there is for them to both be a data layer and also a place where agents can come and use the tools. And to be fair, Salesforce is offering this to all agents. It's not like KOA is going to be the only reasoning tool. In fact, they had a strong claw partnership that I saw when I was at Dreamforce. You can use other agents to bring them as well. They have a whole agent control panel. So they have a whole thing they are basically positioning themselves across three levels and I think it's
[11:40] really instructive to see from a future of SAS perspective because on the one hand they have that data layer advantage which I've argued before is powerful in the age of agents who need a tremendous amount of data. They also have the ability with that data and distribution to construct effectively an agent gateway that allows companies to have a trusted surface through which to engage with their data agentically. And so Salesforce could be in a place to say trust us with the agents. We will handle everybody's agents and by the way here's
[12:11] KA our agent that can help specifically where it's useful with CRM data. That's a pretty strong place to be and that's a place where you can start to have a much more nuanced conversation about value. And by the way, if you are wondering how do companies get to a stronger conversation if they're trying to sell SAS in 2026 and 2027, it's about moving past a simple demonstration of value into a multi-value stack. Agents are going to force software to talk about how they impact
[12:42] multiple parts of the stack in a way that we have never seen before. I think that Salesforce is sort of the OG at that right now, too. They're reinventing themselves. Now, let's go even farther down the stack. So, we've talked about Meta, whom you wouldn't think is in this category at all. We've talked about Salesforce that sort of invented the category. Now, we're going to talk about the company that invented software that we all use every day, Microsoft. Why are they in this SAS conversation? Because they have the same strategic dilemma. They are in a place where they have a
[13:12] data layer. They're in a place where they're positioning themselves as a control plane for agents. And they also have all of these productivity tools that they're used to people using through direct interfaces. Tools like Outlook, tools like Excel. And increasingly, the direction that work is going is away from those interfaces toward asking your agent to manipulate data and process it. And you're managing your agents. And so Microsoft is in the same dilemma. Isn't it interesting that we can go from like a tiny startup,
[13:44] which we're about to talk about, all the way up to the biggest companies in the world and they're all wrestling with this. This is because we are talking about the future of work here. And so I talked about autopilot in a recent video. I won't give you the whole story. You can go back and watch that video. Autopilot is a persistent agent in preview mode that they shipped. And it's effectively part of Microsoft's bid to complicate the conversation when they talk with their enterprise partners. They want to be in a position where they can say, "Hey, value proposition we
[14:15] bring to the table. We're managing your data. We're a secure place to store it with Azure. We obviously have the dev tools there in the Azure development environment." And on top of that, of course, we have the whole Microsoft productivity suite that you're used to and you love and no one is going to walk away from entirely because it feels too risky. And oh, on top of that, we bring trusted AI. We bring AI that nobody ever got fired for buying, right? Uh so we bring C-pilot. We bring autopilot now. We bring the way you will do work in the
[14:45] future. And that's really like their value proposition is essentially saying we're stacking these bricks and we're adding more value over time. And because of their scale and their distribution and effectively the trust that they're monetizing, they are making the case that they are the ones companies can trust with AI. You notice how, by the way, we've been talking about AI for a few minutes here and the labs have not really come up. So open AI hasn't really come up. Anthropic hasn't really come up. I did a whole video just recently on how open AI is starting to play into the
[15:15] enterprise space. I think that's a lot of the story of Devday. I talked about that. Anthropic is certainly a big player in the enterprise space. These players are starting from a different position. They don't have your data. They don't have existing distribution. They're trying to gain that distribution in a matter of a few months. And so really the play if you are a lab is simple. It is about saying bet on us for intelligence. Bet on us to be the best in the world at knowing how to apply
[15:46] that intelligence to your data. embed on us for future intelligence. And that is true regardless of which lab you're in. Now, there are nuances. I talked about some of the differences. Anthropic has leaned very heavily into the enterprise strategy and OpenAI is newer at leaning into enterprise because they've had such success on the consumer side and focused a lot on consumer side. But at a businessto business level, those are the value propositions they're putting on the table. And a lot of where the future
[16:16] conversation goes is how their value proposition ties in with some of the existing players. And that's why it makes sense that Claude tied up with Salesforce because in a sense if you want to be in a position as a buyer where you're not locked in and every business leader, spoiler alert, every single business leader doesn't want to be locked in. That is the thing they're nervous about in the age of AI. And so if you don't want to be locked in as a buyer, as a seller, you have to sell optionality. And so when Salesforce and
[16:46] Claude partner together, you get the position where a buyer can get like whatever they want, right? They want Cloud, they can get Claude. They want COA, they can get COA. They want a different model, there's a control plane for that that Salesforce offers. They want forward deployed engineering support for AI in their business. Well, there's forward deployed engineers from Anthropic that can help them with that. So they have a lot of options and that makes for a more compelling value proposition in 2026. in 2027. Now, let's talk about a startup. We've talked about big big companies like Salesforce, like Microsoft, like Meta. We've talked about
[17:16] the labs. Let's talk about an aggressive startup that is still scaling up. Kind of a big company, but still startupy. Door Dash. Now, Door Dash, you would think there's no way they're in the SAS category, Nate. You're crazy. What are you talking about? Well, they're not selling you software. They're selling you lunch. But the way they're selling you lunch is actually pointing to the future of software. I promise you this makes sense. So on September 30th, they announced their text ordering experience which you can just like type in the US and it will just send you food, right?
[17:46] It knows your data. It knows you personally. It knows your favorite lunch spot. It will just send it to you, right? And on top of that, they announced a limited beta of their MCP, right? MCP of course is famously the plug that Anthropic invented that allows agents to plug into data. So Door Dash, yes, they want you to use their own assistant, but they want to be available when you use anybody else's assistant. Does that sound like a common pattern? Do you start to see the connection here? So either way, they want to be in a position where Door Dash gets the order.
[18:16] In other words, the future they see is that you are typing / Door Dash at the command line and you are getting your favorite lunch order sent to you real real fast. Or you're texting them and you're getting your favorite lunch order. Either way, Door Dash wins even if you never ever ever touch their app experience again. So losing that moment in the interface doesn't mean losing the customer transaction and doesn't mean losing the customer value. An agent might even bring in orders that would otherwise have been too much trouble for
[18:46] the human to organize and figure out. So you might think, "Okay, Nate, we've gone all this way. It sounds like you're saying every software company is going to need to provide an MCP server, right? And maybe they need to have their own agent, too, like KA." It's not that simple. There has to be something on the other end that the customer still needs in the age of agents. And this is why I have been talking about the importance of data and context so much. If two providers offer equally good information
[19:17] and equally dependable execution, yes, in that world, connecting them may make it easier to compare and replace them. Being reachable by an agent in 2026 and 2027 is going to get you considered in the conversation. It is table stakes. It does not establish why someone ought to keep paying you. You have to decide that and you have to deliver that whether they touch the interface or not. And as a buyer, you have to understand that that's the world you're operating in now. And then you have to understand especially that that's the world your
[19:49] team members are operating in. Remember when I said at the beginning of this video we would talk about how complicated it is for individuals to figure out how to talk about this stuff? Part of the issue is that there are so many ways for agents to work with data and that people have evolved such idiosyncratic work styles with AI so quickly. And so if you are in the business of buying software, you have got to be aware of the prevailing patterns of actual AI usage inside your
[20:19] team before you make that decision. And and as a team member, the best thing you can do if you are working and you are using AI, if you have a special trick, there is a temptation to say it's my special trick. It keeps me productive. I get a good score on my annual review. I don't want to talk about it. I've literally talked to people who tell me that. I am not making that up. Don't do that. Don't do that because you're going to lose out if they switch out the vendor. Any part of your stock as an individual at work is then brittle because CFO may decide that the token
[20:50] cost for whatever AI you're using isn't good enough and doesn't know about your use case, etc., etc. Now, I'm not saying everybody's going to go to the CFO and the C CFO is going to process every single person's individual AI you. No, we we all know that's not going to happen. But socializing it to your manager, helping your manager understand the situation is going to help your manager better communicate software needs up the chain. There is a way to do that that is actually going to be productive. Okay, let's go back to the recruiting story at the start and let's start looking at specific software and
[21:20] the different jobs specific software does in the category because I think it's instructive and we're going to get super small here, super startupy. So let's jump in first with crust data. Crust Data has a case study about the firm, a two-person recruiting business that built a custom clawed skill using crust data's APIs instead of buying another packaged recruiting tool. A skill supplies instructions for the AI and API let software request data or actions. And the founders effectively
[21:50] supplied their own candidate selection criteria and then cross data just supplied the structured information about people in their work histories that fit that criteria. And so they work together to make that workflow useful. That is really instructive as we think about where buying decisions are going and where software is going in 2027. Effectively, the business decided to keep buying the provider of the data and the structured data agents could read and it chose to build a search workflow over the top in partnership with that
[22:20] data provider instead of going with a point solution, another SAS tool that digests the data and presents it in interface. Let's jump over to another part of back office. Now, let's look at payroll. Getting an AI to explain a paycheck is a very, very different proposition from replacing the system that gets people paid. So, Workday has described its payroll agent as using the company's configured payroll data and rules to effectively identify issues, suggest fixes for reviews, and improve
[22:50] the quality of the payroll experience. Now to get that right, they have to be building on a ton of existing knowledge and existing workflows underneath that interaction. Someone still needs to have records and configured behavior and approvals and dependable execution so that workday can deliver an experience like that. An agent may actually make a system like that more valuable because it helps people use it more effectively. Agents help us with complexity. And
[23:20] that's one of the huge patterns I'm seeing at every level in business. Whether you're a tiny startup or whether you're a massive company, agents are being used to help people cut through corporate complexity. And so that brings us back to what you're buying when you buy software. Right now, the recruiting example I shared gives us a way to talk about the difference between a useful data supplier and a workflow a customer could build. One of the patterns we see there is that you may be in a position as a data supplier to help move up the
[23:51] stack and partner with your customers to build solutions that work for them because AI has made solution building so cheap. Payroll is a bit of a different beast, right? I don't know of any startup that wants to handcode their payroll rules because the compliance risk is so high. And so in that situation, just giving people the comfort that you have the rails rock solid for that high compliance use case and they can sort through the complexity of bringing someone on in a new state simply and they can get the execution
[24:21] done. That is a huge play to make. But do you see the common thread here? The common thread is that you need to understand your position against the data layer, your position against the application layer, and your position against the agentic layer. Those are three different layers. And I have been talking about individuals at work. I've been talking about leaders at buying companies. And I've been talking about companies that build and sell software. And we're all in one of those categories. All of us have to be able to
[24:51] talk about our work in terms of the data layer, the application layer, and the agentic layer. If you're at work as an individual, you have to be able to articulate how you use AI and which tools you use and why the data layer matters to you, why the application layer matters or does not matter to you, and how you're using the agentic layer to accelerate. If you're a buyer, you have to understand the complicated and fingerprint-like pattern across your business. Every business is unique. Your data situation is unique. Your
[25:21] application situation is usually like a mosaic. It's very complicated and the layer over the top, the way people use AI tends to be even more complex than that because it tends to be tiny teams or individuals that are using AI in a particular way. And you have to make sure if you simplify and you start to coordinate across a big piece of that that you're not cutting out a bunch of productivity you didn't know you lost. And then you have to be able to understand the seller side, right? If you are buying, you have to deeply understand what your software vendors are offering. And I think so often the
[25:51] conversation there stops at do you have AI? What is the AI? Do we have options to not give our data away? That's just the start of the conversation. What we really need to talk about goes back to do you have the data layer figured out? Do you have an application layer that makes sense now? And do you have an agentic layer that gives us flexibility? And that's really like if you want to talk about the keys to the kingdom and how it makes sense to sell and build software. Why I am optimistic about the future of software, it's because I think we finally get a chance to reinvent what
[26:23] software means. We're not going to be bored as software builders, guys. It's so much fun, right? We're going to have so much fun as product people, as engineering people, as sellers figuring out how to position this because we get a chance to say this is where we are adding value in a way we never could before. And I think that the crust data story is a wonderful example of that. You can move up the stack and say we can add value here. We can help you with your workflow construction because we understand how structured data needs to be fed to agents. Now you might not think traditionally that a SAS vendor
[26:53] would be really really good at that. But crust data is an AI startup and so they are good at that. And so when we have these conversations, when we think about what we're selling, when we think about how we use applications, I'm basically encouraging us to take apart and be reflective about what we're building, what we're buying, what we're using so that we understand what the core elements are. Now you might think n this basically means the application layer is dead. It is not as simple as saying the
[27:24] application layer is dead. Why? Because applications have to be decomposed into interfaces and workflows. The workflow on top of the data can be highly valuable. And this is why I gave you the workday story. The workflow that workday operates at payroll company ADP is another payroll company. That workflow is highly compliant. That workflow is not going anywhere. That is an application. It's not data. It's structured workflows on top of data. The
[27:55] interface may or may not be there. The interface may shift dramatically. I do not think that we will all be using the command line all the time. I have seen enough knowledge workers who are scared to death of the command line. That is not happening. But absolutely, we are going to see AI woven into traditional interfaces more and more and more over the next few months. And companies that deliver structured workflows are going to be able to offer that with a lot of confidence. And so really, if you're in
[28:26] the application conversation, the question you need to ask yourself is, how do I deliver extraordinary value with my workflows that people do not want to touch as buyers? And if you're in the buying conversation for workflows, you have to ask yourself, what is the cost of allocating my engineers on my AI native effort to internal back office operations? Is it worth it or do I want to allocate them to my competitive advantage? And that is
[28:57] often the real conversation we're having around investment in AI. Where do we invest the marginal euro, the marginal dollar, where do we decide we're going to go after this particular application and it's going to deliver extraordinary value? And so when we talk about the companies that are on our home screens and we log in as an employee and we see, you know, the Microsoft icon, we see Teams, maybe we see Slack, right? Maybe it's a Gmail company. Whatever it is, when you log in and see those companies and you're like, why did my company
[29:27] choose this? This is the conversation that's starting to happen and it's going to affect you. Because if you have a particular way you use claude code, if you have a particular way you use codec, if you have a particular way you use co-pilot, those are all parts of the way business is generated at that company. And if people don't understand the way your workflow works, that's going to matter. And so what I want to encourage you to do is if you are an individual, if you are working at one of those
[29:57] companies and and you are listening to me and you're like, Nate, this is all really heady stuff like what do I do? Talk about the way you use AI. Talk fluently about the way you use AI. start to understand how to talk about the data you're using, the workflows you're using, where you're finding workflows less useful when you use an actual software application with mouse clicks and everything, and when you don't, because the more you share what's working, one, you make your whole team better, which is a huge issue with team
[30:27] productivity and AI. And two, you are going to enable the company to support you better over time. Now, there's always going to be the people who come back in the comments and say, "I have a toxic manager. or they hate AI, this and that. Look, I I know we all have some of those difficult spots, but on the whole, this is the better approach. If you're in leadership at a company that buys software, and almost every company buys software, this is essential for you to understand. You got to understand the fingerprint usage. You have to dig in with teams. I keep telling leaders, do
[30:57] not sit there at the high level, at the board level, and say, "We understand AI at the company. We just slap AI over the top." That is not going to work. You have to dig in with teams and understand what they're actually doing. And then you're in a position to talk about the data layer and the workflows you need done and the application layer and the agentic layer fluently. You're in a position to say this is what we actually need, especially if you've thought about the outcomes you're driving. And then if you're on the we sell and build software site anywhere from a giant established
[31:27] company like Salesforce or Microsoft all the way down to tiny startup like crust data you want to be in a position where you are saying this is our answer not today but this is our answer in 2027 when it comes to software you are effectively selling the future people who are buying you now are buying a belief that you will stay upto-date on AI and that what they are getting is relevant when the next AI model comes out tomorrow. And so you need to be able to talk about the idea that your
[31:58] workflows are set up in such a way that they're plugandplay. It's easy to plug AI into them. They're set up in such a way that the data layer is easy to access and get to from a variety of interfaces. They're set up in such a way that agents can read it effectively. And if you deliver agent, it's a multiplayer system. And so your agent is great at X-P specific workflow or X-P specific understanding of data and other agents can come in and also be helpful. That is a winning proposition. That is a proposition that I see coming out
[32:28] organically across different parts of the spectrum. And I think one of the most interesting questions in 2026 and 2027 is which companies are going to be able to most clearly articulate that when you ladder all of those layers together, the differential value is actually going to accelerate the business because businesses increasingly are buying outcomes. And that should matter for you, the individual. They are not looking for you to adopt AI. Spoiler alert, that's the baseline assumption.
[32:59] They're looking for all of us to be able to accelerate the entire business. And that is a scary proposition if you're at work because you think, is my team actually accelerating the business? Do they actually drive to outcomes? But I'd rather you hear it from me than hear it in a difficult annual review. So that's actually the conversation that's happening. That is actually what leaders are asking themselves about at these businesses is how is AI actually accelerating our outcomes, our sales, our revenue, our profitability, whatever it is. And that is the position that
[33:30] successful sellers need to take and they need to have honest answers. Okay. So what did we learn? What did we get to in this video? What do we think about? Number one, we have to understand in 2026 and 2027 how complicated the software stack is becoming. It is complicating the way buyers buy. It is complicating the way sellers sell. And it is complicating the way you and I have to do our job. And so I started this whole video talking about someone doing her job in recruiting. It's a real
[34:01] person I talked to. I just anonymized it a bit. In that world, the question that she had to face was, "How do I communicate that the AI I worked with, which was a paid AI provider, is so essential to my job, that if I go back, we're going to lose weeks and weeks of progress on these candidates. I'm going to have to reinvent the whole workflow and it just won't be the same. She had to have a difficult conversation with her manager about that because her manager hadn't realized that there was
[34:31] so much locked up in the relationship between the AI she was using and the data and the application and it was a mix of computer use and it was also a mix of the directly querying the data. It was like her own unique combination and she'd built skills for it and everything else. There are so many of us in that category and my challenge to you is please have that conversation proactively and soon share the way you work. So there you go. That is the state of software. That is how agents are complexifying every part of the buying
[35:01] and selling stack. But my goal here was to help you think through and understand with a really clear framework. This is what actually matters, right? You have to understand how your data layer works. Who's providing data? Who's providing structured workflows on top of data? who is providing maybe an application experience on top of that and how flexible is that? Who's providing an agentic layer over the top of that? How do you think about how those come together and drive outcomes? And I got news for you that used to be something that only a few people in a few companies had to think about. We all
[35:31] have to think about it now because agents are forcing us to vertically integrate stack and all of us are now having to think about our entire stacks and everyone is now a decision maker. I started that recruiting story because effectively she as an individual was a decision maker in how AI is used at her company. That's true everywhere now and we have to pay more attention to it. Okay. Tell me the decisions you're making with AI. Tell me your favorite AI applications that are small, that are tiny, that are like cross data. Tell me your application stories that show how
[36:03] we're moving away from the interface. Talk to me about the disasters of vendor acquisitions. I want to hear about situations where someone bought an AI story and it did not work out. Love to see the comments. It's going to be fun. And I'll see you next time.