Briefly

TRANSCRIPT

00:00
Adam Stofsky
Hey, Laura, how are you?

00:12
Laura Belmont
Hi, Adam. I'm great. Thank you so much for having me.

00:15
Adam Stofsky
I'm very excited for today's video. We're going to talk about AI and contracting for sales teams. Like what salespeople need to know about AI contracting, which I'm sure a lot of salespeople didn't think they were going to need to know a lot about when they started selling. But I think there's a lot here. So why don't we just get started with a. I don't know. A basic question. Why is, how is selling different in the era of AI? Why? Why? Why is selling anything involved with AI different?

00:46
Laura Belmont
Yeah, Selling AI or AI related products and features really feels like this perfect storm of technical complexity. Meeting new customer concerns. To start with, AI features are really just harder to explain than traditional software and SaaS that we're used to selling, and in part because it's so new. But a large part of this is just the way that AI works and how it's different from the way that traditional software works. When we're thinking about traditional software, we say it's deterministic, meaning if I do A and B, I'm going to get skills C. And that's much easier to sell because you know what the outcome is going to be. And when a buyer knows the outcome, they care less about what's happening under the hood.

01:34
Laura Belmont
But AI is probabilistic, meaning when we do A and B, we're not always getting C. Maybe we'll get D or maybe we'll get seven or eight. So when we're not entirely sure about what the outcome is going to be, that black box and what's under the hood becomes more critical and. And customers tend to struggle with really evaluating something that they can't really see or understand.

02:03
Adam Stofsky
It's interesting because that's. In a way, that's the. That's what's great about generative AI, right? Is that you kind of don't know what you're going to get. That's what makes it exciting. That's what makes it generative. Right. It's making something new.

02:14
Laura Belmont
And that's what's great. When we think about it in this creative context and we're talking about generating content, it gets a little bit trickier when our expectation is that we know what the outcome is going to be. So it's incredibly important for our sales teams to understand that and really get what AI is and how it's different. I will add another point of why selling here is really Difficult is the focus on data. So in other scenarios where we are buying software, yes, we may be entrusting that vendor with our very sensitive data, but we basically are thinking, okay, we trust you to hold onto it and help execute a function.

02:57
Laura Belmont
The generative part of AI raises a lot of concerns because we're seeing buyers worried that I'm giving you some of this data, but could our data end up in an output that some other third party has access to? So data is really having its moment. I think people are much more concerned of what's happening and what's going to happen with it. And then we layer in regulatory uncertainty of are the rules going to change my product and how we have created it? We have the fact that so many stakeholders are coming and telling us that they want AI and they want us to use it. There's just so much more focus right now on these tools than our typical procurement process.

03:40
Laura Belmont
And so it's really important that our sales teams really understand the technical underpinnings as well as really understand what the customer is going to be worried about.

03:51
Adam Stofsky
Right. So yeah, these things are somewhat true of other products, but they're just like that much more true for AI. I think it's interesting that people are actually asking like, how does this, like how is, like what is this, what data is this model trained on? Like what, like how is it using my data? These are questions that sales folks need to at least know the basics of, right?

04:13
Laura Belmont
Exactly. There are certain questions when we're talking about AI features that every sales team should be armed with and should make sure that our companies are arming them with. So the first really is just from the business perspective, like what's the use case? What does the AI actually do? This is true for any product anybody is selling. But what is the use case and what are those outputs? And be able to explain it pretty quickly to non technical people. So that goes without saying. For sales teams, there's nothing different there. But when we're specifically talking about AI, there really are some threshold questions that our team should be able to explain. The first I'd say is really the model architecture. So understanding is this a model that our company developed ourselves, so is it our proprietary product?

05:04
Laura Belmont
Is this something that's incorporating a third party large language model? So does this kind of incorporate OpenAI's ChatGPT or Anthropic's Claude? Was it fine tuned? Meaning we're using that large language model but we're tweaking it so people are going to want to know that because terms and conditions flow down from those third parties. Another really important point is how is the model being deployed? Is it on the cloud, Is it being run locally, is it API based? And that is really important because different deployment methods come with different security concerns of who has access to the data, how is it accessed, how is it transferred? Right. That our security teams are going to want to know as part of their review. And then I'd say the third really big one is what are you doing with customer data?

06:01
Laura Belmont
So the big question people tend to ask is, are you going to be using my data to train your model? Right. That's a big thing. I want to caution. Also there's, I think, a more nuanced point there that I also want to know, are you going to use that data for any other purpose? So we're focusing so much on are you using the model, are using the data to train the model? We should also be looking at, could you use it to develop or improve other products and services. So what are you doing with the data? Is it just to kind of give me my output or are you going to do something else with it? Those are the main, I'd say the top buckets that buyers are interested in.

06:41
Adam Stofsky
Okay, so let me just kind of see if I can recap. I like to try to recap. I don't always do a good job of it, but let's try. So, so basically the bottom line is customers, I'm assuming primarily like sophisticated B2B buyers, but I'm guessing a lot of consumer buyers too. Right. Excuse me. Will have these questions, information about the product for all the reasons we talked about, but the kind of categories of questions that salespeople should really pretty much across sectors have the answers to is what are the use cases for this AI product? What are the outputs? What is it actually doing? Two is, what is the model architecture? So like, is it proprietary? Have we developed our own AI model? Is it based on a large language model or some other huge or 4 unknown kind of AI model?

07:31
Adam Stofsky
And then what are, what is the product doing with customer data? Whether it's my own data from a consumer or the data of my many customers, if I am a business customer of that company, who's selling the AI product? And this is not just like what is the, are you training the model on our data, but what else are you doing with that data? Did I recap that well?

07:54
Laura Belmont
Exactly. Perfect summary. And those are really the gating questions. And that's where you know, if somebody doesn't like answer, you can see procurement tie ups and hold ups. There are definitely more questions that come later. How long is the data retained? Other questions about security and models like hallucinations.

08:13
Adam Stofsky
Right. Does this is the model, what does the hallucination read exactly?

08:16
Laura Belmont
Like you have a human reviewing at a certain point. But I think those are really the key ones and I will just flag Use Case is really important because regulatory requirements can flow down from Use Case. So again, this is something that's really different when we're talking about AI, it's not just about like what is the technology, it's how are we using it. Because the risk level is very different if we're using that technology to maybe make a logo that we're going to use in an advertising marketing campaign or if we're using it that same model to analyze personal information or sensitive information. So that's why Use Case is so important and why it's a little again, different than traditional software.

08:59
Adam Stofsky
Right. Wow. Okay, great. Well, that was a great intro for sales teams. Laura, thank you so much.