[Automated Transcript]

Zachary: Customers are really good at telling you that they have a problem.

They're not as good at telling you exactly what the problem is, and they're really terrible at telling you what the solution should be. And so if you kind of follow that degradation of information and you infuse that into your QBR, 

and you start to ask more and more questions about like, what's missing? When are you doing this? When are you doing your forecast reviews? You ask more and more questions. You know, maybe start to realize they don't need more stages. They need fewer stages, and it needs to be simpler because what's happening is their team is getting too confused and they're just plotting a category down on an opportunity and saying it's gonna close, and that's ruining the forecast. 

Deep: Hello, I'm Deep Dillon, your host, and today on your AI injection, we're joined by Zach Hawley, founder and CEO of Steerco. Zach holds a BS in marketing from Arizona State University and has spent years leading customer success teams. He founded Steerco to calculate real time health scores, generate QBR slide decks in minutes and flag at risk accounts [00:01:00] automatically.

So customer success and account management teams can spend less time on reporting and more on strategic customer work. Zachary, thanks so much for coming on the show. 

Zachary: Yeah, thanks for having me. I really appreciate it. 

Deep: Awesome. Why don't we get started with my kind of usual first question is like, what did people do before Steerco, uh, solution?

What, what, what was that? Process, like what's different with your solution and, you know, maybe walk us through an example so we get like a good flavor for it in a specific context. 

Zachary: I've been working in customer success for about a decade in account management the two years before that.

engagement with the customer has always been paramount in what you have to do in customer success. That's, that's kind of why customer success was invented. It was invented because we have this large install base that will inevitably renew. They don't want to keep talking to a seller, they just wanna know how to use the product or know how to get the most value out of the service as as much as they can.

[00:02:00] And so the CS core. Activity that they had to do was talk to the customer. that started with a lot of how-tos and training, you know, this is way back in like 20 12, 20 13, um, you know, when I was just a baby and, and customer success. it really started to ramp up and evolve into these deeper engagements, more colloquially known as business reviews.

They can be quarterly, they can be called an executive business review, an annual business review, something like that. this was a real core anchor piece to what a customer success manager did with a customer. And these meetings were intended to be highly, highly valuable for the customer. It was all about what they've done so far.

What problems were they facing? Let's plan together for what you needed to do next. They were, they became best practice because they became a source of customer satisfaction. Customers would come in and they'd leave and they'd say, oh my gosh, I know exactly what I need to do next. And they'd feel so loved and cared for, and they'd [00:03:00] feel like they had just the whole red carpet rolled out for them.

and it was wonderful, and it became a source of almost automatic renewal. and then started to evolve a little bit more. The practice started to evolve of asking a little bit more questions, digging a little bit deeper, and the products that started to evolve and upsell and cross-sells started to come out of it.

And so what c SM started doing is they started taking more care in manicuring what these meetings look like. Well, that takes more time. And as the SaaS economy grows, there's more accounts, that get put onto people's books, and you can't really take that much care and time into every single interaction.

So what happens is, I. You're left with two choices, which is you either water down the engagement and try to do it for everybody. So you spread it across evenly like peanut butter or you prioritize and you only do these types of meetings for your strategic accounts. You can imagine, yeah. Go ahead. 

Deep: I'm gonna jump in for a second.

So, why don't we like walk through a little bit more detail on What [00:04:00] a specific QBR is in a specific context. Because when I think of customer success, In a kind of fairly complex SaaS product, world, you know, there's kind of a lot more than just the quarterly level business goals.

That's right. That get done, right? Like, there's like a ton of day to day. Mm-hmm. There's a ton of places where things get stuck. a lot of the customer success activity is in this day-to-day stuff. So what are these, like what's an example of it? Is this, is this going back to the original thing that like got the engagement to happen in the first place and we're talking about a very high level of business goals and where we stand relative to them 

Zachary: Yeah. So a customer typically buys a solution because they have a really big problem that they need to solve, and that problem is measured on a very specific. Set of metrics, right? What happens in the day-to-day is during that implementation, oftentimes the CSM is partnered with people in the customer organization who didn't make that decision.

Their job is to just go execute. And so you get questions back and forth of how, what does this button do? It'd [00:05:00] be really cool if you could do this thing over here. And by the way, this button has rounded corners. We really like it with pointed corners, you know, it gets really, really nitty gritty.

The QBR is an opportunity for everyone to take a step back, level up the conversation. Oftentimes, the executive sponsor or the the buyer is brought back into the conversation as a check on. Are we seeing the forest through the trees here, or are we focusing on the wrong things? You wouldn't believe it.

How many times I've had a QBR with a customer where we spent the last three weeks trying to fix a problem, but during the QBR, we discover that problem is not even relevant to why. 

Deep: It's very easy to rabbit hole on something. Right? Totally. Especially, yeah. Especially as you move down the execution chain within a company.

Mm-hmm. You know, you wind up with people that are more immediately chasing a particular problem. you know, they start spinning around. They're not sure why they're doing what they're doing. The sponsor can get, frustrated or they check in and they, you know, they're like, well, that's where it was a month ago or two months ago.

Why, why are we still there? Right. 

Zachary: But you, to your point, there's so much day-to-day activity that [00:06:00] happens that now you need to recap back in these business review conversations. You need to level it back up and say, all right, here's all the stuff that we've been doing. Here's the impact that it's been making.

Back in the day, you didn't really have too many levers to pull, to pull reports on. It was like product usage and like support tickets. Mm-hmm. That was really it. But then there was the infusion of more survey data. We started surveying customers more. there was the infusion of. Different types of surveys.

So not only are you surveying just at an NPS level, which is, you know, would you refer this? It's CSAT and psat, and then you segment by who's using it the most and who's not using it. And then you have all these mini, minor projects that are going on. The data became really overwhelming. And then the way that I coached my teams to set themselves apart was, Hey, by the way, try to help your champions, connect it to what's happening at large in the company.

So if the company just announced a major acquisition, with another company and you think that our software or our product will [00:07:00] can help with that acquisition, make sure to mention it. But to do that, you have to have your eyes elsewhere, all while still focusing on the project and the deployment and the engagement at hand ultimately.

It became too much. And so what happens again is you, you either just choose to ignore that stuff and you just try to get as much data in front of your customers as possible and sell them on really good news. or you do that really high level engagement, but you only do it with a few customers.

And so up until recently with the advent of AI which is really good at parsing all sorts of data from all over the place and providing it back to you in a digestible form, you really had to make a choice to either water down your business review or do it with only a few people. And we asked the question at Steerco, why can't you just do a deep engagement with everybody?

And we feel like we're proving that. 

Deep: one of the things that you bring up is an important challenge you know, like if you're a consultant or a customer success manager, let's say a higher stakes client. for you to really understand things at a business level, you have to [00:08:00] understand the client's business.

which in and of itself is a fair amount of effort and energy and is always dynamic and changing. So walk us through what's different with your product what are all the feeds that you're tapped into? What are the data sources into your product, and what kinds of interpretations are you doing?

Zachary: Yeah. You're, you're totally right you need to understand your customer's business, but what's happened is because the subscription economy has grown, everybody's businesses have gotten larger and they've realized they can't just keep hiring CSMs to keep up with the customers.

I just talked with somebody the other day who their enterprise team manages a hundred separate customers each. Yeah, they try to verticalize 

Deep: a little bit. Mm-hmm. Like, you know, I mean, if you're doing projects in healthcare, I, it's, it's its own lexicon, its own like, confused, set of partners that interact, right.

You're starting from scratch with each customer success manager, they're not gonna be able to really dive in at that business level. They're gonna maybe deal with your product. 'cause [00:09:00] that's a repeat across all the customers, but contextualizing it for the client might be hard. 

Zachary: Yeah. So we point our AI solution at not only your CRM and like your support ticketing, system, but we also pointed at your product usage.

We point it at your customer success platform. We're pointing it at your survey data that you're gathering, and we're pulling all of that stuff in. But at the same time, we're also pointing all of our AI externally. We're pointing it at sources like PitchBook. At LinkedIn, we're pointing it at sources like the Edgar system, which scans 10 Ks and 10 Qs, and you can grab the latest data there.

We're also, pointing it at the, press release wire as well. So we're really trying to bring in contextualized information about the customer that you're working with quickly so you can really understand. And then we actually take it one level up where we track when's the last time you had an engagement with them, and if you are about to have another engagement with them, you can see what's changed, [00:10:00] So we, we are scraping all this data. We're, we can gather it every time you click this, this button that says, you know, start engagement and it's this pre-written prompt. Basically allows you to say, okay, well last time you talked about X, Y, and z this time where, and you're about to meet with them. These X and Y have changed in these ways.

Make sure to bring this up. And we don't wanna over prescribe it because we don't believe that AI is really replacing the human, it's just replacing the manual work they have to do. So what we've done is we've allowed the AI to say, here's the data point you wanna show, here's a screenshot for your slides.

and also here's three to five questions you might wanna ask your customer about this data point. Because one thing that I've had a lot of success with and the teams that I've managed, I've had a lot of success with is in A QBR, if you're the only one talking, it's a bad QBR. You need to be able to use the information that you're bringing to open up and engage the customer more and ask deeper questions.

Deep: What's the point of all the questioning? Like are you basically trying to expand the discovery and get [00:11:00] more scope of work? I mean, that's basically your bias. Mm-hmm. As a customer success person. Right. And then also to like cement the satisfaction with the existing services.

Zachary: Yeah, I might, just a slight nuance to what you said, I, I would say that A CSM only recently has become charged with expanding all scopes of work at the core. I think A CSM is really just trying to understand and make sure that they get to the root solution of the root problem. And what I've learned in my decade plus of working with customers customers are really good at telling you that they have a problem.

They're not as good at telling you exactly what the problem is, and they're really terrible at telling you what the solution should be. And so if you kind of follow that degradation of information and you infuse that into your QBR, the first software I sold was, Sales, analytics software. It was about reviewing your pipeline, reviewing your funnel. I would work with VPs of sales who would say, something along the lines of like, our forecasts are not accurate. you need to fix that. They ask you like, we need all these [00:12:00] different stages.

And you start to ask more and more questions about like, Why are you saying it's not accurate? What's, what's missing? When are you doing this? When are you doing your forecast reviews? You ask more and more questions. You start to realize they don't need more stages. They need fewer stages, and it needs to be simpler because what's happening is their team is getting too confused and they're just plotting a category down on an opportunity and saying it's gonna close, and that's ruining the forecast.

Deep: I wanna go back to what you actually put into your, yeah. Your LLMs and where you get it from and how you transform it. Then I wanna understand better, like what the interface looks like. Is it just Question answering or conversational? Or is there something else about, so you mentioned a number of reports, um, that get generated.

Mm-hmm. I'm assuming most of those are external, uh, products. can you walk us through them just like kind of one at a time let's just talk about each one. So, 

Zachary: yeah. So like A CRM is an external product, right? Yeah. So you use Salesforce or HubSpot. We connect with that, right? So 

Deep: you, yeah. So you've got like Salesforce or HubSpot.

You guys have built an [00:13:00] API pipe into the Salesforce. So you have like a full set of interactions between the customer that your CS person is supporting and their customers. 

Zachary: That's right. 

Deep: Uhhuh. And so now, that's a pipe that you guys update to GPT or, or whatever. that seems pretty straightforward. and then similarly you had some other reporting ones that you were mentioning. 

Zachary: yeah. So like if somebody has like a data lake, like a, like a quick site or something, we do the same thing. We build an API to their data lake and pull that stuff in.

But we also pull publicly, we ask, the deep research tool of our, of our API, to go out and search the web and pull all of the most recently publicly available information about this particular customer. Which is what nobody else is really doing right now. Everyone's focusing on, and that's some 

Deep: like news, like business wire and stuff, or could be, 

Zachary: it could be the news.

It could be, like I was saying, Edgar, which is the SEC filing site, right? So a company files their annual report and the CEO in the annual report always has a letter to shareholders. In that letter to shareholders, he or she might [00:14:00] say, we're focusing on X, Y, Z this year. If one of those initiatives maps to what you do as a CSM, you may want to mention that because this now connects you to A-C-A-C-E-O initiative.

Deep: So one question I wanna ask you is, with the exception of the private information, but this public information's already, Kind of exhaustively indexed by open AI and The others. So are you guys. Massaging, manipulating it and contextualizing it before you reason with it.

And are you doing it on the fly or is this an apriori task, or are you relying on the background knowledge of the LLM? 

Zachary: No, we're doing it on the fly. what I realized when AI, like OpenAI started to become really popular, you know, the chat, GPT, the Gemini, all, you know, perplexity, all that stuff.

the company I was working at at the time had this, policy, you know, everyone was very scared of ai. They're like, do not use it. You're not allowed to use it. Don't put sensitive customer information that didn't these tools, well, you know, everybody's using everyone. Everyone said, sure. They [00:15:00] crossed their fingers and then they just went to their personal laptop, their personal phone, and they started doing it.

And while that feels like it's more efficient. It's still not, because now everyone's doing it separately. So there's no consistency across the board. Everyone's training their own, GPT to do what they want it to do. it's expanding on, good rep behavior. So the good reps who know to go search all these external sites and bring that stuff in is great.

But then you've got reps who are just not that great, maybe just not that knowledgeable, maybe not that tenured. They don't know what to look for, and they're hoping the AI will just do it for them. And it, it hallucinates a little bit and fills in the gaps and it's not that great. And so what I was noticing was there was a real yearning for a, pre programmed prompt.

Really? Yeah. Which is like, not that we can prompt better than any individual, but what we can do is we can consolidate it, make it consistent. We can firewall it against, the greater. Company so that things don't go back and train the model. You can feel comfortable [00:16:00] uploading sensitive customer data.

Ultimately, it's leaning into the solution more than what I think a lot of publicly traded companies are doing right now. And so we're doing it on the fly, but we're prepackaging the prompt. 

Deep: Yeah. So you user asks the question, you translate that, you figure out what data sources to go to. You formulate anything from a search query to a, a SQL lookup or something to an API call you retrieve raw ingredients.

You send it to the LLM for a response generation. Maybe you send up some profile information about the asker, the requester and the customer or something like that. 

Zachary: Yeah. I'll actually go back one step. The, the CSM can eventually refine the response by doing that process you just said, but we've actually backed it way up where we don't even say that the C SM should ask a question.

We assume we know what the question is that they're about to ask. Yeah. Which is, Hey, help me generate this QBR. And so what we've done is there's no prompting to trigger the [00:17:00] process, which is where I think, by the way, AI is going. we've exhausted this idea of having an assistant where you can chat to them and, and generate something from scratch.

We've said, no, no, no. We know what you're about to ask. Click this button. We'll do the task for you already. And then if you wanna go from there and refine it using a chat bot, you absolutely can, but we're going to actually automate your entire workflow at the click of a button when you want us to. 

Deep: So is that always generation of these reports and so you have like a whole kind of ag agentic back and forth reasoning to That's right.

Kind of formulate the report. 

Zachary: Yeah, that's right. Uhhuh. 

Deep: Are there other use cases beyond the report where you're digging into the LLM. 

Zachary: Yeah. So not only are we the, the quarterly business review is just one. but the other thing we're, we're populating a bunch of different opportunities for works.

Another, another big form of manual work that A CSM has to do is a success plan. This is typically at the beginning of an engagement with a customer where you're outlining what it is you wanna achieve, what metrics do [00:18:00] you wanna measure, and how will we know that this partnership is a success at the end of your contract?

Right? These, by and large, started really well, and then like anything became kind of showy and really just more something for performance. they were hard to continue on and very manual long term, because if you're gonna pick these KPIs at the beginning, they may adjust, they may, whatever. So what we decided to do is we said, okay.

Scrap all that work. We, we can use our agent to do it. Click creed success plan. You can refine it, you can pick the KPIs. It will then automatically track those KPIs live throughout your tenure with the customer. And then this can really be a live document that you and your customer work on where you can pick a KPI and be like, oh, it's starting to shrink.

Or, oh, you know what? We're not focused on that anymore. Let's adjust the success plan really quick. And you can go to the AI and say, Hey, new initiative, this is what we're doing. We're no longer focused on this. And it will snapshot it, archive it, and build a new success plan for you, which is something that success that CSMs are [00:19:00] supposed to do.

But it's wildly manual. And so the ability to do this is gonna let CSMs really focus on the conversation with the customer. Like, wait, why did that change? What's the initiative that changed? What's changing in your business that's changing the KPI that we need to now focus on? Rather than just trying to keep up on, well, what's the benchmark and what do we do?

And how did it change, 

Deep: what level of changes are you seeing by, you know, the CSMs? Once you've generated this rapport, are they just like accepting it and shipping it off? Are they making massive changes? Are they, is there like a lot of back and forth to kind of massage the output and, and sort of dig deeper?

Like what, what are you seeing? How do you measure it? How do you know you're not just putting, you know, your own people to sleep and they're just checking stuff off and letting it 

Zachary: Yeah, yeah. Generally in our, in our pilot groups, we're finding people that are, they're accepting about 70 to 80% of it, and they're using the last 20 to 30% with the customer.

So they're actually [00:20:00] meeting into steerco with the customer and talking about the success plan that was generated, and then refining it on the margins, and, which I think is actually really good. It means that the AI is doing a, fairly accurate job, but it's missing the nuance, which you would expect.

And we always tell our CSMs too, it's gonna get you 70 to 80% of the way there. It's, it's probably not gonna get you a hundred. So be careful. 

Deep: the CSMs are like, are they actively vetting the, the information coming back from the LLMs and how do you deal with citations and just hallucinations and all of that?

Zachary: Yeah, the best part is we've, uh, really programmed our AI to source every statement it makes. So when you click into it, you'll be able to see its rationale. It goes back to a backend. It says, well, we pulled this reasoning from this part of the product usage or from this article, or whatever. So you can always understand where the AI is getting its stuff.

But to your point of how do they vet it? the encouragement that we give is that you don't vet it by yourself, vet it with your customer. And what you're doing in that regard is you're now [00:21:00] focused not on each other. You're focused on what are we doing together? Is this right? Did what this technology has spit out, right?

Is it right, is it wrong? And then people love to correct. Not just each other. People love to correct anything. So when a customer sees something, they're like, Ooh, it's close, but I actually would say it's this. Or, you know, that's not quite the KPI we need, we actually need this over here. It's 

Deep: curious, it's, it's curious that you're inviting your customer's customer into and acknowledging that an AI generated this thing.

Like, Tell me like how you got to that as opposed to letting you know your CS folks pat themselves on the back for being more efficient and never bothering to tell their customers about it. 

Zachary: I think that's where the stigma that AI is actually coming from. I think it's people pretending they're not using ai and so they pass something off and then everyone's like, well, was that ai?

Was that not? So what we decided to do is we said, Hey, lean into it. This is gonna help you. It's not gonna be perfect, but it's gonna help you create a relationship with your customer. If you could talk about how cool this technology is and talk about where it was wrong and where it [00:22:00] was right, and hopefully what it did.

Was it started the conversation for you, which by the way is the hardest part. What questions do I ask? How do I get to this part? If it gets you to that point where now you're just refining, you're not creating, both of you can have a really constructive conversation. And by the way, the, patting themselves on the back CSMs don't pat themselves on the back when a success plan is created.

They pat themselves on the back when the customer says, wow, this was great. I really appreciated this conversation. They don't really care about the actual document itself. Are you 

Deep: seeing any, I mean, it seems like there's a risk of, relying on the bot to place the narrative framework you know, and now the dialogue is steering around what bots think so what do you guys do to mitigate and manage that?

Like is there a even a strategic outlier part in the process where the CSM is orchestrating. Going for the gusto and having the bot put the whole report together. Sure it saves time, but it also, you know, ends up spewing a lot of stuff that, anchors the narrative.

And it might not be in the thing that [00:23:00] really matters. It might just create a bunch of FUD that you waste your time with. 

Zachary: you have worked with AI before. It's very clear. Obviously that's all we do. All we do is these systems. I know, I can tell, It's fun to talk to somebody who's been around because one of that was one of the things we started realizing in our testing and in our qa.

And so what we decided to do was implement a feature called strategic guidance. What that does is it allows you to, in any period of time, you know, you can, pick a strategy that you wanna implement across all QS or across all success plans, across all account reviews and account plans, whatever, and you can make sure that that strategy is.

inherent in every part of what you're doing with the customer. So let's say there's a VP of customer success and they have found through all of this data, they've been sifting through, that the leading indicator for churn is executive satisfaction. If the VP at the customer is not happy, is a huge leading indicator for churn.

It doesn't matter what they're doing or how much they're using the product. If I'm a VP of [00:24:00] Cs, and that is something I've found, I can now go into steerco. Add strategic guidance that says, Hey, in your success plans, in your qs, make sure you're controlling for executive satisfaction regardless of what all the other data says.

You need to make sure that that's a focus, because we know that if we improve executive satisfaction, we improve their odds of returning. And so to combat what you just said, which is where the agent or the bot. Ultimately puts its own narrative in, based on what the data is saying. We have allowed, individual CSMs to control the strategic guidance of their book managers to control the strategic guidance for their team and VPs and C-Suite all the way down.

the view, the vision is that we can actually get, macro strategies as well. So maybe there's really great thought leaders in the space. I know like the CEO of Gainsight, Nick Metta, he is fantastic. He's constantly out there, uh, talking about what CSM should be doing and how they should be talking to customers.

Our dream is for him to actually come out and say, Hey, using Steerco, here's what your strategic [00:25:00] guidance for your QBR should be this quarter. And anybody at any company using Steerco could say, oh, I wanna use M Meta's guidance, and they can click that and boom, that's how they'll operate on their QBR or success plans.

Deep: Yeah, what have you found overall, it sounds like your goal is to go and try to. To some extent capture the quality of QBR that was done on, you know, maybe white gloved customers and like, take that down a notch or two so that you can hit like a broader reach and then try to jumpstart, more strategic conversations, whereas before they maybe were being avoided or just didn't end up happening due to the day-to-day kind of operations issues.

Is that your assessment? 

Zachary: it is. I think one, a addition there is that, a lot of times CSMs became overworked and so the engagement was semi deep and some, some CSMs got really good at pretending that they did a really good job when really what they were doing was somewhat performative. slap a new coat of paint on it and good to go.

This is a really great way to spend the same amount of effort that you would do doing that, but [00:26:00] actually creating something that's of value. So it's, it's leveling you up and being able to do it more, better, faster across your entire book. 

Deep: when you're the, customer success manager, and you do this, you're sort of like putting your opinion out there.

Mm-hmm. And it's sort of naturally filtered through the lens of what you are incentivized by. If you're the customer and you receive that, you kind of know that, are you guys positioning this thing as a third entity that's, maybe not obviously on the side of the CS manager. Is that kind of the positioning or is it more like, a, an extension of the CS manager?

Zachary: Yeah. It's meant to be a neutral arbiter. So it's meant to be the, thing that sits in the middle, analyzes the data and just spits out what the data is telling it to spit out, which is what every customer expects of their CSM. But csm, like you said, have a bias towards what they're compensated on and what their incentive I've done.

It should be what the customer's also 

Deep: incentivized. usually what that is is to hide the [00:27:00] warts of the product. Mm-hmm. to avoid bringing too much light and attention to features that the product manager told them they're not gonna build. Mm-hmm. and then trying to minimize the amount of time and work that goes into this engagement, which usually means not trying to do a bunch of stuff.

it's very much an outcome of the SaaS model. Like it's a fixed flat rate for whatever category of consumption the customer's in, and you're trying to minimize time with the customer and maximize happiness, so, 

Zachary: mm-hmm. 

Deep: That ends up in all kinds of decisions, whereas this kind of approach is really trying to like hone in more on customer and what they wanna actually achieve that could end up really screwing up the plan of the CS manager. Do you run into that at all? Or, 

Zachary: That's the biggest risk that we've heard in, all of our prospect interviews is like, what if this actually just tells me what the customer wants and what the customer wants is not what we offer.

my response to them [00:28:00] is they probably already know. and you are doing a disservice by just trying not to talk about it. with the amount of data that's available and the amount of crowding that's in SaaS, most of these SaaS competitors are starting to eat each other up publicly, and they're calling out each other's warts and it gets pretty dirty and it flies around.

this is where you can take back control of the narrative and just start to lean into it and say, Hey, you know what? I know that you think this is really important to you. This is something you really wanna do. Just so you know, this is where we think we can get you a lot of value. if, if that's not for you, we should part ways, but I think that something that customers are really yearning for is authenticity, and they're starting not to get it, which is why you're starting to see a lot of people out there saying like, I can't wait to just deal with a bot because I don't want somebody to just slap a graphic on a slide, send it to me.

And they never even looked at it. Right. They, they want to know Yeah. What's actually happening with their engagement. 

Deep: Yeah. I mean, one of the values I've seen of customer success [00:29:00] that's very difficult for a bot to replicate. And I think it's probably the core value of a, of a good CS manager Is really just the fact that the customer has somebody to like look at, talk to get together with. It's that, that human touch side. Do you think there's an opportunity in your kind of framing here to enhance that? Or do you think, are you trying to like get the bot to play some of that role in some of that empathy?

Like how do you guys think about that? 

Zachary: No, we really want to allow the CSM to have the space to think about their empathy, to think about the customer, to think about all that, to meet the customer where they are. The problem is with the amount of work that every employee now has to do, the amount, the number of accounts that they have to manage, there's no space for it.

They're too busy in the data, they're too busy in the changes. The delta to think about, well, what does this actually mean to the person I'm talking to? How are they gonna be able to present it to their boss? This person now, after our conversation has to walk into an ELT meeting and executive leadership team meeting.

[00:30:00] Explain that when they stuck their neck out for our solution, it worked, right? Yeah. Or it didn't work, or something's not going well. And a lot of times that part is lost because there's only so much how many hours in a day? And so they're spending all their time just trying to get the data that the customer wants and they're not thinking about what it means.

Hopefully this can eliminate that first part. So you can actually start to think about what does the narrative look like? What does the person need to feel? How can I help them along in that path? And how can I ultimately teach them rather than just tell them? 

Deep: So I'm gonna switch gears a little bit. And that's like, should this be done? And what are the ethical ramifications of it? And not only with respect to your specific kind of manifestation and incarnation of it, but like what are the second order of effects?

Like let's assume for a moment you're wildly successful. Part of, what I'm after is. If we rewind 20 years ago and we had a conversation with Mark Zuckerberg and we talked about what are the second order effects we're worried about from you, I think he probably would've [00:31:00] said just about nothing.

But he certainly wouldn't have said, oh, I'm really worried about 14-year-old girls who were suicidal ideation rates going through the roof. I'm really worried about orchestrating or facilitating the orchestration of a genocide in, Myanmar because I, don't wanna bother getting who or fluent in the language to curate it.

Those are not things that anyone in Facebook or you know, meta set out to do. But these are things that are very much tied back to that company anyway.

I'm curious one of the things that I feel we as technologists and I put myself in this bucket sort of fail at is we're all in this giant ecosystem of here's the cheese you need to run down this treadmill to get the cheese from the VCs to put the money into the company, and then you start growing.

But relatively few of us sit down and a really imagines the kind of second order effects of success, what happens here? Both the good and the bad, but I'm mostly interested in the bad. If you're wildly successful. 'cause I get the good, it makes sense. [00:32:00] as an end consumer, you're interacting with a SaaS product you know, that is now much more effective at helping you with your customers. you get to think about your business in more depth. I feel like the value proposition you're bringing to the table seems pretty clear, But what goes south? how do you think about it? Yeah. 

Zachary: Yeah. Well, I can tell you it's certainly not, a political uprising in the Congo.

That's probably not gonna come outta mass adoption of steerco. Right? Yeah. But I see your point, and, and something that I've been thinking about a lot is, as SaaS has grown, the number of customer success managers as a position, the, a number of recs that have been open has grown proportionally with it.

The problem with that is companies are finding out that's really not tenable. there's like a faltering effect where people think kind of binary and they think, well then we need to eliminate customer success managers altogether, and it needs to be a revenue center, and we need to go account management, and they need to drive, upsell and cross sell, and they need to be driving revenue, especially in a market like [00:33:00] this.

And our hope is that what we do is we change the conversation a little bit. We say, no, no, no, no, no, no. That you don't need to go to zero, but. You know, maybe the negative effect of this is it doesn't open as many customer success opportunities in the future because more CSMs can manage more customers, which means that it kind of halts that straight proportional up, You can start to, make it more asymptotic. Right? And that's okay. I think that that is fine if you create a, a finite ratio that A CSM can manage, I think that's really, really good. What I don't want to occur, and, and what is starting to happen at the ground level is, there's been a lot of pressure on CS teams to, to drive immediate revenue.

And that's just not what CS teams were created to do. They were created to drive long-term value. There's literally a, a metric around it, LTV, lifetime value of the customer. That is what CSMs were created to do. And somehow that's been lost recently because the departments cost so much, and so they're saying.

Start [00:34:00] bringing in the upsell. Yeah. Start selling your customers, you know, and I don't like that. I, I would rather get back to let's help them build the lifetime value of customers. But to your point, not to deflect. I think what that means is if we can allow CSMs to manage more accounts, that means that there are fewer CSMs that are ultimately needed in the marketplace.

But I would rather fewer than 

Deep: zero. Well, I mean, I don't know. Maybe, but part of your argument is that, defacto CSMs can't handle their load and they're being pushed up into these white glove arenas in the more high value customers. that generally jives with me. You know, like, I don't know if it's completely true, but my sense is that, there's probably a lot of truth there.

Mm-hmm. One of the things I wonder about a lot is, and this isn't even unique to your domain, but I. I get how a company's best CS managers are gonna be required. moving forward. I have a harder time with the B and C players and understanding what happens to the mediocre in this new world where, and I'm talking like one or two years out, three years out.

Zachary: Hmm. 

Deep: [00:35:00] Where, where the models are, picking up more and more of the slack. And we live in this, very achievement oriented society, but they're just like a massive percentage of the population that's never gonna be the top 1% at anything.

And that doesn't make them bad people, and it doesn't mean that they don't have a right to live well. so what happens in an arena where the. bots are pushing the envelope, the level of capability and reasoning is going up faster than you. Or I can probably keep straight because it's, you know, every other day there's new models coming out that are just so much more powerful than the ones that I looked at from just a few months ago.

And so that means your service is gonna get more and more and more capable. Mm-hmm. What does the world have to offer? two types of people? A, that kinda mid performer and B, the, the new kids coming outta school, right? Like, 

Zachary: yeah. 

Deep: 10 years ago if you were a software engineer, you got a good five, 10 years before you had to start managing software engineers.

Mm-hmm. The day you come out, you're basically doing all the things of a software manager [00:36:00] just to interact with GPT. ' everybody's being forced up into the strategic thinking. Like, how, how do you see that?

Zachary: Yeah. I think that it depends on how you grade a, B and C player. So, I, a quick story. I, I've managed a lot of CSMs in my career. One that I managed recently, man, I had a conversation with his manager. So many times, about how he's a great CSM, but he just can't figure out how to stay on task or create a presentation or pull the data that he needs to pull.

It was like the second that needed to happen, he just shut down. But holy moly, if you could get him on the phone with a customer, it was lights out. They loved him. Yeah. They thought he was the greatest thing. He thought he made them feel so special he could educate them on what they needed to do so quickly.

And we even joked back in the day, like, I wish we could just get him an assistant so that he didn't have to do this. But because he had to, he became, you know, honestly like a C player. Right. Yeah. And so when you think about [00:37:00] like, how do you grade B and C players? I think this takes the people who are good at the core of the job.

Or maybe failing on the margins. I think if this eliminates the margins and they can be showcase what they're really good at, I think this really helps them for new people. I have told a couple of prospects, that we're talking to right now and we've been talking about like, how can this be beneficial long term?

I think this, shallows the learning curve so quick because no longer are you overwhelmed with all these different data sources that you know need to go pull reports from. And you know, what does a customer wanna know? You now can click a button. It's right there and it's up to you to have a conversation, a real human conversation with the customer about what this means.

And you don't need to feel scared about asking, is this good, is this bad? Because you just brought them something they didn't know already. So you're already starting ahead. I think to the extremes, the people who are really average or below average at all the things, this is gonna eliminate them into [00:38:00] oblivion.

But I think the people that have a core strength that you saw somewhere in your interview, or the reason that you keep them around this is going to help them, or the technology that's emerging is really gonna help them shine, and help them do what they do best. 

Deep: Yeah. One of the things that I wonder about is like, a lot of times if I'm in a, deep new area that I don't understand and I'm, going back and forth with GPT on it and, you know, using one of the models that have a lot of high reasoning, I'm pretty confident in them.

I find that there is an entire process just to get to the point where you even understand it and something like a deep strategy report for a client. Sort of presupposes a deeper understanding of their business. There's like a lot of stuff in there that I imagine for you to stand behind it and, talk about it, sort of presuppose that you understand it.

And it feels to me like you might need to start building and tooling to even understand the report. 'cause these things are gonna get really, really good, really, really fast. And maybe that's where your [00:39:00] conversational capabilities have come in. or maybe you wanna like, test the CSMs before they present them.

'cause it's not, yeah, I mean, right now maybe you're seeing 70, 80% efficacy in reports, but you're probably gonna be seeing a lot higher than that, you know? Mm-hmm. You're, you're gonna start seeing reports that, you know, people don't quite grok all the elements of it. And it could be the customers too, it could be your CSMs, but it feels to me like an arena that's gonna be become increasingly important.

Is the ability to get what the bot said understood by the listener. And it's not, 'cause the bot's not saying the right, stuff. I'm like on the advisory board at my, university's, electrical engineering department.

So I was just back there for a few days and it struck me that I was like, oh my God, it feels so nice to be surrounded by like such smart people it kind of hit me like there's not a ton of places in society where folks are very comfortable and used to interacting all day every day with people who are capable of passing PhD qualifying exams.

So extreme domain experts. 

Zachary: Mm-hmm. You 

Deep: know, oh [00:40:00] one, the oh one model is, scoring in the mid nineties on, PhD quals. It's gonna be, even higher in the next year. And for you to talk to somebody, not just one entity, but it's the equivalent of being in grad school, being able to pick any expert on any field, throw them into conversation.

It takes a lot to be able to have those conversations. Like that's, 

So it feels to me like that's what's gonna be increasingly missing is like, humans are just not gonna be as smart and they're gonna, we're gonna have to make them smarter to be able to keep up. 

Zachary: I think CS is an, in an interesting position in this, this conundrum that you've laid out in that, customer success managers are experts of the product that they manage, but they're not experts of their customer's business.

what we've programmed in steerco, specifically back to the earlier part of our conversation, is we program questions for the CSM to ask their customer. And the idea is, it's okay to not know, you know, how this [00:41:00] software or how this product impacts your customer's business. 'cause you're not in the rooms your customers in.

and you can teach somebody a product and they can get their like PhD in that product. But what is really hard and what a good CSM does is they do code switching. They're able to say, okay, my product does this. You need that over there. How do I help you connect the dots? Yeah. And what questions can I ask to where you connect the dots?

It's very ineffectual for a CSM to come in and be like, I've got the answer. Here's the playbook. This is what you do. It lasts like three weeks. Right. But if the customer can get there on their own because of the questions you ask, one, they've solved the problem and they've, they look like a rockstar, but two, they associate this dopamine hit this feeling of success with the person they were talking to when it clicked.

And I think that, again, you know, not to, I, I sound, I feel like I'm sounding very Pollyanna about what AI can do, but I do feel like it's, if you understand and start from [00:42:00] a, an assumption that you're not trying to replace critical thinking, you're trying to replace manual tasks that really helps you. If you go into an arena where you say, I also want it to do my critical thinking, for me, I think that's where you're gonna get into a lot of trouble.

I think it's also gonna make a. A couple more mistakes and, and maybe it starts to reason a little bit more to where the recipient doesn't really even understand what it's saying because it's so high level. I really do feel like what we're really trying to do is build a tool that takes away the manual work that might involve some reasoning but not critical thinking.

And those two things I think are nuanced and different. 

Deep: interesting. I mean, I probably disagree with you. I probably think like, even if you're trying to do that, that these models are gonna start to take, the critical thinking and do really well with it. and you customers are gonna, I mean, 'cause they're gonna look at it and they're gonna be like, wow, okay.

That's really interesting idea. if you're prompting it, well, if you're being clever about it. And, providing the right structure and guidance to the [00:43:00] models, cause it's just like a level of thinking that probably wasn't regularly applied to this stuff before.

Mm-hmm. but I think that somebody saying something brilliant is entirely pointless if the listeners don't understand it. Right. And so I think that's probably where your challenges is going to be as you get better at this. And as the models get better is like getting both parties to gr and come along.

Yeah. I know this sounds really weird that we're in this world where the machines are smarter than the people, but I think we largely are already there. So 

Zachary: we, we've been there for a long time. I think there have been professions that I think could have been wiped off the face of the earth because of technology before ai, I mean, you can look at real estate agents.

Yeah. You know, why do real estate agents still exist? I mean, the technology truly does exist to do everything that a real estate agent can do. Yeah. 

Deep: Because you just, you, at the end of the day, when you're gonna make such a big purchase, you just need a human that you trust. You need to 

Zachary: feel, you need to feel to stand next you 

Deep: Yeah.

That's ultimately what it's about. you mentioned this, [00:44:00] person that was really good on the human touch side. I think you might find that, that ends up being your best employee in a two or three years down the line.

Like that person that's like super high eq, super strong capability to make a customer feel like good after they talk to them. Because a lot of the smart stuff that we've valued for the last 50 years is gonna start to become table stakes When really high cognitive reasoning capability is commoditized.

And that's already happened. 200 bucks a month with unlimited, oh one access, you know, that's already an entity that scores a five on all the AP exams scores in the 90 percentile on all the PhD quals. I mean, like, it's a really deep thinker on some level. It's also very fake and like shallow and parody and it's very easy for it to screw up.

But on raw reasoning tasks, it's pretty impressive. So, I wanna end with the question that I always end on. Let's, lets fast forward five or 10 years out, everything that you're sort of imagining and [00:45:00] envisioning, you know, happens. What does the world look like? how has the world benefited from your work and maybe what are the areas that you're concerned about?

Zachary: I think it goes right to what we were just talking about, which is we stop over indexing on. Somebody's ability to memorize data, to pull together fancy charts and, you know, there's literally a World Series of Excel right now that is televised. Right. I think we, I think, by the way, it's fascinating.

It's really, really good. But, I think we stop over indexing on that. I think you start to see a lot of leaders emerge that are just really effective communicators. You know, we made this shift a while back from Steve Jobs, you know, in Silicon Valley. He was known as this like gruff direct person. and then a bunch of founders came out of that and they just decided to be jerks, right?

Because they were like, well, Steve Job is, is Steve Jobs. 

Deep: So it was good. 

Zachary: Bingo. And, and they had these fancy degrees and all these great credentials. So they felt the need to be able to do that. Well [00:46:00] then it was like, okay, well now that's actually not really that important. You do need to be smart and kind.

And I actually wonder if, uh, the future looks like somebody who's kind and articulate. And I think both of those things are probably what's gonna be, coming out of this. You need to be able to connect with people because people are gonna be able to sniff out. If you used AI to write that paper, they're, they know I've used AI to write stuff.

You can tell it sounds robotic, even if it doesn't sound robotic. It's just a matter of 

Deep: tailoring the prompting though, that, that maybe, 

Zachary: maybe, but it doesn't really, I don't know if I've ever connected emotionally with a piece that, AI has written truthfully. And I can tell you that if connect with 

Deep: any, if you've connected with any pieces in the last year, you've connected with pieces that have had AI involvement.

Zachary: Different, different from pure AI written. I guess what I'm saying is I think that there's gonna be much more of an emphasis on. How do you communicate and what is the emotion that you can convey? Yeah, 

Deep: I definitely agree with you that the oratory skills are [00:47:00] gonna be increasingly important.

Presentation skills mm-hmm. Are gonna be increasingly important. the social aspects I think will be increasingly important because at the end of the day, these decisions are, are still very trust oriented and trust anchored. Like to keep the system, get rid of it, switch it out. Like all those things they're not just kind of cold hearted, factual decisions.

we're social creatures and so we, we need some of that social input. but at the same time, I do think you can connect with AI words. I'll ask you a, a simple exercise. This is something that a friend of mine, he had to give a commencement address at a high school.

And I said, oh, well that's, super easy. He's like, I can't get anything out of, GBT. I was like, well just tell it to write, your commencement, speech as if you were, Sylvia Plath and you were spending the weekend right. With, with Jerry Springer. And that's it. And then all of a sudden he reads, he is like, oh my God, this is so good.

Zachary: But, but to your point, and maybe there's the answer somewhere in between, it still depends on his, ability to, orate that [00:48:00] speech, to emphasize the right word. Oh, yeah, yeah, yeah. To insert pregnant pauses. Right? Yeah. One thing that I'm, I'm talking to a lot of, investors about right now that they're seeing in their portfolio companies is they backed a lot of, founders who are really smart and technical but they can't sell.

And more and more technical founders are starting to emerge because of AI that can help you write code. I mean, heck, I wrote the MVP for my own app. I don't know how to do that. Right. But I was able to use ai, do it. 

Deep: It's, it's, it's nuts. Like we have, uh, these kind of reporter types like that write some content for our blog.

And I don't know, like two years ago, like I was having a heck of a hard time finding folks that could really, you know, write anything that meant anything about ai. Now we've got like this 28 step process and they passed things through this and that, and they asked this question and they work on this strategy and they interact with, with our data scientists and eventually out spits an article and I read it.

I'm like, man, for somebody who's like, doesn't know anything about ai, this is such a good article. [00:49:00] Yeah, 

Zachary: yeah. 

Deep: And the truth is that by the end of that process, they actually do understand their topic pretty well. So I think, you know, my hope for all this. is that, people just get better. 

Zachary: Yeah.

Yeah. 

Deep: Like I think that we all get smarter, we all get better, we all get more empathetic, we all get more better at communicating. We also just get smarter because, I mean, the analogy I use is like, hey, if you took a three-year-old and raised them, in a house full of, professors, they're gonna get smarter.

And that's not to say that only, professors are, are like that, but, I think whatever you're surrounded by, you get better at.

And all of us being surrounded and the bots aren't only excelling frankly at analytical and intellectual endeavors. like, my wife knows this, she's an artist, and she notices it a ton on the empathy side and the EQ side. Mm-hmm. Because they've spent a lot of energy.

to their credit, like OpenAI and Anthropic and stuff, they've spent a lot of energy understanding that people are lonely. You know, [00:50:00] people don't have somebody to talk to. 

Zachary: Yeah. 

Deep: these bots like help give them some confidence 

Zachary: it's naturally, or I guess, not naturally a people pleaser. And I think one of the things that you'll start to see, you know, you mentioned if a person grows up in a household with professors, they're gonna be naturally smarter.

Very soon we're gonna have a generation of children growing up in the house of professors everyone's going to, because chat GPT is around and they have access to technology and phones, and they're gonna be able to have that. So they are, in essence, going to be growing up around that. Well, if everyone's leveled up from an intelligence standpoint, what becomes the differentiator in how the cream rises to the top?

And I really do think it's the ability to connect with other people and the ability to communicate that 

Deep: have yet to meet somebody who disagrees with me on it. But if you do let me know. But the one job I know will not ever be taken by a bot is like a bartender. nobody wants a robot to give him a drink.

Zachary: I don't know. I think, alright, I'm gonna try to push back. I don't totally disagree, but I'm gonna try to push back here. have you been to a [00:51:00] bar where you can just get the pour, like you just put your glass under and it pours the, the one you want?

I 

Deep: have not, no. 

Zachary: They're great. I live here in Phoenix and uh, we have Waymo Waymo's. Fantastic. By the way. Do you know why? It's fantastic? Because I don't have to talk to somebody. I control the whole experience myself. I have my things I need to do. I don't feel rude.

There's no awkwardness that I need to feel Ah, so now this, this car can just take me around. I think the bartender's very similar. Like I don't have to talk to a bartender. It's not an awkward conversation.

It's just a bot , we do also get all those, waitresses or waiters that come by and, and you're in the middle of a really deep conversation with someone and they come by and then the tone that you're in is deep and intense and they come in and they've got this energy that's bubbly.

It's like me coming by. It's like a golden retriever saying, how's everybody doing today? And you're like, holy moly. I was on the verge of tears 'cause this person's story. But I guess I'm good. I'll have more, fries, please. Well, I appreciate 

Deep: that. I appreciate the pushback there. I think you're probably, you know, I actually have a, friend that's working on a.

physical [00:52:00] Android bot with all the fake skin and everything and, and really Oh, wow. And really focusing on the emotional, uh, facial rec gestures and all that. And it's, it's pretty wild stuff. And making like crazy progress. 

Zachary: but here's where I am actually worried about this though.

Remember we talked about EQ and Yeah. And, uh, and understanding emotions. People don't want friction. They don't want pushback. AI is not going to argue with you. You said people suffer from loneliness and, I think that if we develop too many technologies that just help people feel good, there's going to be an artificial, uh, 

Deep: I think this is, this is probably the most prescient few words in this conversation.

In 20 years, I would bet my house that all the problems we've seen from the cell phone, is gonna come from toddlers with their little AI stuffed animals that have never had anything pushed back on them. No other kid came up and smacked them and took a toy or bit 'em or whatever.

No adversity at all. No grit. 

Zachary: Yeah, 

Deep: I think it's gonna cause the whole [00:53:00] next generation of psychiatrists and psychologists issues. Like they're, this is what they're gonna deal with. But anyway, 

Awesome conversation. I've had a really good chat. thanks so much for coming on the show.

Zachary: Thanks, deep. I really appreciate it. Thank you so much.