Episode 39: AI and Sales with Matt Walsh

Host Rob May has a conversation with Matt Walsh, President, and Co-Founder of Noted Analytics. Tune in for a conversation on AI and sales, what Noted Analytics does, and much more. 

Subscribe to AI at Work on iTunesGoogle Play, Stitcher, SoundCloud or Spotify

Rob Circle Headshot

Rob May, CEO and Co-Founder,
Talla 

Matt Walsh headshot

Matt Walsh, Founder,
Noted Analytics 


Episode Transcription   

 

Rob May: Hello, everybody, and welcome to the latest episode of AI at Work. I'm Rob May, the co-founder and CEO at Talla. And today my guest is Matt Walsh, who's the president and co-founder of Noted analytics. Matt, welcome to the show. And why don't you tell us a little bit about what Noted Analytics does, and then a little bit about your background before this company?

Matt Walsh: Thanks, Rob. Happy to be here. Like the local podcast here right in Boston. So a little bit about Noted Analytics. The idea from it came from my 10 years in sales roles. I used to work SAP prior, and just found myself frustrated with how much time I would spend updating CRM from my meeting notes. I also spent some time in sales operations roles, seeing the power of if you can get good information from the field, you can package that in a way and deliver insights back to the team that makes them more effective in their sales cycles. I thought well, how can we bridge that gap of capturing information that the team collects and package it in a way that actually helps them close more business?

That's the idea of Noted Analytics. It's taking all that data and revenue that's left on the table, and make it actionable to drive more revenue.

RM: Tell us a little bit about some of the AI and machine learning things that you guys are working on that are related to this.

MW: Where we're starting first is in categorizing the information that reps capture. I was a big Evernote user, right? But what I found was Evernote wasn't doing anything for me. I had this unstructured notes. Then I have to do all my time placing this into certain fields, moving it around. So the first step that we've taken on our AI path is auto-categorizing information as you take it. If I say send the PDF to Rob, that's the next step-- which will then update the next step field, create an action, a task for me to complete.

If I'm capturing use cases, right, so then as you start to go back and look at it, then you can say, all right, well what use cases are actually driving success our organization? Which ones are leading us down a path that there'll be a lot of tire kicking and will go nowhere? And the goal then is to bring that information forward in a sales cycle. So maybe it's meeting two and you hear about a use case. Maybe it sounds great. You're an account executive. You have happy years. You're excited about this, but actually, you know what? Every time this comes up, this isn't good. We should try and steer it to another use case or guide the conversation in a different way.

RM: Are you basically sort of like compiling models and deploying them, and they're doing things? Or is there is there a feedback loop here that it starts to customize it for a specific customer? Tell me a little bit about the process and what's required of the end user, so that the tool can sort of get trained and learn how to categorize those things.

MW: That's been part of our evolution as well. Initially, the rep would take notes. And then they would apply their own categorization, just as you would use a tag on Twitter or LinkedIn. Some users found that to be something they understood. Other users were not so used to doing that. We added a template feature. You can go in, let's say it's a discovery call. You want to learn who you're meeting with, what are the pain points, what are the next steps. Starting to auto categorize that. From where we stood with our beta customers and templates, now we have enough data to do the auto recommendation.

When the tag comes up, there is a checkmark or an x. You have that feedback loop to make it smarter going forward. Obviously, the goal is to continue to refine it, so it's working at a company level. Then as we get more and more data, to make it specific to that individual user.

RM: Do you talk about AI during the sales process to your customers? Is this something that they latch on to? Are they coming in looking for a smart solution? Is it a top tier messaging? Is it a secondary messaging for how you do this stuff? Tell me a little bit about that process.

MW: For the customer, it's not top tier. For the customer, it's more around the efficiencies gained and saving reps time. If you could save three hours from every sales rep a week, that's the messaging that kind of resonates, as well as having that information in their Salesforce. As they do their forecasting and reporting, garbage in, garbage out, right? Now we're getting better information in.

On the investor side, and then it becomes top tier. It's a different conversation. What I'm finding is that, and I’d notice this too in my SAP days, technology for technology's sake, business decision makers don't care.

If anything, they're overwhelmed with how much they've been over promised and under delivered. So it's what we're delivering. It just happens to be AI that's facilitating.  

RM: One of the things I've seen happen with a lot of tools the market is as you're making an AI version of some stuff that exists in other formats that is better, sometimes it's hard to differentiate with your top level messaging, how it's better. Sometimes your competitors can make sort of similar claims, like we have an LP, we have an LP. One of the problems with your differentiators sometimes being technical is that your average buyer doesn't really know how to tell who's telling the truth, right?

I deal with so many IT people, I give the example of like let's say I come to you with an anti-aging pill, right? I'm like hey, this one's based on this science out of Harvard, and this was based on the science out of MIT. You don't really know how to evaluate the science, or which one might be a better pill. Until they've just been out there for a long time, and you have some other way to do, you're looking for some proxy signal. How was that playing out in the sort of sales AI space? Are you concerned about misinformation and miscommunication in the market? How do you drive a wedge through that? How do you help the buyer figure out what's true and what's not?

MW: I feel like a lot of buyers are going in having been burned, and are looking at it understanding how hard the challenges and being skeptical at the outset. What we've been doing is making it extremely easy for them to get going, where we're listed on the Salesforce app exchange, download it. Let's get going see it for yourself. We could talk about how much efficiency you're going to gain. We could talk about how much data you're going to see from this. But rather than talk about it, let's put this in your hands. Let's get feedback from the reps, and let the data speak for itself.

RM: How do you talk to your customers about training AI in the product?

MW: We try and have that be as seamless as possible. The introduction of the templates was a great way of people take notes in an outline structure. Just in following a normal outline is how we're training the model. That's our approach in general with Noted Analytics, is allowing sales reps to work how they want to work. Nobody wants to go into Sierra. Nobody wants to do that. The overriding feedback from reps is, it never gave me any value, right?

But I do take notes in my Word doc, or my Evernote, or whatever. Hey, continue to do that. And the next iteration of where we're going with this AI is then being able to provide insights back to the rep.

Let's say a number of people on your team have identified a particular use case. Is it one that leads to success? Then suggesting that based on hey, this is a similar company based on their industry, their demographics. That kind of information being pushed to the rep is the next thing that's going to be okay, this is what's going to make it sticky in their usage.

RM: Let's take a step back from Noted Analytics a little bit, and let's talk about the industry more broadly. What are some of the trends in AI for sales and CRM that you find most interesting? Are there other companies that you really think are impressive, or other things that you see your customers using that you like to recommend to people?

MW: Yeah, absolutely. It's interesting too in terms of the tools that people are using. It initially seems like a big push in automation, right, being able to get more messages out, being able to personalize those messages, using standard scripts that come from marketing, and inserting them in there. Which helps, but then it gets to the point where now it's everyone's getting bombarded with these messages that are seemingly personal, but aren't, right?

RM: Right. Other tools that I've seen a lot of people come-- or just about how much success they're having with are call recording is Gong.

MW: In terms of being able to playback a meeting and say, OK, how is that objection handled? Or when pricing came up, how did you handle that? And so I think having those and being able to showcase which ones are valuable for everyone else to hear has been a big, big success I've heard from other managers.

RM: We use Gong here. We haven't been doing it for a very long time, but I really like that tool. I think it's pretty cool. How do you recommend people keep up with the industry? This is the industry that moves fast. Do you think your average executive at a tech company, do they need to read the AI news? Do they need to know what open AI's doing? As you think about you guys are building features into your products, do people have to go to conferences and keep up, or do you do let the user's drive it? How do you how do you see some of this going in terms of where AI innovation comes from particularly on the appliance side?

MW: It depends on the line of business. I mean, certainly on the marketing and sales side, I think you have to be knowledgeable of it. I try and keep myself knowledgeable of it. My technical co-founder is certainly much more of a practitioner of it. It's such an ingrained part of all companies and technology in general. I mean, every company calls themselves a technology company now. I read an article in Sweet Greens saying that they're are a technology company, right?

They're are platform and salad is their content as they say. I think that's the mindset, and everyone has to stay current on it, whether it's attending conferences and just meeting other thought leaders and people in that space.

RM: Do you find there are certain types of companies that are more successful with rollouts for tools like this than others? Can you profile that?

MW: The best way I think about this too is in my previous roles when I was at SAP selling SAP Hana, an in-memory database, it was one of those classic tools that you buy in itself, it doesn't do anything. You have to come up with a good use case, and then you really find that. And the companies that were willing to experiment, willing to try something, willing to see if it fails, learn from that, and iterate, and iterate, and iterate were the ones who were successful. The ones who got caught too much and how much of an ROI, what's our time to get our money back, and get consensus, they just paralysis analysis, and just couldn't move forward.

RM: One of the things that we've seen a lot, not just in Talla, but across my AI investments and the startups I've invested in is that the buyer's journey for a lot of these tools is very poorly defined. It's taking shape. They're trying to figure it out. And you can see that because their requirements change while they engage with you, right? And so what's interesting about that is I think if you're in that phase of the market, where it is tough to define your requirements, sometimes it is going to be hard to justify the ROI. You just need to go out and know that hey, you need to try this stuff, you need to experiment.

I like to give people the example that when the internet came along and you could now build a web page, companies had a design process, but it was for paper graphics. What some companies did in 1997, was hey, we want a web page. Let's use our paper graphic design process to get a web page, and give that designed piece of paper to a web developer, and say make this into our web page. That's not really the right process. You should use different tools. You should use a different workflow. You should think about it differently because you can hyperlink. You can do things that you couldn't do on paper.

The way to really attack these new technologies sometimes is I agree, it's definitely the experimentation angle. Every company has got to do it. You've got to accept that yeah, you're going to make some mistakes, but like your output here is the learning. You're going to get smarter. What that's going to allow you to do is deploy that stuff effectively because you understand it.

MW: That's a classic technology story. I remember reading something about the shift from steam to electric power in factories. Before when it was steam power, all the heavy equipment would be closest to where the generator was. Then it would kind of fan out based on how much power they use. Switch to electricity, they had the same layout, and like, wait a minute. We don't need that. We can have power going everywhere.

It it's the same kind of thing. It takes that like second evolution. First, you're experimenting with it, then you learn, and then you can rethink a whole new way to really maximize the gains from it. Do you ever think forward about potential ethical issues in AI use of sales? Is there a chance that 10 or 15 years from now, I have a chat bot just knows how to push my buttons and can sell me anything, and it's just literally I'm buying all kinds of stuff I don't need? I guess number one, do you think that's a possibility?

RM: Number two, how do we make sure that people use this responsibly? And what does it even mean to use it responsibly?

MW: I think to your point, in being pushed to buy something, and also maybe ignoring you. You want to buy something, and they're like no, we're disqualifying you. Why? I want to buy. I can certainly see that. In general, where I see AI is the human element is going to continue to evolve as to how we are involved in that, but oversight. Before, it used to be we were doing 80% of say manual work and then 20% of just reviewing our output. If AI is going to drastically shrink down a lot of those tasks, then our whole entire focus is going to be in that oversight.

I think being able to look at what's happening, and see these trends, and jump in, and say something's wrong here. We're pushing this person too much. We're eliminating a huge group, and stepping in and being able to call an audible, and redirect the AI.

RM: Is there any application of AI outside of sales, outside of your area of expertise that you're particularly excited about-- either something that you've seen that you think, wow, this has a ton of potential, or maybe an opportunity that you see that, you say I don't know why nobody's applying AI to this yet. They should be.

MW: Some of the areas I find most exciting and slash scary are the Google Voice, or that scenario that Google did where they have your assistant call and schedule a haircut. I mean, just the applications of that and being able to just these are the five or six things that I want you to do, completely outsource them, and have them being able to negotiate that schedule, work within your calendar, I think that's very exciting.

The scary evolution of that is then what if it just continues to run its own life, or managing your own life? I find that interesting. In terms of areas that, I'm not sure if I have a good answer for something that's not quite done yet.

RM: Well, it's interesting. I think it took off in 2014. Just sort of 2017, there was this explosion people trying different AI things. What happened in 2018, 2019 was everybody sort of consolidated around figuring out things that were actually both technically feasible and economically viable. They were trying lots of things. We tried some at Talla that just weren't going to work economically now that might in the future.

MW I think a lot of what's changed is actually I would think about the products where I would actually look at customer acquisition. I would look at what are software areas, if you're going to do software product, that are easier to acquire customers than other products, and other categories? Then I would launch an end product there. I've seen that AI companies are a little slower to ramp sometimes, because in addition to sort of an MVP, or product market fit, you have this idea of model market fit, which is you're trying to predict a thing, or automate a thing, or whatever. Can you do it with the data set you have at a high enough probability? Sometimes you can.

What about you? To turn that around, what areas do you think are pretty exciting that haven't been explored yet?

RM: Pretty much everything. I have 63 investments now, and like 52 or 53 of them are AI-focused. I've seen a lot of stuff. I really am a big fan. We were talking about at the dinner that I'm hosting before this. I'm a big fan of AI hardware and where it's going to go. The thing that I think people are missing is we've had 50 or 60 years of this x86 architecture. It works like this. Your memory's here. Your computes here. We move things back and forth like this. Now to start to come out with these you know spiking neuron chips, and these neural network targeted chips, and all those kinds of things, you're seeing people revive a lot of old analog circuitry techniques to do some of this.

They're initially going to be used for sort of lower power, faster versions of things we do today with GPUs. What'll happen is after you have a few years of people programming on those chips, they'll realize well, I can do something about this that I couldn't do with the old architecture. And I think you'll see a real explosion in innovation from that. I can't predict what those things will be, but it's very similar to like the internet coming along. People going to cool stuff with this. I don't know what it'll be, but it'll be something cool and different, because when people grew up on it, it's a new way of thinking.

That's what I'm most excited about. I really, really love human in a loop models because a lot of the investors, I look at some of the best companies. I introduced them to a lot of investors back and say 2016, who would say this is a services company. Because they've got this human in the loop. But most of them that have been good have done a good job of converting those services into machine learning tasks, automation. And so some of them are doing pretty well. I think that's going to continue because I think a human in the loop is a way that you can get a lot of proprietary data sets.

MW: Absolutely.

RM: Last question. You work in this industry. How do you feel about all of the hype that people put forth around killer robots, and taking jobs, and everything else? Is this something you guys think about or talk about? Do you ignore it? Are you concerned about it? Do you think the hype's misplaced?

For the most part, I ignore it. It seems like it's something new when I first started hearing this. And then it's really interesting when you go back and just Google. This same message has been played forever. It's always the machines are taken over. It's always just different iterations and sophistication of these machines. It seems like now, they're getting closer to the point where they could maybe.

Because they're getting more and more capabilities to do so. But I'm not at all worried about it. I don't spend any time thinking about the Boston robots that can jump over our box and do hardcare is now going to kick down my door and harm me.

RM: Good. Well, Matt Walsh, thanks for joining us today. If people want to find out more about Noted Analytics, what's the URL?

MW:  Notedanalytics.com.

RM: Great. Thank you everyone for listening. If you have guests that you would like us to invite onto the program or questions you'd like to see us ask, please send those to podcast@talla.com. With that, we'll see you next week.

Subscribe to AI at Work on iTunes, Google PlayStitcher, SoundCloud or Spotify and share with your network! If you have feedback or questions, we'd love to hear from you at podcast@talla.com or tweet at us @talllainc.