What's Next for iRobot?


In this podcast, Motley Fool analyst Jason Moser and host Deidre Woollard discuss:

  • Why Amazon ended the iRobot deal.
  • Where iRobot could go next.
  • If Sofi is building the next big bank.

Eric Siegel, author of The AI Playbook, explains some of the challenges facing companies looking to adopt artificial intelligence.

To catch full episodes of all The Motley Fool’s free podcasts, check out our podcast center. To get started investing, check out our quick-start guide to investing in stocks. A full transcript follows the video.

Should you invest $1,000 in iRobot right now?

Before you buy stock in iRobot, consider this:

The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and iRobot wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.

Stock Advisor provides investors with an easy-to-follow blueprint for success, including guidance on building a portfolio, regular updates from analysts, and two new stock picks each month. The Stock Advisor service has more than tripled the return of S&P 500 since 2002*.

See the 10 stocks

 

*Stock Advisor returns as of January 29, 2024

 

This video was recorded on January 29, 2024.

Deidre Woollard: Amazon says no robot. Motley Fool Money starts now. Welcome to Motley Fool Money. I’m Deidre Woollard here with Motley Fool analyst Jason Moser. Jason, how was your weekend?

Jason Moser: Hey, Deidre. Just great, how about yours?

Deidre Woollard: Watched a lot of football.

Jason Moser: Yeah, there was a little bit on TV, wasn’t there?

Deidre Woollard: Just a little bit. Well a lot of times I meet with you on Mondays, we’re talking about Big M and A deal today we are going the other way we’ve got a big break up may be a small break up depending on your view of it. But Amazon has called off its acquisition of iRobot, so Amazon doesn’t seem to want a Roomba anymore. Why do you think that is?

Jason Moser: No, it’s un-deal. Un-deal.

Deidre Woollard: Yes un-deal I like that.

Jason Moser: Honestly, I’m not surprised it felt like this was something that was almost inevitable given the EU’s concerns regarding the deal. I don’t know, to me this was a deal that always it just never seemed to make sense. I understand Amazon is just trying to get more and more of data and I think the data argument works to an extent. But, as we’ve talked about before, data is data at some point it’s becoming just unlimited. It’s frankly it’s very easy to get so then it becomes a matter of like what you do with the data. It’s not really a matter of having the data as much as what you can do with it and I think it’s fair to say that Amazon does a pretty good job of doing stuff with data. But what they would get from a company like this never really was clear from the start. To me I feel like, listen, I respect they realize there’s no path to this still happening. It’s going to be more trouble than it’s worth. They pay what, a $94 million breakup fee, that’s a little bit more than a 10th of a percent of the cash on Amazon’s balance sheet. It impacts really nobody on the Amazon’s side and so from that perspective, I’m not worked up at all about this deal on happening and frankly. I feel like I’m a little bit more happy that it’s not happening because it just never really made sense to begin with.

Deidre Woollard: Maybe it was a tech play a little bit, there was that fear about the data. If it’s important enough that the European regulators were trying to block it, I don’t know what that maybe says about the relationship that companies are having with regulators. We’re seeing this all over the place sometimes deals are taking extra long to go through. We know the FDC is looking at companies like Amazon, like Microsoft. It’s got its eyes on big tech.

Jason Moser: Yeah.

Deidre Woollard: Do you think that’s part of this for Amazon? It just thought, this is a tiny thing it’s not even worth being a blip on the radar at this point with it.

Jason Moser: Yeah, no question. There’s no doubt that giving these things another look is very much involved these days, particularly when it involves big tech. When you look at the numbers its just compare that to the scope of Amazon’s business. This is a deal it does not impact Amazon at least. You could look 10, 20 years out and say maybe what could they do with this deal on that data that might come from a deal like that. That is obviously you’re speculating very far out and so from that perspective again, to me it just feels like they realize the juice isn’t worth a squeeze. This isn’t that big of a deal. It’s just not worth getting into and I understand the EU regulators, there’s a little bit more of a concern over antitrust regulations overseas versus here. But it does feel like over here, domestically speaking, that is becoming more and more of an issue. We’re seeing more and more big tech, these types of deals are being put under the microscope and understandably so, it doesn’t seem like these are deals that matter as much to the acquirers. For Amazon this just isn’t really a big deal, for iRobot obviously it was a very big deal and this was a bit of an exit strategy and perhaps this was something that they were thinking from the moment they went public. Maybe this was part of the exit strategy plan there. I don’t know it puts iRobot in a very difficult spot right now whereas with Amazon, just no worries at all. They’re going to be able to keep on just going around business as usual so we’ll have to see what comes up in regard to iRobot.

Deidre Woollard: Yeah, that’s the part that’s worrisome. This deal has been in the work since August 2022 and when that announcement gets made, it’s not that the company goes full stop, but there’s always that pause. You don’t quite have to push as hard, you know that this thing is coming and then all of a sudden it isn’t. They announced they’re laying off about a third of their staff. Amazon had already dropped the price on the acquisition, so it already signaled the value here is slipping. The Roomba brand, I was talking to you before the show and you mentioned that you bought a Roomba competitor recently.

Jason Moser: Good idea.

Deidre Woollard: Does that brand still have value?

Jason Moser: Well, there is value there. But I think that what we’ve seen is very quickly, what was it bit there’s that old Warren Buffett quote regarding like the innovators, the imitators and the idiots. You have the innovators that come in there and get things going and then you have companies that come in there and try to imitate and they take part in that growth and then you have idiots that really come in there and don’t know what they’re doing. We’re at this stage, it feels like in this particular market, I’m not talking about just robot vacuum cleaners, but all of these things that iRobot was focused on, we’ve seen a lot of imitators come in and these imitators have done a very good job. Just anecdotally, I went through and did a lot of research over the holiday season on one of these particular products. The robot vacuum cleaner that goes back to its station and unloads itself. I was just surprised, not really surprised, but I was just a little bit, I took note of the fact that the products that iRobot has, they just weren’t getting the greatest reviews. Having had an iRobot, a Roomba from early on when these products really first launched, it was always one of those things where you thought there’s potential, but it’s not something that’s necessary and it’s not something that’s sufficient. In other words, it doesn’t take care of my problem. It’s neat to have, but I don’t need it and I think we’re probably starting to move a little bit past that point where they start to become sufficient. They’re not necessary by any means, but there’s clearly a lot of competition out there and it does appear that iRobot and their products are playing second fiddle, if not further down the line and that’s a big problem. If you’re talking about making, a one to $2 billion acquisition of a company. You got to keep that stuff in mind, like is that brand on the up and up or is it something that is, witnessing some competitive pressures there and it certainly feels like iRobot was witnessing a lot of competitive pressures.

Deidre Woollard: Yeah, it has the name, but the name alone is not a mode at all.

Jason Moser: Yeah, it’s not a verb yet.

Deidre Woollard: Well, it’s sort of but it didn’t quite get to that status. The other thing I think here that I think a lot about is the promise of the Internet of Things and there’s been that trend, the promise of the Internet. Everything’s going to talk to each other. Your computer, your refrigerator, and everybody needs to be connected. This is the way of the future. That promise hasn’t fully come through, in my opinion. I think that’s part of this too, is that we’re becoming less entranced with these types of items.

Jason Moser: I fully agree, I feel like there’s a lot of potential when you consider the Internet of Things, particularly as it relates to consumers. But the promises have not really been fulfilled and I think a lot of that honestly just has to come down to the reliability and having used some of these things I’ve tried to incorporate certain types of these things into our lives at home. Amazon, Alexa, connecting it to the light switch, being able to just talk and make something happen. It’s clever technology, it’s really cool and when it works, it’s really impressive. The problem is, it’s not reliable. You know what’s undefeated? Deidre The light switch. The light [laughs] switch is undefeated. When [laughs] I want the lights to come on, I flip the switch and they’re on. But it took probably, I don’t know, two or three times where I said, hey, Alexa, turn on the lights and she said, can’t connect, don’t know. I was like, OK, I’m out. It’s not reliable. We need to get to that point where this stuff is just reliable and maybe there’s a redundancy factor that comes into play here, with Cloud providers, they kind of have that back up to provide that redundancy. If we can get to that point with the Internet of Things and smart home where there’s that redundancy and that reliability, then I think we’re really onto something. We’re just not there yet and it’s not to say we won’t get there, we’re just not there yet.

Deidre Woollard: Yeah, I think that’s true. I want to move on and talk a little bit about earnings. The big earnings are coming later this week. But the one that we had today was SoFi. SoFi is interesting, this company started off as a SPAC. You have the former CFO of Twitter at the helm, Anthony Noto. But what’s interesting is just how fast it’s grown. The customer base grew 44% year over year. Now that’s only 7.5 million customers total. It’s not coming for Bank of America‘s throwing anytime soon, but this is interesting. Should we be taking SoFi a little more seriously?

Jason Moser: I think so, personally. There are two things I really like about SoFi today. One is, you mentioned Anthony Noto. I think he’s the right leader for this company at this stage of his life. He’s a great advocate for the brand and I think he’s done a wonderful job bringing the business where it is today. But I also think just SoFi was born from student loans. Ultimately it started as a student loan solution. That resonates with a demographic that I think is key to the company’s success, and so when we see them getting seven-plus million members, clearly much of that member base skews younger. As we know banking is very sticky. I think this makes a lot of sense for companies like SoFi, you get companies out there like public.com. Any companies that are out there really trying to offer a new way of looking at finance and helping the younger generations approach these issues. There’s a lot of potential there because banking can be so sticky. once you get your banking relationship involved, it’s just a lot of work really, to unwind it. The longer that relationship pursues, the less inclined you are to really want to unwind it. I told you before like, if you look at me, for example, now I’m an old codger, but $10 of marketing spent toward me on SoFi’s part, that’s $10 wasted. I’m not going to switch because our banking relationships have been so established for so long, I don’t want to unwind that. But for younger generations that don’t have to worry about unwinding it, they’re just getting started.

They’re looking at these solutions that SoFi is presenting, and they’re thinking this looks like something I could use. This looks like something that’s going to be helpful for me. You know we saw in the press release they’re offering opportunities in regard to investing in alternative funds. Whether you’re a big believer in the alternative funds side thing or not, that’s one thing. But then you look at they took that next step in actually offering the educational side of it too as well. For us here at the fool obviously that’s a big part of what we do every day. We are big believers in the education side of things, and so I think there are a lot of reasons to believe that SoFi is on the right path here. It’s just it’s a difficult business, no question, but really it’s very understandable. It’s a big land grab. They’re trying to get those members in and lock them in by giving them more services and making those services a bit more a part of their day to day lives.

Deidre Woollard: Well, even just the word members that they used to describe their customers I think is interesting. Because part of what they’re doing is they’re giving it that, I hate to call it like a millennial feel, but there is that like you’re part of this club of SoFi and you have this trust with us. We’re going to just keep adding products that you might be interested in and we’re going to do all of these different things for you. I think when you have that stickiness you can do that pretty quickly. I think it’s easier to do with a smaller user base. I think it’s going to get a little bit trickier. They’re nimble in a way that a big bank can’t be. But at some point they have to start playing more with the big boys.

Jason Moser: Well, you’re right. At some point they’ll have to approach that. But I absolutely agree with you on the members side of things. Reminds me very much of what American Express has done through the years. There’s just something there, they create that sense of you’re part of something a little bit bigger. You’re part of something that is, I don’t want to say exclusive, but there’s something special to it. When you can make people feel good about something in which they are participating well, that typically is going to result in a longer term relationship.

Deidre Woollard: I think that’s really the case, is that they’re building this thing for the future versus just part of it’s the land grab and part of it is like we’re offering high interest savings and things like that, but it’s really building this thing for the long term that matters. Jason, I want to ask you one more question. We talked a little bit briefly about the alternative investing thing that they announced today. Do you feel like the younger generation is interested in this? The sexy part of it of course is you can be part of Kathy Wood is Ark Investment. But the other part of it that’s in with that is 6,000 mutual funds. Well, a [laughs] little less attractive. How do you think that’s going to land with maybe a younger member base?

Jason Moser: Well, that’s a good point. I think that’s really where the education part comes in. I think most of us would agree that mutual funds are yesterday’s investments, not terribly focused on mutual funds as they’re not the most efficient. Most of them tend to underperform the market. We typically don’t point people toward mutual funds for the most part. But generally speaking, the alternative investments, I like the idea that they’re bringing more options out there for people because ultimately their goal is to bring more people under their umbrella. That’s where I think really the education part comes in. I’m happy to see that they built that part of the site out to help educate investors, to at least give them some better understanding as to what they may or may not be getting into. Again, younger generation, they don’t know as much just because they haven’t been around as long. That’ll change quickly, of course. It’s nice to see SoFi really focusing on that educational side of things.

Deidre Woollard: Absolutely. Well, thanks for breaking it down with me today, Jason.

Jason Moser: Thank you.

Deidre Woollard: We talk about a lot of stocks on the show, but it’s just a peek at the motley fools investing universe. This year we’re rolling out a new offering. It’s called Epic Bundle. The service includes seven stock recommendations every month. Model portfolios and stock rankings, all based on your investor type. We are offering Epic Bundle to motley fool money listeners at a reduced rate. As a thanks for listening to the show. For more information, head to Fool.com/epic 198. We’ll also include a link in the show notes for you. AI isn’t as easy as flipping a switch. I talked to Eric Siegel, author of the AI playbook, about the lessons we can learn from big companies pursuing AI objectives. Well, I’m excited to talk about the book because it hit on some anxieties I think I’m having when I see so many companies needing to deploy AI. There’s so much hype right now and I worry that a lot of that hype, it turns very quickly to disappointment if things just fail to launch. Tell us a little bit about some of the hurdles that companies encounter.

Eric Siegel: Well, there tends to be a big disconnect between the business and tech side. Data scientists make a predictive model that’s meant to target fraud detection and marketing and credit, financial credit risk management, etc. Then it doesn’t actually get deployed because the business stakeholders just aren’t quite ready the get wet feet. They weren’t involved enough in the project. Taking a step back there is a lot of hype right now. Generative AI products like ChatGPT that generate text in an almost seemingly human like manner are extremely impressive. Something I never thought I’d see in my lifetime. I’ve been in the field for more than 30 years, but the value of them is probably overblown. There’s a lot of value for having it write first drafts of English, of natural language, and of code, but I’d like to sort, pivot the listeners. Let’s not forget the established enterprise use cases of the underlying technology which is machine learning. Learning from data to predict, which is used to improve all the large scale operations. The Holy Grail for improving operational decisions is prediction, whether you’re going to click by lie or die, the outcome or behavior that will directly inform whether to contact you for marketing, which add to display, whether to issue a credit card, whether to audit you for fraud. This is where the established track record is. This is older than generative, let’s call it predictive. But it’s not old school, it’s where the vast majority of established opportunities still exist. Its potential has only barely begun to be tapped. There’s a lot of industry leaders, but there’s far more that are behind and aren’t quite making that connection between BizInTech, so there’s a lot of initiatives that actually fail to deploy, and what’s needed is a couple of things, an established industry process or practice, and that’s what I talk about in the book. But perhaps more importantly, and first and foremost, is some ramping up. That’s what I’d like to espouse. There’s nothing intimidating about the basic idea of learning from data to predict and then using those predictions. What all business stakeholders need to learn about the use case, about any particular project that’s meant to deliver value is three things, what’s predicted, how well, and what’s done about it. You might predict who’s going to click by lie or die, commit an act of fraud, turn out to be a bad credit risk, etc, whatever behavior and outcome there would be to predict. Then what’s done about it is, let’s act on that in order to drive individual decisions. How well is, what are the metrics? How well does it predict? What kind of ROI would the project potentially deliver?

Deidre Woollard: Well, I like that you made that distinction between the generative AI that we’re currently experiencing and playing with and the machine learning that has been existing in our lives for years. Some of the challenges with machine learning are there’s the tech side and the deployment side, but there’s a mindset issue there as well. How do you help businesses wrap their brain around that?

Eric Siegel: Well, the mindset is basically something we’ve been talking about since the big data movement and well before that, which is, you need to be data driven, you need to be empirical. There’s a place where computers trump the gut. I trust my GPS to tell me exactly where to drive. Every time I try to outguess it and think I know the residential streets better, I often get the sense that I was wrong and it knew where the traffic was and it optimizes. There are certain things machines are just better at, including learning from a large number of historical examples to predict, and prediction is the Holy Grail. This is what it means to apply science to the improvement of business, to be a specific, the improvement of our large-scale business operations, all the main things we do as organizations. The trust thing is moving along. But part of what’s going to make a big difference with the trust, and also ultimately the antidote to hype is to focus on concrete use cases and get the stakeholders, for example, people who are in charge of the large-scale operations consisting of many decisions, that stand to potentially be improved by the predictions delivered by machine learning, because that’s what it does, it learns from data to predict, and by getting everybody on the same page, and ramping up in detail on what’s predicted, how well, what’s done about it. This is driver’s ed, not auto-mechanics school. You don’t have to pop the car and see where those park plugs are to drive a car, but you need a lot of expertise about how cars operate and the rules of the road. Likewise, you need this expertise to run machine learning projects, and in that way, take the best practices in applying science to business and get the thing successfully through to deployment.

Deidre Woollard: Well, all of these driving and auto-metaphors are leading me perfectly to your example of UPS in the book, which makes perfect sense. You’ve got 16 million deliveries a day using machine learning to optimize. Sure, that makes a lot of sense. But it was not that simple at all. Tell us about the story of Jack Levis and how he got machine learning into UPS and some of the challenges he faced.

Eric Siegel: An incredible story. Jack Levis was the lead, he’s more recently retired, but he completely revolutionized the way they actually optimize that delivery of the 16 million packages, and the way they do it is they optimize delivery by predicting delivery. In particular, what they optimize is exactly the allocation of packages to delivery trucks at all the different shipping centers around the US. The end result of doing this, by the way, in conjunction with also prescribing driving routes, which I just mentioned a moment ago, which also helped, together this produces an annual savings for UPS of 185 million miles of driving a year, $350 million, eight million gallons of fuel, 185,000 metric tons of emissions. But the only way to make that improvement was not just the number crunching, it was the deployment, it was the actual integration of the predictions of what’s learned from data to actually change existing operations. Today, we’re so fascinated with the core technology. As a data scientist, I’ve been in the field more than 30 years. I was, and in some ways still am the same as your typical data scientist, we love that idea, it’s the best of science, learn from data to predict. It’s amazing, you’re making discoveries that hold in general, in that sense you’ve actually learned a truth from historical data, and in that way, data is experience.

But what matters is actually getting it to deployment. It’s like right now we’re more excited about the rocket science than the launch of the rocket. Jack Levis very much faced that, and had to go up the organizational chain and then down. I cover that in a couple stories in the book, pretty much the opening of the book and then the closing, I circle back to the UPS story. Because first he had to convince an executive, and he literally took him for a ride and showed them what happens and how it’s counter-intuitive and potentially more valuable. Then he had to convince the people actually implementing integrating the new process change, namely the staff workers who are loading the packages following the prescribed instructions on the loading dock, loading them into the trucks. They were resistant to change. Change is a hard thing, but if you’re going to get an improvement, you need to implement change. That’s why these projects need to be seen as business operation and improvement projects that involve change management rather than just being tech projects. The tech is important, the analytics are extremely important, that’s what we’re leveraging, but first and foremost, it’s a business project.

Deidre Woollard: In the book, you also talk about the FICO score, you call it the most famous deployed model. It’s amazing. It receives 20 billion records, which is just like terabytes of raw data every month, a petabyte every five years, so it’s got all of this data coming in. How did they wrangle all of that?

Eric Siegel: Well, actually, those stats that you just listed pertain to FICO’s fraud detection model, which stands in contrast with the FICO’s score. There are two different products that FICO has, and that’s the funny thing, is that, FICO is so famous for the credit score, but probably a bigger part of their business, and something that affects you and I much more frequently is the fraud detection model. Doing what I just mentioned, in real time, banks are using it to decide whether to authorize your payment card transaction. In fact, FICO had very much as the corner on that market. For two-thirds of the world’s payment cards, 90% in the US and the UK, every single transaction is determined on the fly, in real time based on the prediction of that very model that’s delivered by FICO, whether or not to authorize and allow that charge. The reason they’ve got so much data to work with, is that, every single bank that’s using it has to have also agreed to participate in this consortium of banks that all provide the data from which to learn which transactions did or didn’t turn out to be fraudulent. Greatly informed, of course, by, when customers complain because they’re like, “Hey, I didn’t buy that.” You see that. That’s the source of learning. That’s what’s called the training data, the positive and negative experience from which to learn. Each one is basically just a row of data. They accumulate this incredible amount of data, they literally provide an updated model exactly once a year to all the banks that are using it. That means even small and medium banks get to make use of the best in class fraud detection model, which is great. I mean, it’s great for society. This is the very integrity of transactions. We need to make sure criminals aren’t conducting unauthorized transactions too often. Yet we can’t have too many safeguards because that’ll really decrease the usability of e-commerce and in person bricks-and-mortar commerce as well. It’s an amazing showcase of deployed machine learning.

Deidre Woollard: Also one of the things that you talk about in talking about that is how much prep has to go into making this model work. I think about it like painting a room. I mean, it’s like no one wants to do the hard part of taping down everything, but really that’s what you have to do with a large language model. Is, there’s so much training. I feel like you’ve studied this thing, like you said, for 30 years. All of us are just learning right now about large language models. What do we need to know?

Eric Siegel: Well, when we go to large language models, Generative AI, by the way, refers to all these large language models that power chat bots like ChatGPT and many other competitors as well as generating image, video, sound, music, so like DALL-E 2. Either way, you’re leveraging the same core technology that I’ve been talking about for these enterprise projects. It’s learning from data to predict. It’s just that you’re applying that technology in a very different way. It’s literally predicting what should the next word be as I’m writing this sentence or this paragraph. Well, technically, it’s the next token, but it’s basically on that level of detail per word. In the case of rendering an image, it’s how should I change this one pixel as I’m continually iteratively going through phases to iterate this incredible image. The capacity for it to generate these images and generate text and computer code is just unbelievable.

I mean, before I was a professor at Columbia, I got my PhD there, during which for six years, I was in the Natural Language Processing Research Group. I’ve seen it fail a lot, and seeing what it can do now is definitely something I never thought I’d see during my lifetime. I mean, they really ramped up the underlying core technology in order to actually leverage this unfathomable amount of data points from which to learn. But just the same as you learn, hey, transactions with these characteristics done by this type of cardholder, this situation, and this time of day, and all the attributes that might inform the chances that it’s fraudulent, just the same as that. It’s saying, hey, look, I’ve written these three-and-a-half sentences so far, given that I should predict what’s the next most likely word. It’s the same idea, same core technology, just being applied in a different way. The question is, how far can we get? What it’s done is extremely impressive and definitely valuable in terms of first drafts that are then used by a human. Measuring the absolute enterprise business value of that endeavor, well, that’s something people haven’t been talking about enough.

There’s been some initiatives to that end. Of course, the value is going to totally depend on the particular writing, let’s say, you have to write 200 letters a day to customer service or something. It depends on the particular language model you’re using. It’s experimental. You can only try it out and see how well it works. But the overarching question is, how far is it going to get in terms of helping and even replicating what humans can do? I say, well look, how much can we actually reverse engineer, the incredible abilities of the human mind, just based on that one part language? That one behavior of humans recorded as all the words that we’ve written. Well, you can get pretty far, but you’re not going to get all the way as far as the general capabilities of humans. I think that’s a red herring when we talk about what they call artificial general intelligence; computers that can do anything people can do. But as far as improving writing tasks, coding tasks, drawing tasks, there’s a whole new level of capabilities, and there’s a lot of creativity coming down the line in terms of the way it could be used. But as far as everyone’s waiting for the big value to strike oil, to discover that killer app, people who code say, hey, that is a code killer app, for me, it’s really helping me. It makes me code 50% faster, maybe twice as fast but depending on the task in maybe the best case. In a sense, that’s a killer app. But I think in terms of the public narrative on the world stage, there’s no killer app. It’s like waiting for it do. The killer app that’s implied in the story is an artificial person, and I do not think we’re headed there.

Deidre Woollard: As always, people on the program may have interest in the stocks they talk about, and the Motley Fool may have core recommendations for or against, sod don’t buy or sell stocks based solely on what you hear. I’m Deidre Woollard, thank you for listening, we’ll see you tomorrow.



Source link

About The Author

Scroll to Top