Sami Shalabi, co-founder and Chief Technology Officer of Maven AGI, joins John Furrier of SiliconANGLE Media to delve into the transformative realm of artificial intelligence and its burgeoning influence on data center evolution. This discussion, filmed at theCUBE's NYSE Studio, is part of the ongoing AI factory series, examining how AI and its infrastructure reshape industries and accelerate technological advancements.
In this video, Shalabi shares expertise in building agentic platforms that not only advance customer experience but are set to redefine it entirely. Highlighting the significance of embedding AI into workflows, they discuss how Maven AGI enables seamless interaction across multiple modalities, illustrating a fundamental shift in automation and integration. TheCUBE analysts, including Furrier, explore these innovations and their broader implications for the tech ecosystem.
Key insights from the conversation emphasize the vital role of integration in harnessing AI's full potential, according to Shalabi. The discussion highlights the importance of domain-specific intelligence in AI agents and the growing influence of developer tools that redefine user and customer experience. With the use of voice as a new modality for AI interaction, significant advancements in AI adoption are expected.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: AI Factories - Data Centers of the Future. If you don’t think you received an email check your
spam folder.
Sign in to AI Factories - Data Centers of the Future.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Register for AI Factories - Data Centers of the Future
Please fill out the information below. You will receive an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for AI Factories - Data Centers of the Future.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: AI Factories - Data Centers of the Future. If you don’t think you received an email check your
spam folder.
Sign in to AI Factories - Data Centers of the Future.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Sign in to gain access to theCUBE + NYSE Wired: AI Factories - Data Centers of the Future
Please sign in with LinkedIn to continue to theCUBE + NYSE Wired: AI Factories - Data Centers of the Future. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Sami Shalabi, Maven AGI
In this theCUBE + NYSE Wired segment from “AI Factories – Data Centers of the Future,” Nebius co-founder and CBO Roman Chernin sits down with theCUBE’s John Furrier at the New York Stock Exchange to unpack how AI factories are reshaping enterprise infrastructure and the future of data centers. Chernin outlines Nebius’ two-track strategy: a multi-tenant cloud built for developer experience and managed services, and large-scale, mostly bare-metal deployments for hyperscalers and AI labs. He discusses the significance of Nebius’ Microsoft deal (described as “up to $20B” and set to become one of the largest single-site GB300 deployments) as both an engineering milestone and a way to feed scale and cash flow back into the core cloud business. The conversation explores why enterprises want “the baby of supercomputer in the cloud,” marrying cloud flexibility with supercomputing efficiency to minimize time-to-value without sacrificing performance.
Chernin details Nebius’ specialization in AI-centric workloads (large distributed training and inference at scale), a platform roadmap that moves beyond infrastructure into inference, fine-tuning and reinforcement learning as services, and a commitment to helping customers build on open-source models for control, cost and data leverage. He traces customer waves from foundational model builders to vertical AI companies and tech-forward enterprises, noting early traction with firms like Shopify and momentum in regulated sectors such as healthcare following Nebius’ compliance milestones. With roots in Yandex’s large-scale engineering culture and meaningful exposure to ClickHouse, Chernin also weighs in on the economics of AI-scale infrastructure (power and capacity as gating factors), hybrid orchestration and sovereignty, and why latency priorities vary by use case – from reasoning models to voice agents – as AI factories become the new unit of value in modern enterprise compute.
In this segment from the theCUBE + NYSE Wired: AI Factories event, Sami Shalabi, co-founder and chief technology officer of Maven AGI, joins John Furrier to discuss the transformative role of agentic AI in enterprise customer experience. Shalabi details how Maven AGI is moving beyond basic chatbots to deploy sophisticated agents capable of handling complex workflows across support, success and sales. The conversation highlights the shift from generic models to domain-specific intelligence, emphasizing that real enterprise value comes from agents that can proa...Read more
exploreKeep Exploring
What is the focus and purpose of the AI factory series and what does Maven AGI do?add
What challenges are faced by AI agents in integrating into existing customer experience systems?add
What has been the impact of engineering tools and AI agents on productivity in software development?add
What trends are emerging in the development tools and customer experience concerning user interaction with AI?add
>> Welcome back over to theCUBE. I'm John Furrier, your host. We are at our NYSE CUBE Studios at the East Coast. Of course, we've got our Palo Alto Studio connecting Wall Street, Silicon Valley and tech as AI and AI infrastructure continues to dominate the marketplace as it enables accelerated computing, accelerated value in software, open source, and ultimately the road to AGI. This is what the AI factory series is all about. Sami Shalabi is here, Co-Founder and CTO of Maven AGI out of Boston. Thanks for coming on today. I appreciate it.
Sami Shalabi
>> Yeah. Thanks for having me.
John Furrier
>> So we're covering the AI infrastructure. We love the speeds and feeds, the GB300s, neoclouds, on-prem is hot. Crown jewel data mixing in with models. You're seeing that from companies like Amazon. You're starting to see people realizing that it's the data. It's the compute scale. It's also the hybrid configurations. This is bringing in fast the agent value, which some have said, including us, it's been stalled. So welcome to the series. Really appreciate you.
Sami Shalabi
>> Well, thanks for having me.
John Furrier
>> First of all, talk about what you guys do first. I want to get into this. I think we think the unlock will be software and the models themselves. Talk about what you guys do.
Sami Shalabi
>> Yeah. So Maven AGI is an agentic platform that helps our customers build world-class customer experience agents. Everything from support to success. So agents that are across the entire customer journey that operate in every imaginable modality. You can chat with it, you can talk to it, you can SMS it, you can WhatsApp it. And we have a series of agent assist products that integrate into people's workflows.
John Furrier
>> You know, that's the hottest area right now. You see even the call center days have evolved. You've seen that as a low hanging use case. Search is great with AGI. You see value of the data on collaboration tools. You're starting to see things like Slack become very valuable. You see things like embed into Slack. You're starting to see a new level of integration. And a lot of these are software abstractions. Talk about your vision on how you see agents coming to the table.
Sami Shalabi
>> Yeah. So when we think about agents, they're actually going to be augmenting the entire workflow across kind of not just how people work, which is I think where we're seeing a lot of recent adoption, which is work productivity. But in our case, we're seeing it's actually going to fundamentally change customer experience at its core. And we're seeing this clearly with many of our customers. In many cases, we're seeing automation rates in the 90th percentile where customer interactions are coming in and really kind of it's helping users solve their issues proactively, meaning before they even happen. So creating surprise and delight moments, and also reactively, which is in the support use case. One of the key aspects of this is that for AI agents to be able to solve real use cases, real ROI, real scale, they have to integrate into the various systems that normally users use. On average in customer experience, most folks use six to nine different systems to respond to any form of customer inquiry. And a big part of the way we've been thinking about this is integrated directly into the user experience, but also solve the integration problems so that people are actually solving their real issues.
John Furrier
>> One of the things people used to struggle with on the enterprise side was all these siloed knowledge banks and knowledge databases. Talk about how that's solved today. Because one, you're talking about a few things, end to end, which is almost horizontal scale, but also different silos. Those are two dynamics that tend not to go well together, but with AI, they kind of jam well together.
Sami Shalabi
>> Yeah. So a couple of pieces around the integration is it's one of the, I think, biggest unlocks with large language models is its ability to integrate systems without actually writing code. We're able to transform data on the fly through reasoning capabilities so that we're able to integrate all these different systems across different workflows. So that's kind of like one piece that starts to unlock a whole series of non-deterministic use cases in a way where we're actually solving real customer interactions that connect various systems. For example, with some of our customers, folks like Roe, the financial services company. I mean, we're answering all sorts of questions related to not just how the product is used, but also intersecting it with the user's data so that it's truly solving their issues, connecting all these various system, whether it's the CRM, the knowledge base, telemetry, in ways that are actually getting to a place where the user experience is getting to answering these questions directly without having to kind of go through complex user experiences. That's, I think, the first piece. And then in order to achieve scale, really, I mean, one of the things we have seen with many of our customers is if you do not integrate this directly into people's existing workflows, you don't get much adoption.
John Furrier
>> Yeah. One of the things that came up on some of these AI factory conversations is a comment I'd love to get your reaction to. I heard a comment just the other day, generic tokens are a waste if they don't know your business. And that was really designed to talk about how token's sake aren't that valuable. Yeah. I mean, having a general model is great, like a jack of all trades, some say. It's like having a bunch of interns. When you actually get the domain intelligence into the agents, that becomes a really key forcing function. So that's one. And then two, with coding assistance, you got protocols like MCP, A2A, and then a boatload of frameworks from OpenAI, SDK, Strands on the AWS side they just announced last week. So a variety of different frameworks, which means that like a robust developer environment. So talk about those two dynamics and how the domain intelligence on the token side is combining with the protocol and framework optionality for enterprises.
Sami Shalabi
>> Yeah. So let's kind of unpack the pieces around the token and the integrations. In the case of the world of tokens, in our particular case, what we've actually gone down the path of is instead ... I mean, a lot of folks have been approaching the problem of using large language models in ways where you just kind of put everything into the context window and hope to God that it actually produces an answer. We've seen at scale that creates a lot of challenges because it starts to introduce hallucinations because you can create contradictions, et cetera. So in our view, in order to kind of achieve the scale and actually build AI that's grounded in the truths of the enterprise, you need to kind of narrow those context windows and the tokens that you're using because it actually is the source of hallucinations. And the next phase of this is actually all the integrations. And actually, there's been quite a bit of progress from a standards perspective to simplify a lot of these integrations. However, when we've analyzed the problem, I think we're early days on the integrations because there are more integration points for enterprise readiness that go beyond things like A2A, MCP, et cetera. Things like how do you integrate into people's data lakes? How do you personalize the interactions? The security models, especially with legacy systems are quite diverse. And when you're trying to deploy this stuff at scale, it's really kind of important that you actually embrace the complexity of the environment. And in our case, some approaches, some folks have been taking is actually copy all the data into the models and into the environments. And what we found actually the fastest way to get to value is to connect to the systems. And we call that the graph of record where we're truly connecting to all these disparate systems, whether it's knowledge, personalized, identifying the user, the ability to actually invoke changes and actions and everything from integrating into people's data lakes and contributing to their data lakes.
John Furrier
>> So you're saying it's better not to move the data, but to basically let it-
Sami Shalabi
>> To connect to it.
John Furrier
>> Connect to it. Let it stay where it is, get it embedded.
Sami Shalabi
>> Yeah, because that's actually where you get a lot. When you start to duplicate it starts to create all sorts of challenges. I think that's the first part. The other part is there are certain types of data that you do not want to duplicate. For example, if you're asking for your account balance, you do not want that copy. You want that from the system of record.
John Furrier
>> When you talk about the developer environment, do you see any trends there that are giving you confidence around agent building? Because you're seeing the coding assistance drive some really interesting change. One is it's easy to migrate now with the ability to clone environments. Number two, the developers themselves, once they get like a couple months into building agents, they discover new things that accelerates kind of kickup on value. Are you seeing similar-
Sami Shalabi
>> Yeah. So we're seeing ... I mean, engineering productivity has definitely accelerated. And in fact, within Maven, we're using code generative tools to drive and build many of our integrations. I mean, we're integrated into hundreds and hundreds of systems, enterprise systems, and it's been quite an accelerator to help with the development process. That's, I think, one aspect of it. It's actually simplifying. It's actually making your ... AI agents are making engineers early in their curb act, I mean, become 10x engineers fairly quickly. So productivity has been fairly significant. So that's, I think, one piece around how the developer tools have accelerated. But I think more interesting for me, because we're focused on bringing AI agents to customer experience, which is not the developer experiences. A lot of the development tools are actually pushing the envelope in terms of what user experience for AI looks like.
John Furrier
>> Yeah. And the customer experience-
Sami Shalabi
>> And a lot of learnings that are emerging from that, that are actually making it a more mainstream because the developer environment tends to be the early adopter. So a real example is that, I mean, many of my engineers are actually using voice as a new modality for interacting with their development environment. And I think we're going to start to see more voice come into customer experience, and we've been quite excited. We just launched our voice agents this year, and we're seeing tremendous success with them. And I think voice is going to be a new modality that is going to accelerate AI adoption more broadly.
John Furrier
>> The voice is the killer app on both fronts, not only the customer side, but also on the developer side. Great point. Speaking of customers, give some examples of how your customers are changing the game for their experience. What's their best practice? What's the psychology like? I mean, little wins go a long way. I mean, you don't have to boil over the ocean. You can just start knocking down wins because they're data rich in terms of the domain specific sets that they're dealing with, whether it's a certain application, customer support. Again, talk about the customers right now. Where are their heads at?
Sami Shalabi
>> So customers, and I think we can kind of unpack this. So we started in customer support because that is where the biggest pain is. It's one of the few functions that have historically never seen any material improvements as a result of previous gen making into automation. And now, AI is entering a place where it's solving, in our case, over 90% of customer interactions without humans. So that's, I think, one piece of it is that there are many customers are looking to scale their top line without having to add headcount. So that's, I think, one piece of it. The second is, especially the more sophisticated and more progressive customers are not thinking about this, "Oh, how can I deflect and reduce headcount? How can I use the savings to actually create a differentiated customer experience?" And this is where when we start ... I mean, we started in support because there is so much pain there and there's so much ... And that's I think the first part. The other one, it's one of the few functions that has a ton of data that you can clearly measure ROI.
John Furrier
>> Yeah. Yeah. And I mean, that's the key. One of the things also with the large scale AI factory conversations is that they're squeezing more throughput out of the tokens per watt. So one, there's an energy bounded problem. So you got the data rich, you got the demand on the compute ready to serve up, and it's the biggest pain point.
Sami Shalabi
>> Yes. And this is why it's been one of the areas where there's been tremendous product market fit. And most of the time is we start there, but the product continually expands across the go to-market function. So real example, ClickUp started up with an agent assist product inside of their ticketing system. They ended up moving into kind of Slack and then expanding. And then this became a tool that's been very popular with their sales team, but it started in a place that actually had measurable ROI and then it just started to spread. And many progressive leaders, as they've been thinking about this, it stops being about how do I bring AI to support? How do I bring AI to success? How do I bring it to pre-sales? Is that where we're advising them and to think about it quite differently is that the world of customer experience is going to pivot from these siloed functions, which are really a function of how the employee pool and how the organization was set up into a world where the customer experience is going to be a mix of proactive and reactive experiences where user problems are solved across the entire customer journey as one unified experience.
John Furrier
>> Well, it's great to have you hit that problem really hard with a great solution. Thank you for being part of our Wired community at theCUBE. Of course, we've got our Palo Alto Studio in New York Stock as well. Also in Marlborough, Massachusetts, you have come up and visit us there as well.
Sami Shalabi
>> Oh, I'd love to .
John Furrier
>> And see Dave Vellante and the team there. I really appreciate you. Put a plugin for the company. What are you guys working on? How big are you guys? Looking to hire? How's revenue? Put a plugin for what's going on.
Sami Shalabi
>> Yeah. So we're headquartered in Boston, Massachusetts. We've raised $78 million. We have over 70 customers and we're creating a tremendous amount of impact for all of our customers. We focus on large enterprise, start-in support, but really kind of a solution that's across the entire customer journey. And we are hiring, offer everything. Engineers, go-to-market, et cetera. We are in hyper growth.
John Furrier
>> Yeah. Well, tell them theCUBE sent you, get a 10% bonus right in the top. Only kidding. I'll say that's your offer. But Sami, thank you so much.
Sami Shalabi
>> Thank you for having me.
John Furrier
>> I love what you're doing. Again, this year's going to be the year of agents really bringing a lot of value to the enterprise, certainly low hanging use case, customer support, easy way to get some wins, and then extending out with the development teams and also the data hygiene all looking good. Congratulations and thanks for participating in our AI factory series.
Sami Shalabi
>> Thank you for having us and talk to you soon.
John Furrier
>> We'll see you soon. Okay. I'm John Furrier with theCUBE. We are here at the NYSE, part of our CUBE Wired community, is part of the NYSE Wired network. Thanks for watching.
John Furrier