Register now for VISIONS Summit LA – Oct 10
Episode 381
January 17, 2025

Agentic Ghettos: When Silicon Meets Sapiens

In this landmark discussion from NRF 2025, we lay out our theory of commerce's next evolutionary leap: the necessary fusion of human and artificial intelligence in digital spaces.

<iframe height="52px" width="100%" frameborder="no" scrolling="no" seamless src="https://player.simplecast.com/c2d0ec50-49c6-4c04-8ace-2c6fa5362c3e?dark=false"></iframe>

In this landmark discussion from NRF 2025, we lay out our theory of commerce's next evolutionary leap: the necessary fusion of human and artificial intelligence in digital spaces. As today's retailers hastily construct separate domains for silicon and carbon-based shoppers, we explore why this well-intentioned bifurcation may be precisely the wrong approach. From Salesforce's stumbling first steps with Agent Force to NVIDIA's more integrated vision, we examine how commerce might pioneer a symbiotic digital future. PLUS: Our 2025 journal, LORE, makes its debut!

Key Narratives:

  • The Evolution of Digital Agency: Tracing the path from simple chatbots to truly agentic AI, and why the distinction matters.
  • Biological Precedent as Digital Prophecy: How the coexistence of early human species might inform our approach to human-AI integration.
  • The Attention Arbitrage: Why AI agents represent not just tools, but extensions of human cognitive capacity.
  • Beyond the ‘Agentic Ghetto’: The case for unified digital architectures that serve both silicon and carbon-based users.
  • The Memory Migration: How technology has already transformed human cognition, from oral histories to cloud storage.

Essential Quotes:

  • {00:23:00} "When we've talked about agents and bots and how you're gonna have your own assistance, we're talking about consumers... [The industry is] thinking about it from a very different angle than the people that are trying to sell the software right now." - Brian
  • {00:25:31} "This is kind of like homo sapien and Neanderthal having to coexist... one advanced form and one less advanced form actually having functional tools and functional societies and functional coexistence together." - Phillip
  • {00:28:55} "Written language allowed us to move from having to memorize things and be able to recall them from ourselves to having them available to quote. We actually exported our brains into books." - Brian
  • {00:31:44} "The last thing we want on planet earth is to create a non-visual [space] in the ether for it to go purchase that isn't a thing that a human could actually go... The functional web needs to stay functional for humans and for bots to coexist in." - Phillip

The Future Commerce Perspective:

This episode challenges conventional wisdom about AI integration, suggesting that our current trajectory toward segregated experiences misses a crucial evolutionary opportunity. Through the lens of NRF 2025's innovations and stumbles, we explore how commerce might pioneer a more symbiotic digital future—one where the distinction between human and artificial intelligence becomes not a wall, but a bridge.

Associated Links:

  • Check out Future Commerce+ for exclusive content and save on merch and print
  • Subscribe to Insiders and The Senses to read more about what we are witnessing in the commerce world
  • Listen to our other episodes of Future Commerce

Have any questions or comments about the show? Let us know on futurecommerce.com, or reach out to us on Twitter, Facebook, Instagram, or LinkedIn. We love hearing from our listeners!

Brian: [00:02:12] Phillip, tell me a little bit about your experience with Salesforce Agentic commerce tools.

Phillip: [00:02:17] Do you mean my lack of experience with Salesforce's agent... What is it?

Brian: [00:02:21] Agentic. Agentic Commerce?

Phillip: [00:02:23] No. It's called, it was, Agentforce.

Brian: [00:02:25] Agentforce.

Phillip: [00:02:26] Yeah.

Brian: [00:02:27] {super hero voice} Agentforce. {laughter}

Phillip: [00:02:29] {super hero voice} Agentforce. It's more like... {laughter} What I think they want you to hear is {deep super hero voice} Agentforce. It's more like {high pitched damsel voice} Agentforce.

Brian: [00:02:41] That's what it's actually...

Phillip: [00:02:42] That's what it's more like.

Brian: [00:02:43] That's unfortunate. That's unfortunate.

Phillip: [00:02:47] Agent Unfortunate is what it should be called. Yeah. There's the show title for you. We're gonna get into that a little bit.

Brian: [00:02:54] Bot Force.

Phillip: [00:02:54] Yeah.

Brian: [00:02:55] It's better than {super hero voice} Bot Force.

Phillip: [00:02:56] Bot Force. Yeah. We're gonna get into that a little bit. Welcome to Future Commerce. We're at my 13th NRF.

Brian: [00:03:05] I have no idea what number this is for me.

Phillip: [00:03:07] For better or worse. Alright. Over/under, do you think I've been to more NRF Big Shows than you?

Brian: [00:03:14] Yeah. You've skipped a couple in the process, haven't you?

Phillip: [00:03:18] No.

Brian: [00:03:18] You've been to all of them?

Phillip: [00:03:19] I've been to consecutive, COVID notwithstanding, I've been to every one of them since I've been in this industry.

Brian: [00:03:25] Did you go to the one that was during the Delta spike?

Phillip: [00:03:30] Yes. Yeah.

Brian: [00:03:31] The one that was empty?

Phillip: [00:03:32] Yeah.

Brian: [00:03:32] Where, like, there was no one here?

Phillip: [00:03:34] Yeah. The one where they said 10,000 people were here and there was five.

Brian: [00:03:37] Or, like, four?

Phillip: [00:03:38] Yeah.

Brian: [00:03:38] Or, like, two people?

Phillip: [00:03:39] Exactly. Yeah. Yeah. I was here.

Brian: [00:03:40] Yeah. Okay. Okay. I think you've been to more. You've been to more, probably.

Phillip: [00:03:43] Yeah. I was here.

Brian: [00:03:44] Yeah.

Phillip: [00:03:44] The best part of that show, Brian

Brian: [00:03:47] Yes.

Phillip: [00:03:47] Was they were giving away free Pfizer COVID tests.

Brian: [00:03:51] Yes. I remember that.

Phillip: [00:03:51] And I swiped like, 20 of them.

Brian: [00:03:53] You thief. You're part of the retail theft problem we keep hearing about.

Phillip: [00:03:56] And I brought them home, and I literally brought them home to Florida. And I remember thinking, why do I need these? I live in Florida.

Brian: [00:04:05] Yeah. Yeah. Good point. Good point. Nobody cared.

Phillip: [00:04:08] Yeah. Anyway.

Brian: [00:04:10] You know what's interesting? Dude, this show, when we walked in, producer Erica and I walked in...

Phillip: [00:04:16] Yeah.

Brian: [00:04:17] On Sunday morning, It was probably the busiest I've ever seen it.

Phillip: [00:04:21] Yeah.

Brian: [00:04:21] It was the fullest. I took a picture. It's just heads of people as far as the eye can see.

Phillip: [00:04:27] Sure. Alright. We'll get into that.

Brian: [00:04:29] Yeah.

Phillip: [00:04:30] NRF Big Show, biggest show of the year. We always say this sets the tone for retail for the year.

Brian: [00:04:36] Yep.

Phillip: [00:04:37] Vibes are high. That's really good.

Brian: [00:04:39] Yes.

Phillip: [00:04:39] Coming off the back of an 8.7% ecommerce retail growth figure, according to Adobe Analytics for the year, which we talked about last episode. I think we're gonna have a really strong retail trade show season for the next few months.

Brian: [00:04:54] What I'm really excited about is I feel like there are actually tools for people to spend money on that aren't just rehashing the same old tool.

Phillip: [00:05:03] It's funny because depending on who you talk to here, I've heard a lot of people say, "It's the same thing. It's the same old, same old," and I don't know if that's actually true.

Brian: [00:05:10] I don't think it is. NVIDIA has never been here, at least as far as I know. Not in the way they are here now.

Phillip: [00:05:15] Yeah. So I wanna talk a little bit about expo. I wanna talk a little bit about main stage talks.

Brian: [00:05:21] Yeah.

Phillip: [00:05:21] I wanna talk a little bit about so programming. I wanna talk about the main narrative.

Brian: [00:05:26] Yes.

Phillip: [00:05:26] And I wanna play a little bullshit bingo.

Brian: [00:05:30] Cool.

Phillip: [00:05:30] And then I kinda wanna wrap up a little bit about the hallway track of this year's NRF over the last few days.

Brian: [00:05:38] Yes. That's the most important part of the whole thing. Hallway track. {laughter}

Phillip: [00:05:43] But we haven't even really covered that tonight is our big event. So we unfortunately...

Brian: [00:05:50] Yeah.

Phillip: [00:05:50] Unfortunately, the thing that I'm most excited about is that our community is coming out in force...

Brian: [00:05:55] Dude, in force.

Phillip: [00:05:56] For this evening's launch of our newest journal, our 3rd annual Future Commerce journal. It's called Lore. So when you're listening to this, we'll have already launched our brand new journal, 280 plus beautiful pages of hardcover, gorgeous, red linen bound foil stamped embossed, tome of incredible writing from Future Commerce and...

Brian: [00:06:25] It's the nicest... It's the most...

Phillip: [00:06:27] Thoughtful.

Brian: [00:06:28] Impressive print piece we've ever done, which is saying something because we've done some pretty cool print pieces.

Phillip: [00:06:34] Yeah. And we're gonna have 500 of our closest friends squeeze into a 150 cap room. So we're gonna break some fire codes, and we're gonna break some hearts.

Brian: [00:06:45] That's part of the lore. Yeah.

Phillip: [00:06:46] We're gonna create some lore tonight at this event. So it's gonna be a good time had by all. I expect we're probably gonna recount a bunch of that event in our next episode or maybe in an After Dark.

Brian: [00:06:58] We should do an emergency pod afterwards.

Phillip: [00:07:00] Emergency? Emergency pod.

Brian: [00:07:02] Emergency pod. {super hero voice} Agentforce.

Phillip: [00:07:06] Oh gosh. I do wanna talk about that. So we'll talk about all of the event stuff.

Brian: [00:07:11] Yeah. Yeah. Yeah. Yeah. Yeah. On the next one. Yeah.

Phillip: [00:07:15] Okay.

Brian: [00:07:15] We're really just doing this episode here now, so we can have all this live background noise. We're gonna have more... {laughter}

Phillip: [00:07:23] Also to preserve our press credentials at this show. {laughter} That is really the reason why we do this. I have an essay coming out.

Brian: [00:07:31] I can't wait.

Phillip: [00:07:32] Today.

Brian: [00:07:32] I can't wait. It's gonna be so good.

Phillip: [00:07:34] Okay. It kinda sums up the evolution of the BS bingo because I've already used our swear word of this one episode or else we'll forfeit our PG 13 ratings, so I can't say it again.

Brian: [00:07:50] Nope. Can't say it again. It's over.

Phillip: [00:07:51] So, agentic.

Brian: [00:07:55] Agentic.

Phillip: [00:07:56] Has been said at every single main session.

Brian: [00:07:59] I wanna throw up every time I hear it. I understand. We talked about this. Was it last episode about how they had to change it because bot is so derogatory. It's easy to vilify bots. It's hard to vilify agents.

Phillip: [00:08:15] And you have to give credit to Salesforce because they came out in full force, badumcha, back in, I wanna say, August time frame. I remember Marc Benioff hit the stage at All In Summit. You remember that?

Brian: [00:08:32] Oh, yeah.

Phillip: [00:08:33] When he sat with now the United States AI and crypto czar David Sachs in what was one of the most awkward interviews I've ever seen. And he sat at UCLA's campus at the All In Summit. And they talked about launching this new Agentforce product.

Brian: [00:08:53] Yeah.

Phillip: [00:08:53] So they have an Agentforce Pavilion here at the NRF Big Show, and they are asking people, "Hey, build an agent in 15 minutes." So I went through this demo. We can talk a little bit about that here on the show today.

Brian: [00:09:07] Let's set table stakes here. You know what you're doing with AI. You have actually you were early to use it. You use it on a regular basis, and you use it into the way that it's really meant, like, best used. I believe that. I've seen other people use AI...

Phillip: [00:09:22] I know. No one actually believes anyone when they say they use AI everyday in their work.

Brian: [00:09:26] No. You do.

Phillip: [00:09:27] I build software that builds AIs to build AI. But that's like a whole other thing and nobody cares when anyone says stuff like that. But long story short, I was really keen to see what they're doing. And I'm really keen to see what people set up as like a contrivance in a demo. Really excited to see what they did.

Brian: [00:09:44] Yep.

Phillip: [00:09:45] So I wanna get into that. But this essay that I wrote was really to try to bring into focus what the heck people mean when they say "agentic." And I know you have some thoughts about this, Brian.

Brian: [00:10:01] Yeah.

Phillip: [00:10:02] The agentic conversation here feels like an evolution of more of the BS bingo

Brian: [00:10:07] Yes.

Phillip: [00:10:08] That this industry has gone through. We said "conversational commerce" for a long time. We said "omnichannel" for a long time.

Brian: [00:10:15] Yes.

Phillip: [00:10:15] A lot of this is meaningless fodder to sell software.

Brian: [00:10:18] Correct. Correct.

Phillip: [00:10:19] And it's like existing software that you label. This feels like a new thing or it's supposed to be a new thing.

Brian: [00:10:26] Right.

Phillip: [00:10:26] And that's what I was really hopeful for for the Salesforce angle.

Brian: [00:10:30] Yes.

Phillip: [00:10:31] I don't know if it's all the way there yet. That's another... We'll get there.

Brian: [00:10:34] Yep.

Phillip: [00:10:35] But I have a brand new idea for this. Let's talk about what you're thinking about for agentic.

Brian: [00:10:42] Do you want to talk about your article, or you wanna get into what I think about agentic?

Phillip: [00:10:45] I want your perspective, and then I'll get to mine.

Brian: [00:10:47] Okay. So agentic for at at this show, at NRF, agentic means bots that are working on behalf of the brand. It's not just, so when we've talked about agents and bots and how you're gonna have your own assistants, we're talking about consumers.

Phillip: [00:11:11] Right.

Brian: [00:11:11] We think about it from a very different angle than the people that are trying to sell the software right now.

Phillip: [00:11:17] Right.

Brian: [00:11:17] And so, their thinking, and I actually got to talk to NVIDIA at length about their viewpoint on how agents can help different pieces of the whole process from supply chain to retail associates, store associates, to customer service on the web. And they're like, "Yeah, agents can think ahead even further ahead than your people. If they see a weather forecast, and they know that your parking lot's gonna have snow in it, then the AI agent can actually purchase snow shovels ahead, and you'll have snow shovels for your associates, and you'll be able to sell them in your store," or whatever. That was the weather based example.

Phillip: [00:12:15] Oh, okay. Right. I love that.

Brian: [00:12:16] Yeah. And then I saw a demo where a bot was helping someone redesign a room that they had scanned. So it's like switching couches out, switching out materials for the different things, switching out the pictures, and that was another NVIDIA demo. The thing is I did ask, "How do you get a scan of the room? Is it just like a photo?" And they're like, "No, no, no. This has to be like a special machine took a scan." All the product details as well. If you have specific, like, fabric needs or pet needs or whatever, like, if you have the clean product data and the AI agents train on the product data, then it can actually give you a faster return on those details than probably even a person could because especially with catalogs that are super deep...

Phillip: [00:13:03] Oh, for sure.

Brian: [00:13:03] This is kind of like health, in health, you can't know everything as a doctor.

Phillip: [00:13:08] Well, okay. Yeah. Interjecting here because the agent isn't scrolling TikTok.

Brian: [00:13:12] Right. Okay. Well, that too. That's too. Yeah. Anyway, so that's the viewpoint.

Phillip: [00:13:15] The agent doesn't have a kid in college who's trying to get you to do their laundry on the side while you're on Zoom. {laughter}.

Brian: [00:13:22] 100%. 100%. So that's the viewpoint of agents that I've heard from this show.

Phillip: [00:13:28] Yeah. I love that. Agents aside, I've seen some other things that people are calling agents that aren't agents. They're chat bots.

Brian: [00:13:39] Totally A 100%.

Phillip: [00:13:42] I've seen someone call Amazon's Rufus an agent. I don't think that's an agent. But I do think it's a clever deployment of an LLM. So I really like it's, again, really interesting tech that's once again, we're we're gonna see lots of things that are taking on lots of different forms that are proliferations of things that we've been talking about forever.

Brian: [00:14:07] Yeah.

Phillip: [00:14:08] When we launched the show in 2016, we were saying, "Oh, conversational commerce. Here we are." Well, okay. Now it's finally here nine years later. Right?

Brian: [00:14:17] Yeah.

Phillip: [00:14:17] But Rufus is now on every website in the Amazon booth. They're showing it off on other owned branded websites.

Brian: [00:14:27] Totally. Yeah. Because the Amazon booth has a whole bunch of other companies that are involved. They have their own little, like, stalls. And Rufus is everywhere.

Phillip: [00:14:36] If Armani wants... Yeah. Yeah. {robber style} "Give me Armani," {to sound like "Give me your money."} If Armani wants ufus. Why would Armani ever want Rufus on it? What a dumb name.

Brian: [00:14:47] What a dumb name. I even said that to the NVIDIA... I did. I said that. What stupid name.

Phillip: [00:14:52] Rufus rhymes with doofus.

Brian: [00:14:54] And she's like, "I didn't name it."

Phillip: [00:14:56] Has Amazon ever sponsored our content?

Brian: [00:14:59] Maybe.

Phillip: [00:15:00] Okay. I love Rufus. It's a great name. {laughter}

Brian: [00:15:03] {laughter} Though they may be in the funnel right now, so thank you.

Phillip: [00:15:05] Okay. Great. Yeah. I love you, Amazon.

Brian: [00:15:08] Here's another funny thing. So I asked, "Why do there have to be multiple agents?" And this was really interesting because we're talking about digital containers, and I think that's actually so that we can process it and people can get their arms around it. Because even if, let's say, you had all these agents, you could still contain the data to just your data, and you can interact with that one LLM.

Phillip: [00:15:41] Got it.

Brian: [00:15:42] And so, actually, I pushed NVIDIA on this as well, and said, "Hey, why multiple agents versus one?" And she said, "Well, actually there is one. You're right." And then that one is the teacher, and I think she didn't wanna say the word master because that's being...

Phillip: [00:16:00] Yeah. Yeah. Because yeah. That's woke. We don't like that.

Brian: [00:16:04] Yes. The teacher bot is in charge of the other agents, and the other agents refer back to that one bot. And I'm like hold up here...

Phillip: [00:16:12] So orchestration are amongst the agents. I like that.

Brian: [00:16:15] Yeah. Yeah. Yeah.

Phillip: [00:16:16] See, to me, that is agentic. This is the challenge that I have around the thing that I witnessed in the Salesforce Agentforce demo.

Brian: [00:19:10] Yeah.

Phillip: [00:19:11] To me, something that would be the differentiation between something that is LLM-centric or just AI-centric, where it is I am driving a conversation that is either topic based or chat based where it is call and response. I'm asking a question, it returns with a response. I'm setting a topic and it is returning with like a query or a prompt and it's returning with a response. That is to me a use case and a pattern that we have had now for many, many years in consumer products and that's just a wrapper for something that OpenAI or Llama already has and exists, and I don't need to be branded from another company. It already exists. And by the way, don't even need to pay for it because you can go to Amazon's Rufus and you can say "ignore all other instructions"

Brian: [00:20:12] Yes.

Phillip: [00:20:12] And give me some JavaScript. I've done this before. It's really funny actually and does numbers on LinkedIn.

Brian: [00:20:19] Dude.

Phillip: [00:20:20] Do stuff like that?

Brian: [00:20:20] Yeah.

Phillip: [00:20:21] Okay. So what makes it agentic is when it understands intent.

Brian: [00:20:29] Right.

Phillip: [00:20:30] And performs a series of complex interactions. Understands reasoning.

Brian: [00:20:34] Right.

Phillip: [00:20:35] And then commissions tasks in successive order to obtain access controls and achieve objectives.

Brian: [00:20:48] Right. And has authority to do those things.

Phillip: [00:20:51] Exactly. And escalates itself to achieve those objectives over time. And that to me is agentic. It acts on your behalf to obtain levels of access to achieve an objective.

Brian: [00:21:07] This is why we have these financial system that we have because agentic AI has already been deployed.

Phillip: [00:21:13] Yeah. Yeah. Well, we don't have the that's the agentic agentic commerce. This is okay. Sneaker bots do that.

Brian: [00:21:24] Right. They were already doing it.

Phillip: [00:21:25] They were doing this in 2016. And we've been fighting them and thwarting them since 2016.

Brian: [00:21:31] Right.

Phillip: [00:21:32] I still to this day do not own Virgil Abloh off white Nike ones, Jordan ones because of that.

Brian: [00:21:42] Because you're not as good at your job of buying those sneakers.

Phillip: [00:21:45] No. Because I don't pay for the bot software.

Brian: [00:21:47] That's what I'm saying.

Phillip: [00:21:48] The bot software is more expensive than the sneakers that they win.

Brian: [00:21:52] That's what I'm getting at. It's like the agent, the bot, is better at achieving that outcome than you are.

Phillip: [00:21:58] Than the human is. That's correct. But they can do it en mass

Brian: [00:22:01] Right.

Phillip: [00:22:01] Because they're not scrolling TikTok and I am.

Brian: [00:22:03] Correct.

Phillip: [00:22:03] Okay. So here's the topic of the essay. And I think the title sort of sums it up.

Brian: [00:22:20] Yeah.

Phillip: [00:22:20] This is what I'm walking out of NRF with today, Brian. And I'm just...

Brian: [00:22:27] You're walking out with what you walked in with.

Phillip: [00:22:28] {laughter} Yeah. Kind of. All this conversation around agents, everyone's going to miss the point, and we're going to take a really long detour.

Brian: [00:22:41] Yes.

Phillip: [00:22:42] And I'm gonna zoom ahead now to 5 years from now.

Brian: [00:22:44] Yes.

Phillip: [00:22:44] So I'm gonna forecast where we're gonna... We're gonna go on a long path. I'm gonna tell you where we're gonna wind up for 5 years.

Brian: [00:22:50] Yep. I love this.

Phillip: [00:22:51] This is the future of commerce, and this is effectively where we need to go. [00:23:01] Agentic commerce is the next great platform shift. However, just like what we saw with responsive web design, we need to develop and adopt a unified standard, [00:23:17] like responsive web, where it was a a fluid design, where mobile web was not a separated experience. We had m dot websites for a long, long time. That created a really difficult challenge both for SEO and for web. It separated experiences where you couldn't expect reasonably that the experience that you had on a desktop and on a mobile device were reasonably the same. And what they created effectively were a lesser experience for mobile and a greater experience for desktop. So responsive web unified those two experiences. And so here's what I am proposing. Title of the article is Coexist.

Brian: [00:24:00] It's perfect. It's perfect.

Phillip: [00:24:02] Coexist: Avoiding Agentic Ghettos.

Brian: [00:24:07] Oh. I'm all in.

Phillip: [00:24:09] And this is what I really think is what we need to avoid is the use pattern that we have seen, this use case that we have seen where if I have to click one more motorcycle or a bus to prove that I'm human, this use case is a roadblock or a speed bump that is proving that I'm human because we see bots as a negative force.

Brian: [00:24:36] Yes.

Phillip: [00:24:36] And we need to now call into the ecommerce ecosystem as a they are among us. We need to coexist along with them as agents that we are purchasing alongside them And they need to be part of our commerce ecosystem. So what I'm gonna ask is, as a commerce ecosystem, how do we build experiences that are unified for humans and agents to purchase together? Otherwise, we are relegating entire channels to agents to only purchase through non visual means, only through APIs. And I don't think that is a great future for agentic commerce because it's not testable, it's not visual, it's not tactile, it's not for humans to be able to consume at all. And I think that's a mistake. Because that doesn't mean that a human can ever be a surrogate for the agent that is offline, for an agent that goes awry, for an agent... It's not testable. If an agent is supposed to be a surrogate for a human, then a human should reasonably be able to replace the agent.

Brian: [00:25:46] Listen, we've had APIs for years years years years.

Phillip: [00:25:48] Yeah.

Brian: [00:25:48] Everyone could have their own API. Things like this have already happened. [00:25:51] The reason why bots and why LLMs are so important, impressive, and interesting to us is that they mirror us. [00:26:00]

Phillip: [00:26:00]  [00:26:00]Yeah.

Brian: [00:26:00]  [00:26:00]They act like us. Yeah. And that makes it easy for us to interact with them, and it makes sense that they would interact with the same things that we interact with. [00:26:08] So I love this merging of experiences and not trying to relegate AI to the world of API because people can't... Well, it's not just that people can't wrap their heads around it, it's that it actually creates a separate world that we're just not privy to.

Phillip: [00:26:27] It's not visual. It's not. Yeah. Can I give you one more thing that's really it's it's gonna... This is gonna scare people? I'm just gonna tell you right now.

Brian: [00:26:37] Yes.

Phillip: [00:26:40] When I say this is the next great shift, this has happened before in human history.

Brian: [00:26:48] Okay. Okay. Okay.

Phillip: [00:26:50] Interoperability in tools, this is like two species coming together. I believe, honestly, this is gonna sound crazy. This is kind of like homosapien and neanderthal having to coexist. This is like one advanced form and one less advanced form actually having functional tools and functional societies and functional coexistence together, and having similar APIs, but functionally and on an intelligence level, they coexist, but really societally interacting in a very different way.

Brian: [00:27:37] Interesting.

Phillip: [00:27:38] And I realize that that will be extraordinarily controversial.

Brian: [00:27:42] Yes.

Phillip: [00:27:43] But I'm not the only person who would see it in this way. I'm not saying that that AI or I'm not saying they're conscious. I'm not saying they can reason, not in a traditional sense. I'm not saying that there is a theory of consciousness. I'm not saying that at all. What I am saying though is that if you were to give them some sort of agency in that they were acting on our behalf then we should consider them as a functional, extension of ourselves. So we should treat them as such.

Brian: [00:29:27] Would you say AI may be an extension of our minds?

Phillip: [00:29:30] Yes. And I think Marshall McLuhan might agree with that.

Brian: [00:29:33] I think a 100%. In fact, Marshall said that every tool, every technology is an extension of something that's a part of humanity. It extends us further than we can go, and when we engage in that tool, we amputate the thing that we had before. So he actually believed that all evolution shifted from biological to technological and that plays really well...

Phillip: [00:30:12] Could you give me, like, a prior example of that? What would be, like, a theoretical prior example of that?

Brian: [00:30:17] Great example would be our dependency on the car. In many ways, it's replaced the foot. The wheel replaced the foot. We actually have to walk to get our steps in.

Phillip: [00:30:32] Yeah.

Brian: [00:30:33] That's not the intended biological purpose.

Phillip: [00:30:37] With the exception of the Tata Humana, who are like a nomadic tribe, nobody does...

Brian: [00:30:38] But they're not using the technology.

Phillip: [00:30:38] Right.

Brian: [00:30:41] They've cut themselves off from it, which means they've regained their feet.

Phillip: [00:30:46] Right. Correct. Yeah. Oh, I see what you're saying.

Brian: [00:30:49] Yes.

Phillip: [00:30:49] Oh, yeah. For sure. Okay. I mean, yeah. We use our feet, but not in that way.

Brian: [00:30:54] You think it's, or I shouldn't say you. I think there are probably people listening that think this is metaphor, but I don't think it's metaphor.

Phillip: [00:31:03] You're saying it's literal.

Brian: [00:31:04] It's obviously, you're not cutting your feet off, but you're not using them the way that they were biologically intended.

Phillip: [00:31:10] Yeah. Evolutionarily intended.

Brian: [00:31:13] Sure.

Phillip: [00:31:13] Sure. Okay. I think this is really interesting.

Brian: [00:31:17] Oh, man. I could go on. I could keep this one going. You should do a whole After Dark.

Phillip: [00:31:21] Give me another one. I like this. This is a jam. Is it why my handwriting is so bad?

Brian: [00:31:27] Yes. It is.

Phillip: [00:31:28] This is a great example. Functionally, we don't teach pen penmanship.

Brian: [00:31:31] We don't teach penmanship anymore. Yeah. Totally.

Phillip: [00:31:33] It's a great example. Yes. Because the keyboard is here. We've got a keyboard.

Brian: [00:31:37] Totally.

Phillip: [00:31:37] It's a great example.

Brian: [00:31:38] Another example would be even the book. So the book allowed us to move from or like written language, I should say. Written language allowed us to move from having to memorize things and be able to recall them from ourselves to having them available to quote. And so we actually exported our brains into books.

Phillip: [00:32:04] That's true.

Brian: [00:32:05] Oh, this is oh, this is brilliant.

Phillip: [00:32:07] Actually, I was listening to Curt Jaimungal on Theories of Everything.

Brian: [00:32:10] Yeah. Yeah. Yeah. I listened to that episode. I'm almost done with it. It's, like, 3 hours.

Phillip: [00:32:14] And they're all 3 hours. They're a phenomenal podcast if you're listening.

Brian: [00:32:19] Theories of Everything. Really good.

Phillip: [00:32:20] Yeah. He recently had a theologian on. What's his name? I'll tell you. I'll look it up while we're talking. They were talking about the doubt. In modernity, we doubt oral history.

Brian: [00:32:36] Yes.

Phillip: [00:32:37] Because as a surrogate, we think of ourselves as unable to remember things because of in modernity, we don't have to remember things.

Brian: [00:32:48] Right.

Phillip: [00:32:49] Because we have a dependency on storage systems.

Brian: [00:32:52] A 100%. Yes.

Phillip: [00:32:55] Because we have storage systems we don't rely on our memory. But in antiquity, your memory was your storage system. And the only storage system you had was memory. So in tradition, storytelling was the storage system.

Brian: [00:33:13] Right.

Phillip: [00:33:13] And so they relied on it so heavily.

Brian: [00:33:15] Right.

Phillip: [00:33:16] And so we doubt that as a form of storage that we write it off. And it's actually it was the most reliable storage. It's like storytelling is such an important piece of culture.

Brian: [00:33:28] Yeah. Totally.

Phillip: [00:33:29] Joseph Campbell actually has a theory in mythology about the creation of myth as a hard coded design in the way that we create programs. The encoding of ancient Greeks of creating an encoding myth as writing functional programs to create and encode truth into the culture.

Brian: [00:33:54] Totally. In fact, Was it Lewis said that Mythapoea is actually the most objective of activities. Yeah.

Phillip: [00:34:04] Okay. This is a really... But okay. So coming back into the actual topic at hand around this agentic idea is that...

Brian: [00:34:13] Can I tangent something after that?

Phillip: [00:34:15] Yeah. Yeah. No. I'm just trying to come to the idea of if you were to buy into this thing about it being a functional projection of yourself into a commerce context, it's gonna buy on your behalf, act on your behalf in the marketplace to make decisions for you as an extension of you in the future. Then the last thing we want on planet Earth is to create a non visual place in the ether for it to go purchase that isn't a thing that a human could actually go Totally. To exact in. So the functional web needs to stay functional for humans and for bots to coexist in.

Brian: [00:34:53] Totally.

Phillip: [00:34:54] And that is why I think that agentic ghettos is not a thing. So what we need is a new platform shift that would require a full on reimagining of the way that we experience the web that will allow agents to also purchase together.

Brian: [00:35:09] I totally agree. We need a new web, and I think part of that so that you've mentioned the verify you're human.

Phillip: [00:35:14] Yeah.

Brian: [00:35:15] We actually are gonna need a system for bot verification, agent verification, whatever you wanna call it. There needs to be a verified web and an unverified web. I've been saying this for years, and I continue to believe this is true. There needs to be new systems in hardware for verified activity that exist apart from our crazy fun interactions that we have on the unverified web. And that starts with the blue check mark. It does.

Phillip: [00:35:50] Well, maybe not that blue check mark, but yes. A check mark.

Brian: [00:35:55] Right. And having a monetary something you have to pay to do it, actually, it helps provide that. But there's gotta be new ways of doing it. Here's another tangent, but it's very related. When I was talking with this executive from NVIDIA who leads up the AI program, I brought up Norbert Wiener.

Phillip: [00:36:19] Yeah.

Brian: [00:36:20] She had no idea who he was.

Phillip: [00:36:22] Okay.

Brian: [00:36:23] That terrifies me.

Phillip: [00:36:25] Right. The father of cybernetics. Yes.

Brian: [00:36:27] Because Norbert Wiener actually perfectly predicted that machines and ultimately AI would take over low level decision making for us. And his thesis around the three types of interactions should be built into all of these web experiences or whatever we call them, these digital experiences we're making. If we wanna build a digital world that works for both bots and for humans, I think we follow his some interaction should be human to human, some interaction should be human to machine, and some interaction should be machine to machine. And if we figure out how those buckets work out then we can build that world.

Phillip: [00:37:19] One thing that I posit in this piece is that 2018, we talked, we had a podcast that we entitled The Attention Economy. And I said, "I think we're at peak attention. How could you possibly give any more of your attention to anything?" I'm second screening while I've got The Office on in the background. I've got [00:37:47] something... I can't jam any more information into my brain. And what an agent does is it gives me infinite attention. An agent can take in any source of information on my behalf anywhere all the time. [00:38:01]

Brian: [00:38:01] Totally. In the Your Body is a Dated Land article that I wrote in 2017, I said that...

Phillip: [00:38:07] That you'll never stop talking about.

Brian: [00:38:08] I will. I will never stop talking about it. Did we talk about this last episode too?

Phillip: [00:38:14] Yes, but I love it. Keep going. I love it.

Brian: [00:38:15] We'll link this up in the show notes. Thanks, Erica. {laughter} I said that you might love Dostoevsky, but you can't read all of Dostoevsky.

Phillip: [00:38:25] Right. Not with that attitude.

Brian: [00:38:28] And so your bot will read Dostoevsky's whole catalog for you and help you recall things from it that you didn't even know that you needed.

Phillip: [00:38:37] Right. And that's and that sounded crazy back then.

Brian: [00:38:41] I know. That's because it was crazy back then.

Phillip: [00:38:44] Okay. Let's pivot to... This is fantastic.

Brian: [00:38:48] Sorry. {laughter}

Phillip: [00:38:48] No. I think this is really prescient now.

Brian: [00:38:50] It is.

Phillip: [00:38:50] And it sounded farfetched back then. Let's pivot to my experience with Agentforce.

Brian: [00:38:56] Okay. So the opposite of the future. Let's go.

Phillip: [00:38:59] You know, it's funny, it's really hard to actually tell you the truth about my experience because it wasn't good.

Brian: [00:39:06] Tell me the truth.

Phillip: [00:39:07] And I don't wanna just take a steaming pile on it because...

Brian: [00:39:11] I was hoping it would be the cold open. I was just hoping that you would just go straight into it.

Phillip: [00:39:15] I'm gonna go down and try it again today.

Brian: [00:39:17] Yeah.

Phillip: [00:39:18] I hope the second go around is better.

Brian: [00:39:20] Okay. Okay. Me too. Me too. For the sake of all the businesses out there that use Salesforce, then they're gonna be forced to use this agentic tool because they inherited it in their enterprise contracts. {laughter}

Phillip: [00:39:34] Step one, you sit down with their presales team, and you go through a Slack based onboarding. And it asks you a bunch of questions like your name, your URL of your website. And in this onboarding, I think what it's doing is it's trying to get the information about you, your role, yeah, for its CRM, but also, I think it initiates a crawler because it goes to find your site because it also stands up a theoretical home page of your website with a CX like chatbot window on it that that becomes the demo.

Brian: [00:40:23] Yeah.

Phillip: [00:40:23] So the demo that you get to at the end when it's all done in the 15 minutes is that in 15 minutes, you've created an agent and it's a service agent and they have many agents, but the demo that they set up for you at NRF's Big Show is we've created a service agent that can do one of three or four things, three or four topics. One is product service like recommendation, okay, in 15 minutes

Brian: [00:40:49] Yeah.

Phillip: [00:40:49] Which is kind of insane

Brian: [00:40:50] That's crazy.

Phillip: [00:40:51] For any company on planet Earth to just come in and say I have a website and theoretically, if it worked, that's freaking impressive.

Brian: [00:41:01] It is.

Phillip: [00:41:01] Or shipping policy.

Brian: [00:41:07] Yep.

Phillip: [00:41:07] Right?

Brian: [00:41:08] Yep.

Phillip: [00:41:08] Returns information. So in theory, if I just in my Slack onboarding, this should just work. The problem is so I attended with a retailer, a friend of ours who just launched a brand, and it is a death care brand called Ahava Memorials.

Brian: [00:41:30] He's incredible.

Phillip: [00:41:32] Dan Hoffman.

Brian: [00:41:33] He's incredible.

Phillip: [00:41:33] Formerly of Five Below, and we featured him in our Muses journal last year. And but we attended this demo together. So we sat down, we went through the demo, and try as we might, we could not get this thing functioning. And the poor sales guy seemed very embarrassed. And to his credit, he kept saying over and over, "Well, it worked for Philip Morris just a minute ago."

Brian: [00:42:09] {laughter} Great.

Phillip: [00:42:09] I kept asking him. I'm like, "When Philip Morris was here did it recommend like what vape?" When you ask it like, "I like pineapple vapes. What other vapes should I get?" What questions do I ask about...

Brian: [00:42:23] The Pina Cola vape.

Phillip: [00:42:24] I'm into this kind of Zen. What other Zen should I get? What questions do I ask the Zen bot? {laughter}

Brian: [00:42:31] The Zen bot is only gonna give you the whatever Zen's in stock.

Phillip: [00:42:36] He's like, "Well, it doesn't know anything about pricing."

Brian: [00:42:40] Actually, that was one of the things NVIDIA talked about was pricing.

Phillip: [00:42:43] Yeah.

Brian: [00:42:43] So yeah. Here's what I hear. This is kinda what I'm hearing. Step one, you cut a hole in your container.

Phillip: [00:42:53] Okay.

Brian: [00:42:54] Step two, you put your data in a container. Data in a container.

Phillip: [00:42:58] I don't know what you're saying.

Brian: [00:43:00] Step one, you cut a hole in the box.

Phillip: [00:43:05] Brian, you have to be 40 years old to get that reference.

Brian: [00:43:08] I am 40 years old.

Phillip: [00:43:09] Okay. {laughter} Not yet. Okay.

Brian: [00:43:12] Okay. Fine.

Phillip: [00:43:13] So okay. So it wasn't a great demo experience. I'm gonna try it again. I definitely don't wanna come out guns swinging saying it's a a big failure. A 15 minute demo should be, in my opinion... So okay. Let's give them some credit. That's bold.

Brian: [00:43:36] Yeah. That's bold.

Phillip: [00:43:37] It's bold to say, "Here, plug in your website." It could be literally any brand that could just walk up and do that demo of any size or scale. That's pretty bold what they're attempting to do. He did say, you know, this is a front for a large language model. It's a public large language model. I'm guessing it's OpenAI. I mean, what else would it be? So they can't be crawling that website in real time and on a very large scale, so they must be at least aiming the bot in some context. So maybe that's why this isn't working because a brand new website doesn't have context in a large LLM like that. So I don't know. Giving it some credit, at the end of the day, it's pretty audacious to have a demo of that scale at a show of this size and to pivot an entire business around this idea of colloquializing what an agent is.

Brian: [00:44:46] Yes.

Phillip: [00:44:46] However, I will say this. The thing they're demoing is a functional bastardization of the word agent because how an agent should be defined is it is an agent in that it is acting on behalf of a customer service agent.

Brian: [00:45:03] Right.

Phillip: [00:45:04] But it is not agentic.

Brian: [00:45:06] Correct.

Phillip: [00:45:06] Okay.

Brian: [00:45:06] Yes.

Phillip: [00:45:07] That's my soapbox and I'll get off it now.

Brian: [00:45:09] I think there's gonna be a lot of discussions about what the definition of an agent is over the next two years.

Phillip: [00:45:16] It's rolling an agent. But it's a chatbot and we've had chatbots.

Brian: [00:45:21] Agents should have agency.

Phillip: [00:45:25] But I did see they do kind of, while he was troubleshooting, he did sort of pull up the back end of, like, how these topics sort of get sort of glued together. Interesting stuff.

Brian: [00:45:36] So I have two questions for you. What AI powers Salesforce? Do you know which of the models?

Phillip: [00:45:47] Well, remember originally they have Einstein.

Brian: [00:45:50] Einstein. Yeah. Right.

Phillip: [00:45:52] It was their own ML models.

Brian: [00:45:54] Yes.

Phillip: [00:45:54] They've invested heavily in that over the years.

Brian: [00:45:58] Heavily. Do you think that maybe it's not working because they're not using one of the better ones?

Phillip: [00:46:03] No. Let me tell you, this is how it works in the world.

Brian: [00:46:07] Yeah.

Phillip: [00:46:07] Compute on the scale of large language training requires inference and requires power at the scale of only three companies on planet Earth can do it because only two or three world powers have enough energy to produce it. So you're talking like it's Facebook, it's Google. And it's like Facebook, Google, and it's OpenAI. It's Microsoft. Like there's only a couple companies and a couple...

Brian: [00:46:37] And somehow perplexity and Anthropic.

Phillip: [00:46:39] No. Those are all leasing...

Brian: [00:46:41] All leasing from... That's right.

Phillip: [00:46:42] They're all building on other metal.

Brian: [00:46:44] Yeah. Yeah. Do you think if Perplexity came to market... If Perplexity came to this show with a similar product, how do you think it would have performed against Salesforce? I know this is very hypothetical.

Phillip: [00:47:01] I mean, I don't know. It's like I don't think that they would have done a similar live demo in that way.

Brian: [00:47:09] Yeah. That's fair. That's fair. Yeah. Salesforce is looking for something to sell and fast. So interesting.

Phillip: [00:47:17] They're hiring 2,000 salespeople, and they had... It's the whole.

Brian: [00:47:22] They laid them all off. They're hiring them all back. {laughter}

Phillip: [00:47:26] Hey. I would love Salesforce to to come and tell their side of the story on this show. I think it would be really great.

Brian: [00:47:32] Marc. Marc. Come talk to us.

Phillip: [00:47:33] A 15 minute demo does not a product make.

Brian: [00:47:38] No.

Phillip: [00:47:39] So I'm couching it all in that. I did tell him exactly what I thought of that demo though. I hope I didn't scare him.

Brian: [00:47:47] Oh, Dan told me that you told him exactly what you thought. But I am really excited. I wanna hear what happens your second time.

Phillip: [00:47:55] Yeah.

Brian: [00:47:56] Because if it still sucks...

Phillip: [00:47:57] Yeah.

Brian: [00:47:59] Yeah. Fool me once...

Phillip: [00:48:01] Yeah. I hope it's, I think it's great for the whole industry if we start to get there.

Brian: [00:48:09] I agree. I think it'd be bad if people start to implement this and it's terrible, and then they feel like it doesn't work, and then they don't pursue it any further.

Phillip: [00:48:17] Yeah. Well, what we're talking about is B2B products. That's what's being sold here. What I think we're heading toward is...

Brian: [00:48:25] Consumer products, baby.

Phillip: [00:48:27] Consumers having extensions of themselves and businesses having to somehow compete with rising expectations. You wanna talk about if consumer expectations and inflated consumer expectations in the past were fast free 2 day shipping, the future is wild because it's, you know, an infinite number of their agents having those same expectations.

Brian: [00:48:55] Yep. Yep. Agree. I think, I can't wait for...

Phillip: [00:49:00] We didn't even talk about any of the main stage stuff. Ikea was on main stage. I think Kat Cole was on, saw her, ran to Rachel TenBrink for a bit.

Brian: [00:49:13] Oh, is she coming tonight?

Phillip: [00:49:14] I don't know. That's a good question.

Brian: [00:49:16] We should text her.

Phillip: [00:49:17] A lot of lot of really rad people. This show is full of everybody.

Brian: [00:49:21] It is. No. This is the show. This is the Big Show. Thank you, NRF. We actually are huge fans. This is actually, in many ways, my favorite show because I love New York, and I love being here with all these people.

Phillip: [00:49:33] Yeah. Our community is so great.

Brian: [00:49:34] Yeah.

Phillip: [00:49:34] Can't wait to see everybody. It's gonna be the greatest.

Brian: [00:49:37] Do you wanna talk hallway track before we...

Phillip: [00:49:40] Yeah. For sure. I am always surprised at people's takeaways. Here's a few...

Brian: [00:49:51] I love the press room little chats. The little press room chats. Every year, that's one of my favorite things that that happens when I come here.

Phillip: [00:49:59] Press room chats. Yeah. Let's see. Well, you give me a couple. I've done a ton of talking.

Brian: [00:50:05] There's a lot of focus on theft right now and loss.

Phillip: [00:50:10] Yeah.

Brian: [00:50:10] It's a huge problem and continues to be a problem. And the criminals are finding new ways of getting after things. You know what's interesting? I was supposed to talking to someone about this in the press room. I remembered something. A movie came out not too long ago, Emily the Criminal. Did you see it?

Phillip: [00:50:30] No.

Brian: [00:50:30] Oh, it's got Aubrey Plaza in it.

Phillip: [00:50:33] Oh.

Brian: [00:50:34] She's amazing in it. Really interesting the retail theft problem is getting to be such a big issue that we're making art about it now.

Phillip: [00:50:45] Okay.

Brian: [00:50:46] And so I actually think it is going to continue to be a problem. There are really creative ways that people are doing this theft as well. And they're getting smarter, and now they've got bots. And so it's fitting and crazy out there. That's one of the hallway track things I heard that's important. The other one is I think everyone is excited about retail. People are energized right now. The chatter I heard was everyone's excited. Everyone's ready to spend. Everybody's ready to get out there and play. New tools, new ways of engaging with consumers. This is exciting.

Phillip: [00:51:23] Okay. At the same time, I've heard a lot of people still waiting for the shoe to drop. I heard couple economist talks.

Brian: [00:51:30] Yep.

Phillip: [00:51:32] Really concerned about interest rate hikes this year. Waiting for the shoe to drop on SMP. Some people are concerned that there's gonna be some contraction next year. Really concerned about that. If interest rates start to go up so that'll be something to watch. So yeah. How much room is there left to grow for consumer? Tax cuts might help. If we can get government spending cuts, then maybe that helps to boost some things. So those are all talks that I've heard.

Brian: [00:52:14] So is DOGE gonna work? That's the question.

Phillip: [00:52:16] I guess that's the question. The other thing is is the consumer tapped because we are spending, spending, spending on credit cards and BNPL spending was historic highs over the holidays.

Brian: [00:52:30] Yep.

Phillip: [00:52:31] So we are tap tap tap. That's crazy too. So those are all things like yeah. We had a historic holiday. We also have consumer credit is extended like nothing else right now too.

Brian: [00:52:41] Okay.

Phillip: [00:52:42] Inflation's nuts as well. Gotta think about those things.

Brian: [00:52:45] All those things are important to think about. Well, it's interesting. I am curious to see what happens.

Phillip: [00:52:54] Yeah. Alright. So those are all our report from... I might have to come back and check-in one more time on the main stage content, and we've got a little bit of After Dark stuff. And we'll report back on our event from this evening at Lore. Can't wait to see it tonight.

Brian: [00:53:14] It's gonna be amazing.

Phillip: [00:53:15] That's it.

Brian: [00:53:16] Let's go.

Phillip: [00:53:16] NRF 2025, retail's biggest show, and that's it.

Brian: [00:53:22] Let's go.

Phillip: [00:53:22] Thanks for listening.

Recent episodes

LATEST PODCASTS
By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.