SaaStr AI Day Live: Beyond the Hype: AI-First Rev Ops with Momentum's CEO

Hello everyone. Welcome to our next AI Day session. I know we've been going fast and furious, so thanks for joining us here today. I'm here with Santiago, also known as Santi. He's the CEO of Momentum. We use Momentum. It's a great tool. We were just talking about how it's a great use case for pushing things into Salesforce. So ask him many questions. This will be a fun one. He'll do a demo as part of it so you can see what he actually means, not just like slides. So drop your questions in the chat and we'll get to them at the end. But without further ado, Santi, thanks so much for joining us and I'll let you take it away.

Thank you, Amelia. Really appreciate it. It's always fun to be given a demo of Momentum at SAS our day knowing that SASA is also a customer. So that's extra cool. But first of all, hi everyone, I'm Santi. I'm the CEO and co founder of Momentum. We are building AI first systems that actually work. And when I say that, what do I mean? Well, you have heard the news many, many times over. This is to quite a bit of popularity, this idea that the majority of AI projects out there are failing today, right? There's the MIT study and it's kind of become popular to come and make a bold statement, say, hey, you're spending your. You're wasting your money trying and buying AI. And the finding we have after spending multiple years in the market selling and making successful projects happen with small and large companies alike, is that the real problem is data. Now let me tell you about data. I'm going to use a bit of an analogy. Bear with me here. One hour worth of conversation. One call that a sales rep or a customer success person has with your prospects or your customers generates about 8,000 words. That is, if you were to transcribe every single word said, that's about 8,000 words. Now on the very left, the note that they write down. Usually if you were to imagine in a physical way, that would be about 25 words. You know, some people don't capture anything. Some people are capturing 50 words. And that includes both notes and fields in the CRM. We looked at our own customers on this stat and we calculated what is the average amount of output from a rep after they hang up from that hour worth of call is about 25 words. If you were to print it in a page, that's about the first line in one page. Now let's extrapolate this. Let's Take it to a hundred reps. This is a mid sized company, right? A hundred reps working a whole year talking to customers. That's four hours a day of conversation per person. In those conversations there's so much value. We're talking product feedback, competitive intel. You're able to coach the rep by seeing what's going on. You're able to better execute the deal and learn how to forecast it. Or all of that is flowing on those 8,000 words in an hour. And a team of a hundred reps in one year are generating 1 billion words. That's about 2 million pages. So if you stack them on top of each other, that's as tall as the Washington Monument. That's 626ft tall. Now let's keep extrapolating because you know sales calls are not the only channel of communication with your customers. You're emailing them, you're sending Slack messages to them. Of course, Amelia and us, we're always on Slack. We chat every once in a while, but for the most part interactions happen just beyond zoom. So if we add phone calls, sms, what about support? What about what happens after you close that deal? Because the relationship between you and your customers doesn't end once you close the deal. You learn about the success, the risk of churn the opportunity from expansion by looking at what happens after the funnel. If you add all of that now, we end up with a stack of paper that is 2,000ft tall. That's the Freedom Tower right here in New York City. Now we're talking about 3 billion words per year being generated by just a team of 100 refs. And now I go back to the CRM. What ends up in the CRM? And of course you can see the right of my slide here. You can kind of guess it. The stack of paper that actually makes it into your system of record today is just 6ft tall. That's 100 people working a whole year having a lot of conversations to get you to only just a couple pages of paper. 0.3% of the data is captured. And of course, what does that lead to? It leads to our source of problems, first of all, revenue, predictability and forecasting. You don't really know what's going on. You only have 0.3% of the story. The bulk of the data is being thrown away. You hire more reps than you need because you don't understand your forecast and your pipeline. Or maybe you have a less. You hire less reps than you need because you're, you know, underestimated. How much you're going to make that year. You end up making poor product decisions because you don't know what the market wants. You don't even know where your competitors are at. Even though your customers and prospects are showing up into this calls or over email, they're telling you what they like about those competitors, what they like about you, what features work, what features don't work. You don't know it because you're getting 0.3% of that attribution and marketing. Basically, what I'm telling you here is the whole company could benefit from properly mining this data. And of course, agents and AI projects could benefit from this data. Now the question is, how do we get a hold of this data? What is the right approach to operationalize it, to gather it at scale? And that's where I think the current set of options at your disposal are missing the mark. Today, the vendors that are selling you call recording, they're selling you conversational intelligence. They're really selling you a beauty, a pretty library of video that gets stored and never watched. We tackle this problem as infrastructure. Here's momentum. All that we do is focus on the data. Our task is to refine it and execute value out of the data that your team is generating. Every day, we store every one of those pages of paper, every bit, every bite. In fact, not only are we storing transcripts, we're storing the video, we're storing the sounds. If you're making a phone call that is five minutes long, we're storing that audio. And we spend all of our time building infrastructure for you to get value out of that asset. Now, the use cases are many. Of course we can coach your team out of it. We can detect churn, we can of course write notes, we can prepare your team for future calls, we can prioritize your product. What's interesting about this data set is it's really powerful in the one to one basis. Call comes in, field output comes out. But it is really powerful in aggregate. What if we could aggregate one year worth of conversation, those 3 billion words? What if we could aggregate every bit of product feedback that happened in the last three months on a specific segment of your customer base? Then the data comes alive and it can help make decisions at the C level suite and not just the rep level, not just the frontline managers. Now, of course, we're proud of our logos. Actually, I don't think I have Sasser here. I guess I technically do because Sasser is in the top right of my slide. But you should be there as well.

You gotta add us.

Yeah, we got you here, I'm just now realizing. But we have many logos, and we're very proud of every one of our customers. They're relying on Momentum today as their infrastructure. Many of them have called recording technology in place. They have gong, they have cores, they have cleric copilot. And they decide to bring Momentum on top to activate that data, to get value out of that call library. Those videos that nobody's ever watching? Well, we watch them. Our LLMs are analyzing them every minute. Okay, I think we're ready for a live demo. How am I doing on timing, Amelia? Am I good?

Yeah, you're doing great.

Awesome. That's what I like to hear. Okay, let's keep rolling. Let me take you through a couple of examples, because the pattern of analysis and activation of data, as I showed earlier, kind of presents itself in many ways. So let's go into the first demo. Let's go into the CRM now. I'll start by saying Momentum is barely a product. We onboard full customers without ever having them login into Momentum because we want to deliver value. We want to extract and activate that data without them ever having to open a new tab in their browser that needs to run automatically. Because automation scales and automation really delivers every single time. There's ROI immediately. So it starts with going to an opportunity and finding out qualitative and quantitative data about that deal. You can read notes about that opportunity. Every interaction we had with that customer is fully written down after the call is over. You can see qualitative insights that get gathered throughout the conversations. This looks back not only at the last call, but looks back at every single call put together from that account and informs it and enriches it with every new interaction. And this is not just calls. This is emails. This is sms. This is slack messages. This is support tickets when they become an account. And speaking of accounts, let me go into what an account could look like. You could have a section in your CRM where you have full handoff notes. As I said earlier, we're not just grabbing one call and analyzing the transcript. We're aggregating months worth of interactions in every single channel to put together sophisticated analyses that give you handoff notes. And these handoff notes are fully customized using prompts. So you could do a handoff Note for a CS team that focuses on the first 90 days onboarding process. You could create a handoff node from NSCR to an ae, or you could create a handoff note to a professional services team that focuses exclusively on the ps, engagement on the project and commitments that we had for them. You could also do product analyses and roadmap commitments and that could flow to your product team. And I will say this is not just qualitative, this is also quantitative. It's very important that you're able to extract quantitative information out of this data because through quantity you can build automations and you can build aggregation, pick list, multi pick list, checkboxes, numbers, dates, all of these are values that momentum is automatically extracting by automating the field inputs and understanding the metadata in your CRM to push that automatically. This can lead some to some really powerful use cases. And I think Amelia were talking about this coaching. This is our actual real coaching dashboard. We're looking at our own reps and we have defined our own set of competencies. The approach of analyzing every interaction and extracting qualitative and quantitative data out of those interactions very much applies to coaching. So we built a first class experience around it because it's such a common use case that our customers run into. In this case we're comparing two reps to each other on a very specific competency like what is their product knowledge. But of course we can double click on a single rep and try to look at how they've been performing. And that's when we can double click into things like qualitative and quantitative insights about them. What are the most common questions they ask? What are the common questions that are our customers are asking from them? How have they been tracking on every single competency we care about or you know, what is the qualitative assessment or where they need to improve what they're doing? Well, all of this comes together in an interface on top of the data, but the data is at the very core. Now this doesn't just flow into Salesforce or maybe the Momentum user experience, it also flows into Slack. I think this is one of the favorite use cases for Saster. It's a very slack heavy shop and most of our customers are, we are as well. So we're able to build these micro workflows that are letting the data flow in real time to the right places. Do you care about product feedback? You create a channel for it. Do you care about risk of churn, you create a channel for it. Do your managers care about every single call that the team takes? You create notifications and these are threaded signals that are flowing into your team. They have the nuance of the conversation, they have links to the call so you can review them. All of it flowing automatically every Time. The call happens now. I wanted to use another example and I'm just about five minutes away from Don and I know these demos need to be real. I couldn't really show you my customers data up until this point, but I want to show you my data as a good example. This is the world of momentum. The world of momentum is all about conditions and prompts. The example I wanted to use was our decision to build Microsoft Teams. We were looking into the next horizontal expansion of our platform, deciding whether we should build HubSpot or we should build teams. Today momentum is a salesforce and Slack experience primarily. But when we decided to expand horizontally to expand our market, we wanted to understand where was the biggest bang for the buck. We wanted to know would we make more money if we built HubSpot support or would we make more money if we built Microsoft Teams support? So we ran a batch analysis of momentum over a year worth of our data. We wrote a prompt and this prompt is fed with every single interaction in our pipeline. It is also fed with CRM data that is part of the decision. Then we can decide to analyze subsets of the data. In this case we chose emails and calls alone. We selected the type of opportunities we wanted to look into like every single stage in the funnel. Actually we didn't look at building pipeline because that was too early in the funnel. But everything past that we cared about those conversations. And then we decided to analyze pretty much everything that happened in 2025. That's seven months worth of interactions now that extracted data backward. This allows you to get insights from fields and questions you never asked in the past. It's like reconstructing history in just a couple of minutes. Now what we ended up with was a deep analysis of our pipeline, trying to understand really what is the delta, what is the payoff on building one versus the other. And of course when we look at the results, we ended up with very interesting findings. If we built HubSpot we would have more of the past pipeline that we would have been able to engage. 135 deals versus Microsoft Teams where we had 125 deals. Now what's interesting about it is There was about $4 million worth of pipeline behind Microsoft Teams, but only $2 million worth of pipeline on HubSpot. What we ended up realizing is sadly the kind of bottom of the market, the slower tier segment ends up being the kind of half sword pole and the Microsoft Teams tend to be more of the enterprise play as we are currently strategizing to move up market rather than down market. We thought we would get a bigger bang for the buck, less deals, more revenue if we build Microsoft Teams. And this is of course an analysis that we ran a couple of quarters ago because by now Momentum has full support of Microsoft Teams and we have a couple of enterprise customers who have already become users of us thanks to this investment. But this is a. I thought it was a good example because you can see the power of data oriented decision making. The ability to store every one of those vectors of communication and run quick and effective objective analysis of the data to make real important decisions. C level decisions inside the company. Not just day to day coaching, data hygiene, operational excellence, which of course is also very much useful and very much powerful. Now I'll go back to my deck. It's really about in the end, automation and data agents are here and data is the new bottleneck. Nobody is missing a piece of tech to build agents. Automations are all over the place. Of course, OpenAI now launched their own agent framework a couple of days ago. We have different technology that we use to build our own agents, but nobody needs more workados or zapiers or openais or N8Ns. What people need is the data to feed to those agents. Great automations with poor context into really bad decisions and very poor execution. So that's my pitch. Go get data for yourself. Don't just buy another agent framework.

Awesome. You've got some questions. There's a VC asking for fundraising. I hope you want to answer that one. And then I've got some more related to the deck.

If I got a. If I got a customer for every time I get one of those emails, I wouldn't need to fundraise. I think.

That'S fun. Okay, so we've got a few questions here. Okay, I'll start with the first one from the live stream. Back to your earlier point. I think this was earlier in the presentation you mentioned about 70 to 85% of all AI projects fail due to bad data. How does momentum ensure data quality when processing? I think you said like 500 million words or something like that. What validation mechanisms do you have in place?

So we have a very complex pipeline of analysis. This is where I think we kind of pull apart from your average diy. I think your average DIY would be something like gong. And then you mount clay on top of it and you run some prompts and then you put that data somewhere. It could be a field or it could be some notion or airtable. For this data to be trustworthy and aggregated at scale first of all it needs to be analyzed it as a batch. You don't want to grab one call and try to extract an insight from that call alone. Every single call, every conversation your team is having with your customers is a sequence of past interactions. And the last transcript counts on a lot of assumptions and establish context that comes from that path. So what we do is we actually grab everything that's relevant from every single interaction, every email, every sms, every call and run the analysis on that. And then by the way, you know, we've been doing this for three years so we found so many patterns that you want to fight against things like hallucinations or non answers that your LLMs will give you. So we have a multi tier validation layer. So we're having things like GPT 4.1 validating answers that GPT 5 is giving us or maybe a Gemini 2.5 on another layer. So we're discarding things like non answers or redundant answers or hallucination by just spending more money on tokens for the for the data to be higher quality. And then by the way, a lot of these stuff, we do it at the intake time, it would become impossible to grab one year worth of interaction then try to run a backwards analysis on every single one of those words. We're actually cleaning up and refining that data at the time of intake and we store it cleaned up. So when you come into momentum and say what have been the three most common objections that I've seen in my business in the last 12 months? We don't go through all 8,000 words, we only go through the data set of objections that we gathered in our data set. So we are basically a data store that purpose built for this use case.

Yep. And then another great question, and I think I had this question to your team when we were first getting set up was when you write back this data to CRM. So you were showing Salesforce in the live demo, how do you handle any conflicts with existing maybe potentially incorrect human data that's already in the CRM?

So it depends on the customer and in the best of the use case, we have customers that are very bold and they go wild and they just immediately turn it on automatically. Momentum is highly configurable so we allow you to do things like select a confirm to write. I think in this case I was showing the demo, we have a drop down and I think Amelia, you ended up in cases like this we can put full automation.

We do automatic.

Yeah, we do automatic what we end up finding and we always advise our customers is to go full automatic. For some of our customers, we've created dedicated renders of their CRM where they have a tab that is owned by Momentum. This, for example, is a tab that would be owned by Momentum. And all the data here is fully automatically captured. We can do fancy things like we could have a field that has a appended version of who is the economic buyer. So you would have the date and the name of the date and the date. So we're appending on top. And they would have one field that consolidates an intelligent answer by looking at that historical. So for example, many of our customers would have four or five different fields for every letter of medicine. They have the consolidated answer, they have the historical. They have a numeric assessment of how good of an answer they think it is looking at both field values together. So we can be pretty flexible in this regard and our customers. And depending on how bold and wild they want to be and how sophisticated the data set they want to collect, they can build it on top of the Momentum framework.

Awesome. Next question coming in. And this is a common question we've been getting today, so we'll probably start to incorporate this into all of our future presentations. But can you speak a bit to the implementation of Momentum? I can also share our how our implementation went just from being a customer now. But for your average customer, can you speak to more of like the implementation? What do they need to prep in terms of data and also in terms of CRM?

Yeah, I think you have. You need to have control of your CRM. You need to have a decent structure of opportunities and accounts. If you don't really are, if you're not really creating opportunities in your CRM or you have triplicate of opportunities, you will have a problem of mapping, which I think is something we want to tackle before we come in. And by the way, I will say our algorithm for mapping properly calls to your CRM entities is second to none. It is a hard problem in the world of gong. Didn't matter if mapping wasn't that great because all you did with that mapping was to find a call in your call library and play it. But in the world of Momentum, when we're writing data into the CRM, we really needed to hone in on exactly deciding and giving you some configuration and control over what do we prioritize. Do we go after renewal opportunities or the account object? Do we go after net new opportunities for renewals? So all of that is taken care of. Now, the process of onboarding is usually about 90 days we have a process with our own CSMs that show up. We have our, you know, our immediate playbook. We usually set up about 10 field extractions. We have the prompts already in place. But I will tell you, every customer will grab our standard prompts for things like product feedback or medic or spiced or band and they will end up making it their own. We give you something that is 90% good and then one week later, after you see the data flow and you tweak it to make it a little yours. Every company has their own way to sell. Then we go into signals, then we go into coaching. By the 90 days we've deployed the Momentum platform from beginning to end pretty healthily, and then you're kind of on your way. We do have a high touch relationship. We check in with our customers anywhere between every two weeks to once a month. So it is a bit of a top down, high touch product. This is not a PLG you would just come and set up on your own. It does require a bit of a lift, but I will say the value of the data ultimately pays off.

Yep. I would say too, for our, from our own implementation of Momentum. First I was like, oh, we're just getting VIP service because they know Saster and they're super fans. And your team was like, no, no, we do this for everyone. I was like, okay, cool. So yes, your team was very hands on in, like what we needed to do to get started, which I really appreciated because, you know, we have at Sester, we have multiple agents, multiple AI tools, whether they're agents or not, like running. And so that was really helpful for us to have all the stuff laid out. I think in the beginning we did like a once or twice a week call and now we do like every couple of weeks. Right. Like, so it does taper down, but it's not something to your point you can set and forget. It's not a PLG thing. I agree. Like, you do have to have the right buy in and also the right person who can set this up. Right. Which I know there's a related question about. Next. But yeah, just from our own implementation, it is something that you guys helped us with a lot and things we saw from it were probably unsurprising. Right. Like we were like, okay, we actually don't have as much data in Salesforce as we should. Right. Like there is, there is such this lapse of data we've had from it being hit or miss. Like I would go into certain accounts that we were starting to look at. Okay, if we're going to, you know, roll in Momentum, how's that going to affect all the accounts and the owners and all that, all the data, all the things you've talked about. And so I was very hit or miss. I would say that some sales reps had really good notes, really good previous calls. I like different recordings or just summaries or notes. They put maybe some contacts attached, maybe some closed loss opportunities, but most of them were barren. And so that was. I mean obviously it's terrible but it was a big eye opener for us when we went to roll this out that, you know, nobody stated as perfect. And I think everyone can say that now coming into implementing some of these AI tools and it's not that it needs to be perfect. I just honestly thought we would have more. And so that was big for us. Of like, okay, we actually don't have like everything you're showing here of like it pushes all this directly to. For us we use Salesforce. It pushes to Salesforce. It pushes directly into that account. It pushes the, the notes, the recording, the follow ups automatically, you know, it'll. Ours will automatically attach the contact and then we've worked with your team to automatically move certain stages. So if it's a early stage and it's sitting in like you know, meeting not held yet and then momentum knows we have the meeting, it'll move it into meeting health, which is quite nice because we're pretty bad about moving that because we just not. We're. We're just not. We're just busy. Right. Sometimes we don't always move the stages properly. And so again, just eye opening from that perspective of I didn't have as much data as I thought I should in all my accounts, which sucked. But now at least, you know, going forward I'm like, okay, we have peace of mind that this is all in there. I know. You know, I can. I. Well, I use the slack a lot. I use the slack more than I would say actually the Salesforce data, the Salesforce data school because our other agents can use it but I use the slack in real time. Right. I have a call. David has a call. Someone else on our team has a call. I can see and like we'll go into the slack during our one on ones and be like, okay, hey, let's look at the notes because I'll see exactly like what they brought up of hey, I had you know, say, oh, somebody asked me for like additional tickets or they brought up this or that and I'll just read the momentum summary to see what they were talking about. And then I know, at least it's in Salesforce, right? I don't have to worry about like, okay, David had this really good call. There are some next steps we have to do. And, and usually maybe he'd put like one or two of those notes because he's better at it than most. But now I'm like, all those notes are there, so we might not have like.

Yeah, I mean, we're talking to Salesforce about this. Like them as a company. Like, we have our own agent force agents that are using this Salesforce data. I think you're right. Ultimately, you end up with the team that engages a lot more heavily through Slack because that is where they're spending their time. They love when they go and they check something in Salesforce and it is there. But then, so the first 90 days is about building that layer of extraction. And then I would say it is a long journey from that moment on because then you start realizing the myriad of workflows you can build on top of the data now that you have it. So we have our own agent that will reconnect with our closed loss deals nine months later. And it will use the learnings of that past conversation, why we lost it, to write a sequence engaging them on what we learned. Hey, we spoke nine months ago. We know you went with X and Y vendor. We know you had this pain point. Did that pain point get addressed? And, and that is. That is, you know, a derivative workflow. But it would never work if we didn't capture the data in the first place at the very beginning of the loss.

Yep. Yeah. And that's such a. I wanna, it's kind of unrelated, but I do wanna ask you actually, because we, I know we're, we're at time, but I just wanna ask you real quick. You mentioned you have a Salesforce agent live. How are you. We just got ours live this week. How are you using yours?

So mine, for example, is writing sequences and engaging with our AA team doing all of these research into the data, momentous placing in the fields to put together that re engagement sequence for us to reach out. Automation reads a bunch of Salesforce fields. It uses your own playbook for re engagement and then it crafts a sequence for the right contact in that opportunity that we lost. Actually every single contact in the opportunity in case people left a year later and kind of lets our reps know that's the next engagement to try to pull it back. And we've actually re engaged, lost deals and we Actually won one that came from that campaign. So we had an agent pay for itself with a close one deal just a month and a half ago.

Okay, I'm not using ours for that, but that's a good use case. Yeah.

As it turns out, warm outreach sequences are really good if your warm outreach is based on actual data you gathered at the time. Right. You never want to say, hey, we spoke to you nine months ago. You want to say, hey, nine months ago you told me this problem. Has that problem been addressed? Because we are better than ever at doing that now. Yep.

Okay, last question because you just mentioned your. I didn't. I didn't know you guys were on Agent Birth. That's cool. That's a good use case. Are there any other like AI tools that. I mean you are an AI tool, but are there any other ones you're using for sales that you like?

I mean we use clay. We love simplifying the enrich enrichment in our CRM. I think of momentum as a first party data. Right. We extract data from your own conversations. First party. But all third party today where you're doing it with clay is cheap. It's pretty easy. We don't have to, you know, deal with any stuffy data vendors to do all of that. I really like ample market for the cold average side of the aisle. Our team is using them for prospecting and sequencing. I don't. We're just a customer and not much more than that. So I got no allegiance other than the product works really well for our team of, you know, today we have almost 40 SDRs in the organization. So it's a pretty high scale on top of it.

Yeah, yeah, yeah, that's. Thanks for indulging me in this one with that. I know you guys, your team. Ashley, your co founder will be at Sass Randall in London and then Jonathan through your. Your LinkedIn in the chat. But any. I know you guys are doing a bunch of stuff at Dreamforce, right. So how can folks see you next week?

Yeah, we have a space outside Dreamforce. Just hit me up on email. Santiagomentum IO and I would love to meet you, buy you coffee. You have a quiet space outside of the craziness of Dreamforce. We'll be spending all week there. We're in San Francisco for the rest of the month and then coming back to New York City. So always happy to grab a coffee with somebody who's curious about this approach, even if it's a VC and you know, want to talk about a fundraiser, who knows.

Okay, well, thanks so much, Santi. And I'll see you next week, and then hopefully we'll see some of you guys there.

Thank you, Amelia. See you then.

Yeah, of course. Thanks, everybody. Bye.