How Cohesity’s new generative AI Assistant, Gaia, unlocks enterprise data for instant insights

Keith Shaw: Hi, everybody. Welcome to DEMO, the show where companies come in and show us their latest products and platforms. Today, I'm joined by Greg Statton. He is the Vice President of AI Solutions at Cohesity. Welcome to the show, Greg.

Tell me a little bit about Cohesity, and then what you're going to show us today.

Greg Statton: So, Cohesity came into being by revolutionizing how enterprises think about data management, data protection, and data security.

We offer four key applications that run on top of our platform: backup—being able to back up all that data from enterprise applications; securing it through anomaly detection, threat hunting, and data classification; and then enabling on-demand access to that data, no matter where it resides globally.

We took a step back and said, "Hey, we want to enable our customers to re-leverage that data." It's often just sitting there as an insurance policy. Our goal is to drive operational efficiency.

So, I’m excited to talk about our Data Insights pillar—re-leveraging that data through a process called retrieval-augmented generation.

Keith: And again, you’ve got a generative AI tool—which I think is table stakes for any company that comes on DEMO now. So we’re going to see a little bit of that too, right? Greg: Absolutely. Keith: And who in the company typically uses this? Is it C-level execs?

IT leaders? Data analysts? Can everyone in the enterprise access it? Who's the primary user?

Greg: I love to say everyone on the planet should be able to interact with Cohesity—but that’s not quite the case. Primarily, we sell into the technology stack—typically the CTO’s office. It’s the core infrastructure operators—backup teams, security teams. So it’s really the CTO and CISO suite.

With these new capabilities, though, we're expanding into the CDO and CIO’s office as well, to help them gain additional value. Keith: And what's the main big problem you're trying to solve? They already have all this data—you're just trying to help them extract more value from it, right?

Why should anybody be watching this video?

Greg: Exactly—we’re trying to help them re-leverage the data they already have, to democratize access to information. The goal is to turn data into information, and then into knowledge.

Keith: I’m sure you’ve got a lot of customers—but for those who don’t have Cohesity, what are they doing today?

Greg: They're likely using legacy solutions to back up and protect their data—methods that aren't efficient or easily accessible.

Keith: So all of that data would just be sitting there and not being used? Greg: Exactly.

It would be backed up using legacy systems, often stored in arcane, proprietary formats that are hard to access or re-leverage.

Keith: Like a big stack of paper somewhere? Greg: Hopefully not! Keith: All right, let’s get into the demo. Show me some cool stuff. Greg: Awesome.

When you access Cohesity through the Cohesity Data Cloud, this is the main splash screen for Helios—our SaaS control plane that manages everything. We’ll dive into our Data Insights application and show what you can do.

This demo assumes the company is already a Cohesity customer and is backing up data. With our new Cohesity Gaia features, they can now create an on-demand index of data embeddings to drive conversations with their data.

Say an admin identifies SharePoint sites, OneDrives, and NAS shares that contain valuable knowledge—they can create a dataset and grant me access. In this case, we’re looking at over 10,000 PDFs historically archived. My boss gave me some very specific tasks—security-related questions for a board meeting.

But first, I want to explore what’s in this data. As we index the data, we not only create the embeddings index—we also do theme and topic analysis. I haven’t read all 10,000 documents; I’m sure you haven’t either. So this helps me understand the core themes.

I can click on a theme, and start a conversation with the data—exploratory data analysis on unstructured content.

Now, back to my boss’s request—he’s heading into a board meeting and wants a summary of unauthorized data breaches from 2018 and 2021. I just joined the company and don’t have the historical context; it’s all locked in people’s heads or archives.

So, I ingested the archives Cohesity had already backed up, indexed it, and asked: “Can you summarize the differences between the unauthorized data breaches in 2018 and 2021?” For this demo, I’m anonymizing company names because it's real data.

The system takes my text question, semantically compares it against 10,000 PDFs, extracts relevant snippets, packages them in a prompt, and sends them to a large language model. The LLM uses that context to generate an answer—with good detail on both events and general observations.

Because it’s using internal data, I also get resource links and citations—so I can download the source and add it to the board meeting materials. Keith: That’s great for explainability—so the AI isn’t just guessing. Greg: Exactly.

You get a much more robust and trustworthy result because you're using your internal data as the source of truth. This isn’t something you'd want to do with a public ChatGPT-type tool—this is proprietary data. Keith: Right. Greg: What we’re doing here sits between full model fine-tuning and simple querying.

Instead of training a model on all your data, we pull relevant content at the time of inference—like handing an artificial researcher a stack of topic-specific papers and asking a question.

Keith: How do you prevent people from accessing information they shouldn't see? For example, someone in sales doesn't need access to sensitive security files. Is there access control? Greg: Absolutely.

When we create datasets or indexes, admins can include or exclude specific folders or file shares. They can then set role-based access controls—so only designated individuals or groups can access specific data.

Keith: How long does it take to set up? Is this SaaS-based, and can it be set up in days, hours, weeks? Greg: Great question.

It depends on how much data you have and where it resides. Everything is delivered through our SaaS portal. Right now, it leverages full SaaS infrastructure, but we’re working on bringing it to on-prem customers as well.

Once the feature is enabled, you could be up and asking questions before lunch. The data ingestion and indexing are handled by Cohesity’s infrastructure—you just bring the questions.

Keith: Is the LLM behind this customized to company-specific data? Or could I ask it random things like directions to a McDonald's?

Greg: You could, but it’ll say, “I can’t find that information.” We’ve put specific guardrails in place. We built this modularly to be an enterprise-grade information retrieval platform. We’ve partnered with Microsoft, AWS, and Google to integrate their language models.

Right now, we’re using Cohesity’s Azure OpenAI subscription, but customers can bring their own LLMs—or train smaller internal models if needed.

Keith: Is there anything else you want to show? Greg: I think your viewers get the gist: ask a question, get an answer.

Keith: So how can customers try this out? Do they need to already be Cohesity customers? Greg: The answer is always yes—I used to be in sales! Current customers can reach out to their account reps to request a free trial of Cohesity Gaia.

If you're not a current customer, go to cohesity.com and sign up for free trials of our cloud or on-prem backup offerings. From there, we can easily enable Cohesity Gaia so you can start exploring.

Keith Shaw: Greg Statton from Cohesity—thanks for the demo. Greg: Thank you so much, Keith. Keith: That’s all the time we have for today’s show. Be sure to like the video, subscribe to the channel, and leave your thoughts in the comments.

Join us every week for new episodes of DEMO. I’m Keith Shaw. Thanks for watching.