Maven AGI Demo — Generative AI for Customer Support

Julia Nimchinski:
And welcome to the show, Lauren McGee, Strategic Sales Advisor at Maven AGI, and Rozzy Stein, Solution Engineer at Maven AGI as well. Super excited for the demo! How have you been?
Lauren McGee:
Good! How are y’all?
Stu Sjouwerman:
Dining off. Good luck!
Lauren McGee:
You will. You will. We are good, we’re ready to rock and roll. Should I go ahead and share my screen?
Julia Nimchinski:
Yep, let’s dive into it.
Lauren McGee:
Alright, perfect. Okay. Everyone see that?
Julia Nimchinski:
Cool.
Lauren McGee:
Perfect! Alright, well, nice to meet you all. As Julia mentioned, my name is Lauren McGee, I run our strategic sales team here at Maven AGI, and joining me is Rosie Risty, who is our Strategic Solutions Engineer. And while it is going to be the Lauren and Rosie show today, I did want to briefly highlight our founding team at Maven, as Maven is built by leaders who have scaled some of the largest customer experience and AI platforms in the world. Our CEO and founder, Jonathan Corbin, spent over 20 years in customer success and helped grow HubSpot from a few hundred employees to over a thousand, taking revenue from $700 million to over $2 billion. And then our founder and CTO, Sammy Shalabi, he is a four-time founder with over 55 patents who led Google News. and mastered the enterprise search for over a billion users. And then lastly, our chief product officer and founder, Eugene Mann, spent over a decade building personalization and applied AI at both Google and Stripe, where he created Stripe’s first applied machine learning program. So together, they have brought unmatched experience in scaling technology, teams, and outcomes. And with those outcomes, that’s exactly what we want to talk about today, is what is Maven AGI actually solving for? We understand the challenges facing customer experience today. It’s outdated, it’s expensive, it’s not built to meet the expectations of the modern users, and enterprises are drowning in labor costs, struggling to personalize at scale, and dealing with fragmented systems. And when that happens, it leaves your customers feeling frustrated with the slow, impersonal, and inconsistent experiences across those channels. And all of that is exactly why we built Maven. Our AI Agent platform is designed to address those problems head-on, scaling support without scaling headcount, and personalizing every interaction and connecting seamlessly across all systems. And let’s dig into how it works, and how exactly we’re solving those challenges. So, as I mentioned, Maven AGI is an enterprise AI platform. And what we do is we unify all of your systems and all of your data to allow you to deploy intelligent AI agents across any surface or channel to enhance both your customer and employee experiences. It is comprised of two components. First, you have AI agents that can operate across any channel, co-pilots, Slacks, Teams, Voice, wherever you need it, Maven can live. And then we have the core Maven platform, and that includes Agent Designer for business users to actually build and manage these AI agents. As well as an app studio for developers to kind of create deeper custom integrations and experiences if they prefer to build them themselves. And how it works is you want to think of Maven as an extension of your team. Anything that your humans look at today to be able to do their job, or serve your customers, is what you want to feed into Maven. And once we have ingested all of your different data and all of your processes, this allows us to create a single graph of record. And with this, what Maven has done is accomplished something all businesses have been trying to do. We allow you all to use one piece of technology that can then look into all of your different systems, create a single system experience to pull all the right answers with all the right context, to be able to drive the right outcomes for your users. And with that graph of record, we then create 3 indices. A knowledge index, a personalization index, and an action index. So this now means Maven can answer questions, it can answer questions contextually, and then it can even take action on behalf of your users or agents. And while our focus is on the support use case, ultimately at Maven, we are on a path to business AGI, where we are solving for all use cases across the customer journey. And how we manage this is through what we call our flywheel approach. First, we integrate, as I mentioned, across all those various systems of record. With that, that ensures our AI has access to the data it needs from tools like Salesforce, Sendesk, and any of the other platforms, creating that solid foundation. And then we deploy and observe. So once integrated, we quickly deploy these AI agents across multiple channels, anywhere that you need them, and then we observe how Maven’s conversations are performing across those channels. Highlighting areas of needs improvement, whether it’s missing knowledge or data access. And then we generate those improvements. So our platform spots any missing knowledge content gaps, we’re able to draft updates as your knowledge base stays current, and so the TLDR is we integrate with your tech stack, we observe performances across those channels, and then we generate fixes. And so what do the results say? Everyone’s asking the same question. Can generative AI agents actually do what humans can do, and can they fundamentally shift the way that support works? This slide showcases a wide variety of Maven’s customers in different industries, leveraging our Gen AI agents and proving that it works extremely well. You’ll see here that Maven is autonomously answering over 9 million support tickets with a 93% deflection rate, affording significant savings and scaling opportunities across the entire customer journey. And then last slide before we do the drum roll effect to show the quick demo, is I like to highlight this slide as we transition into the demo. Because Maven is built to support every type of customer that we know all folks have. Whether it’s knowledge seekers that prefer to self-serve at a help center and do research to find answers to their questions. Or maybe it’s the validator, who wants to have a conversational exchange, maybe over text, chat, voice. And then there’s ticketers. Those are people like my mother, who prefers that white glove service, she really wants to talk to the human and have the human help her. All of those are powered by Maven’s one brain, and we can meet them on any channel that they want to engage on. Now for the demo, what Rosie’s gonna do is we’re gonna simulate a customer journey of an end user that has purchased something through Lowe’s, the home improvement store, and we’ll be focused on the validator customer profile, as well as the ticketer, through the different channels of email, chat, SMS, voice, and then Copilot to show you how we can support it internally from here. So with that, Rosie, I’ll let you take it away.

Rosie Riste:
Yes, thank you. And so, like I said, or like Lauren said, I’m Rosie, and today I’m a super avid gardener. My hardware of choice is Lowe’s. I buy all my supplies from there, but today I have a small problem. I lost my order confirmation, and so I wanted to call in and check, what is the status of my order, because I want to know when I’m going to get it. I’m really excited about it. So the first thing I did was send an email to the Lowe’s support email address, because Because, it’s a… not a very, like, intense question right now, I just need to know. So, I sent in what’s my order status, and Maven recognized who I am based on my email address, and it looked up my account and my order information based off of that information. So, this is good, it looks like my order was placed, it’s already been shipped, it’s on its way. we’re in good shape. So, no more problems here. However, let’s say it’s the next day, and now I have a more significant problem. So, I need a faster response, so I’m gonna go right to the Lowe’s website, I’m gonna go to this very handy chat widget here, and you can see that Maven has already recognized who I am. And I’m gonna go ahead and say, oh my god, you sent me 600 tomato plants instead of 6. So that’s obviously quite a big issue. I don’t have room for 600 tomato plants, so Maven is gonna recognize this, and it’s gonna make a decision to offer me a phone call to provide an even deeper connection. And it’s being shy, because we’re on a live demo, and so everything works slower online. Let’s give it a second here. Maybe we’ll do one more shot here. Sometimes when I, have it hanging for a while, waiting for it to demo. It slows down. In these chat scenarios, this is all one brain. The email and the chat and, later on the voice, all of these use the same Maven brain. So it’s the same knowledge and actions that, across all of these systems, but they can be segmented as well by user, by surface, and so forth. Oh, come on, Maven. Can’t show me up today like this. I think we’ll… yeah, we’ll refresh that, and then if maybe you want to show the co-pilot experience, then we can come back to it. We’ll give it a minute. And so, we’ll come back around to this guy. But the, let’s go to the… move over to the ticketing system. So, in some scenarios, the customer, is having a problem that the business would prefer a human in the loop for. So, for instance, if someone is trying to do something like installing a circus… a circuit, in some scenarios, this is one where they would prefer this to go directly to a human agent. And so, we have the co-pilot here to help. This has… this is currently a ticket in Zendesk, it got here because we emailed Lowe’s, or because Maven, from a different surface, handed it off to a human. And so the co-pilot is here to help me, as a human agent, answering this question. So, I’ve got a little summary here telling me what’s going on, in the whole conversation. If there was a lot of back and forth here, then, this would give a good summary. And then it gave me a suggested response. Again, this is all coming from the same brain, it’s the same information, but in this scenario, some of this information is targeted only for internal human agents. Because we want to review it before we send this to the end user. This is a great response, so I can go ahead and insert it and send it away, but if I wanted to, I could have interacted with Maven here to tell it to change it, shorten it, add a piece of information, maybe there’s a discount I can offer them, that type of thing. Additionally, there’s a section here for research. This is a good place for your human agents to do research into internal policies or so forth that we don’t want to expose to the response side. And this is a way that you can get any information without ever having to leave your workflow. So I’m gonna give this one more shot here, and see how it goes. Because I really want to take care of these 600 tomato plants. Okay, there we go. So now it’s saying we’ll call you, because again, this is a somewhat serious situation here, so we want to give you a really deeper connection. So I’m going to go ahead and say, sure, go ahead and call me. This would be great. And I’m getting a phone call on my cell phone right now. Hello? Hello! Last time we spoke, we were discussing an over-delivery of tomato plants. You mentioned receiving 600 plants instead of the 6 you ordered. Are you available tomorrow for a pickup at 9am? Yeah, that’d be great. Excuse me while I change screens here. Your pickup is scheduled for 9 a.m. tomorrow at 123 Garden Lane, Cincinnati. Just leave the plants outside, and we’ll handle the rest. Is there anything else I can assist you with today? No, that’s great! I am super excited to plant a whole garden. That sounds wonderful! I’m sure it’ll… So, that was the voice agent talking to me, and you’ll notice that the voice agent also went ahead and sent me an SMS message. It grabbed the phone number that I called in from, and then it sent me a message dynamically based on what we were having a conversation about to make sure that I had this confirmation. This is Maven right here in the SMS. I could continue to have a conversation with it about anything that Regarding my own orders. general Maven, or General Lowe’s knowledge and so forth. Again, this is just one more surface that Maven can be exposed on, and you can see that the conversation is keeping context across all these different channels so that the user always gets what they need wherever they are.

Lauren McGee:
Wonderful. Any questions for Rosie or I?
Julia Nimchinski:
Thank you so much for the demo, Lauren and Rosie. Yes, we do have a couple of questions here. So, one of them… One second is… Yeah, I mean, you kind of addressed this, Lauren, in the beginning. Where are customers seeing the fastest ROI? And maybe you could just share a couple of customer stories that inspire you the most.
Lauren McGee:
Yeah, we’re seeing a lot. You know, folks are either looking to reduce cost or improve efficiencies, and we’re able to solve for that across our whole customer base. One of our recent go-lives is actually Paris Hilton’s new beauty brand called Paravy. Where, their chat is powered by Maven, and we’ve actually even really leaned into the personality of Paris. So the persona, the tone, the vibe when you engage with that chatbot is very much like her nomenclature. So for them, it was about improving the customer experience and being on brand. We’re also doing a lot of folks in the fintech space that, you know, have high security profile transactions that really need to be secure, and that Maven needs to be able to fetch that accurate data right away to be able to handle it. So it’s leaning into a lot of that personalization and action index with kind of our foundation flow, where when we plug into all of those integrations and data points. We can answer those questions contextually and be able to take action upon that.
Julia Nimchinski:
Yeah, it… Megan, you addressed the questions here. One of them is, related to reasoning-heavy support cases. how it… basically how it handles it, and then, about sensitive scenarios and… and accuracy. So if you can just double down on that one, that would be great.
Lauren McGee:
Yeah, for sure. So, first, to handle on sensitivity and accuracy as well, we have full security guardrails in place, too. And the secret sauce with Maven is actually our proprietary search algorithm. That when we ingest all of the different data, we chunk it and we look at different signals, like freshness and authoritativeness. And Maven is smart enough to know that a feature release from yesterday trumps a blog post from a month ago. So because of that, our proprietary search algorithm then only feeds the most accurate info into the LLM to generate a reply. And so because of that, we’ve actually been benchmarked against 31 other competitors in the space for having the highest quality tied to that, and that comes in with that special sauce that I mentioned our founder, Sammy Shalabi, did at Google News. So a lot of that goes into it. And that also tails into the security and guardrails that we have in place as well. We fully redact any PII whatsoever. There’s a whole audit list of all the things that you can redact out, so any of that information that’s coming through the channels, it’s actually starred out. We don’t process it, we don’t look at it, we don’t even want to risk Anything with that. So it’s truly clean data in and really clean data out.
Julia Nimchinski:
Really impressive tech. One second, and we’re transitioning to our next session. Before we do that, what can you share regarding the roadmap for 2026?
Lauren McGee:
Yeah, the roadmap for us is we’re really leaning into proactive use cases and really being predictive. So, support was our entry mark into the market, and we’re spanning that across all use cases across the customer journey. So how do we be proactive with customer experience and proactively flag for churn risks and things like that?
Julia Nimchinski:
Awesome. And what’s the best next step? the best next step. With our invoicing, yeah.
Lauren McGee:
yeah, it’s for you guys to check us out and book a demo, and we’d be more than happy to show you the magic of Maven.
Julia Nimchinski:
Awesome. Thank you so much again.
Lauren McGee:
Thank you! Thanks, everyone. Have a great one.