Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Deep Learning with PolyAI

Deep Learning with PolyAI artwork

051: Secret Agentforce: Declassified! And more from Dreamforce 2024

Oct 2, 202418 minAudio
ShareXLinkedInFacebookWhatsAppEmail

Our own Brian Thompson, VP of Product Marketing, joins CEO Nikola to discuss Salesforce's recent AgentForce announcement. This new initiative is presented as a big shift toward autonomous AI agents within large enterprises and while at least one the dyad sees it as a rebrand of Einstein Copilot, others may view it as a step towards making practical AI solutions mainstream. Talk points include the competitive dynamics between Salesforce and Microsoft, the challenges enterprises face in genuinely integrating and deriving value from AI, and the complexity of transforming organizational workflows and data management. Follow PolyAI on LinkedIn Watch this and other episodes of the Deep Learning pod on YouTube

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

Brianif we're being really charitable, it's crawl, walk, run. It's, how do we ease people into, this brave new world, but on the other side, it could also be like, and this is something that happens with these large organizations that they're Doing, they're painting a grandiose vision to make up for the fact that, Their actual product is behind the curve

NikolaHi everyone. And welcome to deep learning with poly AI today with me, I've got Brian Thompson, our VP of product marketing Brian, welcome to the podcast. Thanks for having me. It's really good to be here. First time. Yeah. Welcome on board.

BrianLong time listener. First time.

NikolaGood, good, good, good. We'll try to get some first quarter resolution in. But look, I think today what would be really interesting is to cover the announcement of agent force by Salesforce. I think like every year dreamforce happened and the Salesforce hype bandwagon has been in like going full steam. So tell us what is

BrianAgentForce? Oh, what is AgentForce? So AgentForce gosh, it's so many things, but really I think at its core, it's this big strategic shift from Salesforce. Leading up to this, it was customer 360 last year and so on. That's what they've been building toward. But now they're saying actually the next stage is, Autonomous dedicated AI agents. It's their chance or their, their play for deploying AI in large organizations in ways that are discreet, ROI focused, and really answering all of these questions that we've had over the last couple of years of how do we start to get value out of these AI experiments.

BrianSo that's one thing. It's this vision that they're painting. Another way of looking at it is it's a rebrand for Einstein Copilot. Because that's where we're starting at least. And they're looking to make new releases. Things are going GA over the next, months, two months, six months, and it's going, but there's a long road to hope.

NikolaYeah. Yeah. If you just read Reddit, some more cynical reviews say that Asian force is the rebranded Einstein co pilot. It's another one is a co pilot for like generative agents instead of the previous Einstein GPT announcement, which I think came last year. Not to be skeptical. I genuinely think that like Salesforce has so many things working in their favor in this ecosystem when it comes to the backend data interactions, et cetera. Benioff is going rampant talking about the fact that they have the best AI. And what really caught me by surprise in looking at, we're a bit like a mouse watching elephants stomp around, but it's very aggressive towards Microsoft, right?

NikolaLike he talks about Clippy. He talks about co pilot not being it. I almost feel like there's a part where he's just talking And the genetic behavior that's like something that we haven't stopped talking about for the past 12 months as he's bringing it into the world of enterprise. What do you think about all that?

BrianOh, okay. So there's multiple pieces there. First of all, why, why is Microsoft this present absence and why are we talking about Clippy and all of that? And I think that there's probably a couple of reasons there. Number one, so many of us were deeply scarred by Clippy. I love

NikolaClippy. I've said that before. Like I, okay, fine.

BrianSome of us love Clippy. But see the point being though, Clippy is like a universal point of reference and really,

Nikolaoh it's

Brianpolarizing, polarizing, which is fine. But polarizing, but well known. And one of the leitmotifs of this presentation and kind of everything that Salesforce is doing is that they're trying to take AI out of the ether and make it really tangible, practical. Results oriented. And tying it to discrete things that people know about. I think it helps ground. It grounds it in that way. But also the other thing that I think they're aware of, because they're, they're smart, right? Salesforce has long connected the legacy enterprise with.

BrianTrends that are happening in Silicon Valley. It's their stick. They can take it and upscale it up level it and all that. And that's been, really a good place for them. But at the same time, though, Microsoft, which has even deeper legacy connections. Built mainframes and so on. They're now connected to open AI. So all of a sudden where Salesforce was able to say, we are a safe enterprise, but born digital, still forward thinking company now, Microsoft can make very similar claims and they have the same kind of relationships with very senior decision makers.

BrianI don't know, I would expect that they're trying to get in front of some of that.

NikolaYeah, no, it makes sense. They're probably fighting for similar budgets. And one thing that many have touched upon is and I think this has been making rounds online and in other places where basically people talk about most AI projects failing, and how a lot of them don't actually provide the ROI that they hope it will. This on the one hand makes sense because it's a new technology and there's a lot to be, to be figured out. But on the other, like he kind of like takes that out and then talks about bringing something that's better than a Clippy, better than Copilot, very anti Microsoft, he talks about doing bake offs liberally, et cetera.

NikolaHe's using very startup lingo, which is very interesting to see, because I think it's simultaneously very like big company. Strong, confident language. And on the other, it's clearly the undertones of being a challenger to someone who's got the first mover advantage is very, very interesting. Similar to Elon talking about grog being better than opening. I don't know. It's not right. I think here as well, but when we look at. That whole thing. And then we look at the demos, right? They're basically showing, Saks Fifth Avenue using it.

NikolaA few people talk about evolving and changing as a company to meet people online instead of in stores about having hyper personalized shoppers. Then they go back to that and it will help augment humans, not replace them. So I feel like there's. It's not really gone all the way towards agentic behavior where it's still just a chatbot with the pre pre prepared like thing that a human can validate and send over. But to me that felt Very cool pilots.

BrianYeah. And same, I think that you can, once again, you can interpret this a couple of ways. If we're being really charitable, it's crawl, walk, run. It's, how do we ease people into, this brave new world, but on the other side, it could also be like, and this is something that happens with these large organizations that they're Doing, they're painting a grandiose vision to make up for the fact that, Their actual product is behind the curve or like arguably so and I think it remains to be seen But I don't know like maybe this is more your wheelhouse Like do you like what do you think?

BrianBased on what you've been able to see and based on what we hear for people that are leveraging

NikolaLook, I think that they're definitely not ahead of the curve, right? When it comes to voice, we've not seen them, right? Like we integrate with them. Salesforce is a great company. We're a happy customer, right? I live in Bridge Salesforce. When it comes to these models like they've had the budding AI practice that's never really made it into the big leagues. For the Ted Lasso fans, like they're like the club trying to make it into the Premier League, but have never really gotten there. Just recently they bought another voice company called Tenex a smaller company.

NikolaSame team that previously got acquired by McDonald's and later sold off the spare parts to IBM. So I think those moves are there much like this announcement. Just to counter Microsoft and preserve relevance in the space. I don't really think that they've gotten to like the demo is a demo, right? What they've shown in terms of the builder and configurator it's not how enterprise voice assistants are built, right? Like it cannot be one prompt with a few words and then off you go again.

NikolaI think for the same vision that Einstein had in terms of text and in terms of helping human Asians. It's great. And I think for that, they have the data, but I don't see why they're trying to compete with open AI with Microsoft and say that they've got better models. It is clear that they can have better, more targeted fine tuned models for each individual customer based on the data, right? This way, I felt a bit like they're really just taking a very broad sweep at. All of the competition needlessly.

Brianso I think broad is broad is what Salesforce does, right? They, and strategically it's a necessity. They're an everything company that it's full stack sales, marketing, CS, like you name it. And so it's one of those things where we're in tech. We, we think of in terms of trade offs, they're one of their biggest advantages is their ubiquity and the, the. Breadth of everything they offer. And by the same token, that breadth like forces them into a position where they have to be generalist about everything and they have to elbow people out otherwise.

BrianOtherwise they're potentially in trouble. But I wanted to dig into something that you said, cause you mentioned that the interface they show is not the right way to build voice AI assistance. But it might be sufficient for other things. Is this, does this I guess given those constraints or given that voice is, It's fundamentally different. Is there ever a world in which it gets properly incorporated into that kind of a, like a one stop one size fits all one stop shop thing, or does it need to be on its own? No, I don't think it has to necessarily

Nikolalive on its own. I think Salesforce is one of the players I think with the strongest, there is a great price to be had in the world of doing a lot of work in an automated way. With the right integrations. I think Salesforce is the kind of a key price asset compared to almost anything else out there to build, a huge number of autonomous functioning voice assistants that do a lot of work for a lot of companies, both in support and in sales and in everything else. They've definitely got a good launchpad to do it. I think they don't have the technology. They don't have the platform.

NikolaThey. And it's not so much about the model. I think about that, but much many of his, right? I think it's really more like full on suite of like solutions and everything from analytics to the actual like dialogue manager inside the voice assistant. Like how do you build and provide the control to people, right? In many ways, I think the fact that they're trying to go so broad doesn't allow him to go after the fact that, maybe we can't have, A fully autonomous voice assistant and they've demoed one and it's not bad, right?

NikolaI think the real issue is if you build it into a, with technology that looks like what they demoed it's not actually going to work that way. And I think they we only need to wait until February and the GA for voice to see how well they get on. We've been doing a lot of work on other models and including end to end ones, and you can do a lot, but equally a lot is left. When people show those demos. So I think everything from the control to the third party integrations to actual safety and dealing with hallucinations and double hallucinations, which is what we call like when speech goes wrong, the agent goes off the rails.

NikolaWhat do you do? I think it's also just a really good thing that they're doing this. I think that we've seen like research of interest and followers that you would not believe on LinkedIn and everywhere else. So thank you, Mark Benioff, much like Sam Altman for all that you do for, bringing to the attention of larger enterprises, the good work that we do as well. But yeah, I don't know. I think that there'll be something we'll see how much I think that they've done a lot historically any dream force to announce rebrand and show. And I think. People are like holding out their breath and waiting to see what's real versus what's just the general showmanship that they do each year.

BrianYeah. And I guess when I think about it it setting aside how achievable any of this is for them or in the short term, whatever. I'm just interested in, In that future state vision that they're painting, right? It's one that as I said, it's okay, in some ways, they're trying to take AI out of it comes first out of the classroom out of the academy and then out of the I don't know that like the, whether it's weird garages of startups or whatever, but they're really trying to take it into the corporate boardroom.

NikolaAnd it's already there. So I think just like in terms of what both open AI and then Microsoft did in the way that. Just in how impressive LLMs are, right? Big enterprises are already doing it. They are late to the game and just trying to come to their own phases. Yeah, they're

Briandoing

Nikolait. Yeah,

Brianyeah. No, this is the point I'm trying to make. They've been doing it, but when you look at and I'm, up to my eyeballs and Consulting research and things like that. And so you look at what people say. It's yes, I think something like 80 percent of large organizations are either currently experimenting with gen AI, or they have plans to do it in the next year. But when it comes to like notions of ROI, they talk about innovation value and other like very, very squishy things. And I think that we're community adjusted EBITDA. Yeah, and so I think that, folks are feeling increasing pressure to let's be practical about this and let's like, think about a deployment that's not just, intellectually interesting, or doesn't just satisfy the fact that our board wants to know what we're doing with AI, but really is something that's going to be a step change in how we, function as an organization.

BrianAnd so that was the thing that intrigued me about, Salesforce, like putting their marketing muscle and their messaging muscle behind this multi agent vision of the future. Because I think for a lot of these large orgs, this might be the first time that they're really being given a permission structure to think in these ways. It's,

NikolaNow,

Briancause it's Salesforce trusted partners saying this is the future. And I think that has some real implications. And so I want to know your reaction to that. No, I look, I think it makes sense.

NikolaI think the. Really in many ways what I found incredible as a B2B founder, the more like we've gone into it is there are, there's basically a dark web of companies that you don't get to sell to. As a small B2B company, right? Like you can sell it to a large bank because they have innovation departments. You can sell to all sorts of companies. You can develop a vertical speciality and then find the right channel to get to someone. But like both Microsoft and Salesforce are companies that, especially Microsoft, have government clients, healthcare clients, other clients that you just won't get to work with if you're a startup, right?

NikolaSo I think you're right. Like them like taking Silicon Valley lingo. Into the market out there, I think it's maybe necessary to use the kind of like confidence and bravado and going against Microsoft, but otherwise, they're there to help companies experiment. And because of who they are and what they do, I think there will be a great many company for whom they are the best stepping stone into that technology, right? Thing that gets a bit hairy here is that, AI is not really SAS, right?

NikolaTo derive meaningful enterprise value from AI, you really need to wrap yourself around like recreating workflows knowledge bases, integrations for that agent to harness, right? So the question is will their partners. Be able to do this just yet. And that, there are two questions, like how well have they enabled them and how good is the platform and can it be done, right? I think it's going to be a pretty tall order for there are very, very many partners to fully use this technology to create those kinds of outcomes yet, just because of where.

NikolaThe technology is not the models, but the application stack, right? Like computer networking, maybe there aren't seven layers, but there's quite a few. And I think that models, LLMs are like the bottom one, right? But who and what company uses it, uses that configurator, how they use it, how do they observe its behavior? When are they confident about releasing it? And again, I come back to if they are to take those baby steps, the first step very much looks like copilot, right? So those are marketing wars between Microsoft and Salesforce. And that's fine.

NikolaWe'll see how serious they are about any of this, but the skeptic in me almost goes if you're Salesforce, you want to talk about this so that if there is a crisis of ROI later when it hits, Microsoft hits you, you just weren't hit in the meantime, because you have. Been seen as a laggard in terms of AI. We see many of tweeting about his models being ready better than anyone else's. And I think, that doesn't really make them stand out next to Elon and the whole cabal.

BrianBecause there's also this like heavy implication that they just have solved the data aggregation problem too with data cloud, that kind of thing. It's so we have everything. And and I just, color me skeptical. I have yet to, look under the hood of even a moderately sized business where everything is nice and in order. But like on that note I think, when I think of, It's like everything that you were talking about. I think it's the data aggregation piece. I think ways of working within the enterprise, like just like the cultural change necessary to roll these things out.

BrianI see as potential obstacles, but am I missing something? What else do you see is like making this kind of hard over the next, let's call it three to five year transition before we're all replaced by robots.

NikolaYeah many ways to feel that. Terms of making it real, right? They talked about Saks Fifth Avenue, then they demoed Sophie. If we call Saks Fifth Avenue now, Sophie's not there. So the first question is, where's Sophie? And I think that kind of goes back into what you talked about. The whole thing is a very messy affair, right? What you need to do inside that company to mobilize it towards having its data processes In a shape where you trust, let's say perfect AI or an infinite pool of really good agents to do the work well, it's a hard thing.

NikolaIt's a transformation, right? So will Salesforce be a part of this world? I think they have a birthright to be one of the winners. And I think there will be I think the pitch is feels a bit recycled, but it as we've seen, I think that we have quite a few customers that are excited about it they'll definitely have a fair shot at trying it out for voice. I don't think they've cracked it. Or are even close yet, but I'm excited to meeting Sophie.

BrianThat makes two of us.

NikolaYeah, cool. Brian, I think we're out of time. Thank you for being a guest on the podcast for the first among many times, I hope. Thank you. Thank you all for listening and we'll see you in the next one.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. What if your contact center became a product team?

    Nov 17, 202635 minAudio

    Three years ago, Simplyhealth's PolyAI deployment was only resolving simple FAQs over the phone. Today, they run voice, in-app, email, chat, and WhatsApp conversations on the platform. Nikola Mrkšić talks with Paul...

    Credit the people on this episode

  2. What happens if your AI agents get 1% better every day?

    Sep 10, 202639 minAudio

    An agent that gets better every day compounds into a very different agent within months. That's the idea behind Wren, PolyAI's new release: an agent whose job is to make your other agents better. Some of what it finds...

    Credit the people on this episode

  3. Who's coordinating your army of AI agents?

    Sep 3, 202644 minAudio

    Companies are pouring money into AI, but not everyone is getting the returns. Ian Jacobs, VP and Lead Analyst at Opus Research, thinks he knows why. In this episode, your host Nikola Mrkšić sits down with Ian to work...

    Credit the people on this episode

  4. Why should CX leaders care about MCP?

    Aug 20, 202636 minAudio

    When PolyAI built its Builder MCP, one client helped shape it: Vixxo Facility Solutions, one of the largest facilities providers in the US. In this episode, guest host Damian Sasso talks with Vixxo's CTO Derek Neighbors...

    Credit the people on this episode

  5. Can AI really hear a call the way a person does?

    Aug 6, 202622 minAudio

    PolyAI just launched Dialog-RSN-1, its first audio-native model, and your host Nikola Mrkšić sat down with its builder, Matt Henderson, to unpack why it’s a game-changer for building voice agents. Most voice AI either...

    Credit the people on this episode