Sponsored by: Glean | From Analytics to Action: Powering Databricks Insights

Summary

Glean connects to the Databricks Data and AI platform through AI/BI Genie, enabling all employees to ask data questions in plain English and receive answers grounded in Databricks governance and permissions, enriched with enterprise context from documents, Slack, email, and meetings.
Glean addresses three core barriers to data access — fragmentation across tools, expertise gaps preventing non-SQL users from querying data, and low data discoverability — through over 100 pre-built connectors and a unified natural-language interface over both structured and unstructured enterprise data.
Customer results include 15x ROI and over $300 million in attributed business impact, reflecting reduced analyst ticket volume, faster decision-making for business users, and broader data-driven decision adoption across organizations.

Chapters

FAQs

What three barriers prevent knowledge workers from accessing enterprise data effectively?

According to this video, the three barriers are fragmentation (the full answer is split across a data warehouse, BI tools, documents, and SaaS point tools), expertise gaps (not everyone knows SQL or how to find data), and discoverability (users do not know where the data lives, who owns it, or which dashboard or Genie space answers their question).

How does Glean integrate with the Databricks Data and AI platform?

Glean integrates with Databricks through several layers: a direct API integration with Genie for natural-language SQL queries over governed Databricks data, a Databricks Workspace connector that indexes notebooks and code, and a Glean MCP integration that enables agentic workflows. Each layer extends how Glean enriches answers with data from the Databricks Data and AI platform.

What makes Glean different from using Databricks Genie alone?

Genie is excellent at answering structured data questions governed in Databricks, but many business questions also require context from emails, Slack conversations, documents, and meeting notes. Glean combines these unstructured knowledge sources with Genie's structured data answers in a single query, so employees receive responses that reflect both the underlying data and the organizational context around it.

What customer results has Glean achieved?

According to this video, Glean customers have achieved 15x ROI and over $300 million in attributed business impact from deploying the platform. These outcomes reflect reduced analyst ticket volume, faster decision-making for business and GTM teams, and broader adoption of data-driven decision-making across organizations that previously lacked accessible self-service data tools.

Full transcript

[00:08] Hi everyone. I'm Kelly. I'm a product marketer at Glean. Um, and today we're going to talk a little bit about how you can leverage data bricks and Glean together for all of your data needs. Can I get a show of hands? I'm trying curious who's in the audience today. Are
[00:21] you on the data team? Can you raise your hand? Okay. Are you on like the business or GTM team? Can you raise your hand? Okay, cool. This is going to be a great session for all of those involved. Um so
[00:34] today's agenda I'm going to quickly walk through why access to data is really really hard especially for knowledge workers that are uh not essentially data people. How glean's context supercharge
[00:46] data analytics. We're going to talk through all the different integrations we have with data bricks and all the things that you can leverage us for. Um and then close out with some real world use cases. I'm happy to take some Q&A either up here or off to the side after
[00:59] the session is over. So, historically, access to data has always been extremely difficult and I'm sure all of you know that since you're all data folks uh sitting in this room.
[01:12] Um data is often fragmented. The full answer is usually split across a ton of different places. your data warehouse, your BI tools, your docs, your tickets, and all of the point tools in your
[01:24] enterprise SAS um that that the enterprise uses. The second is the expertise is a blocker. Not every single person in your organization knows how to use SQL or knows how to get answers to
[01:38] their business questions or leverage data to do so. And so that prevents a lot of our business users to actually be able to leverage data decision- making. The third one is that a lot of data are
[01:51] are often uh hidden. And so users don't actually know where the data lives, who owns it, or even which dashboard or genie space to use to really go and answer that question. And so there's
[02:04] really a a data gap here, a skill gap here, and also a discovery gap here. And what this results in is that there's a ton of data bottlenecks, right? I'm
[02:16] sure all of you know you get a ton of ad hoc tickets that ruins your sprint and now you have like 10 other things that's on your plate that you didn't know that you had a week ago. And for the business users and those of you that are in GTM,
[02:29] those results in blind spots and uh poor decision-m because you just don't have access to the data that you need. So what does Glean do? Glean consolidates basically unstructured data
[02:42] that is um often in your SAS tools like your Jira, your Slack uh with structured data that's often in a cloud data warehouse like data bricks. And so some
[02:54] of these things is we were able to actually merge all of these things uh for natural language analysis. And so what data bricks and glean really unlocks is trusted and actual datadriven
[03:07] decisions. So we talked a little bit about how data is fragmented everywhere. What Glean does is now that it allows users actually query consolidated data in a single interface and not only that
[03:20] we allow you to use natural language to do this and so you can talk to your data in real time without the SQL expertise that was often a blocker. And lastly you can discover data seamlessly. So you can
[03:32] find AI and BI dashboards uh or Genie spaces in seconds across all of our platform in Glean. So Glean search, Glean assistant where you can talk to it um as well as Glean agents if you're
[03:45] building it within our agentic platform. So I'm going to take a step back here and talk a little bit about what is Glean. Glean is a work AI platform which
[03:57] is really three things. It's an enterprise AI assistant which helps every single knowledge worker to reimagine work with AI. Um we are also a horizontal agent platform. So we enable
[04:09] AI transformation across every single department. Um and all of this is really built on an AI system of context. And what that means is we help you understand your company's data, your
[04:22] people and your work in order to give you more accurate and performant answers. Here's a quick overview and architecture of the clean platform. At the very very
[04:35] bottom, this is where we have over hundreds of pre-built connectors to all of your enterprise data systems. One of the things that you as data people know is data engineering is hard. Getting data into places is really, really hard.
[04:48] We make that really easy with these pre-built connectors so that we're allowing you to kind of uh consolidate that structured and unstructured data together. Once we do that, that's how we kind of unlock all of the great things with Glean. Glean search allows all of
[05:01] your knowledge workers to then search across all of these different places in a single place. Um, and then like I talked about earlier with Glean assistant where you're able to chat with all of your enterprise data, but now you
[05:15] can also write to all of those different enterprise tools on top of just reading from those enterprise tools. Um and obviously we have firstparty agents uh in all in Slack, Zenesk, etc. as well as
[05:27] thirdparty agents with Glean MCP. But what's the bread and butter of Glean is that we are a context layer for
[05:39] everywhere that you build and use AI. What this means is that not only would do we use the context that we know about your company and how you work and all of the tools within Glean, if you want to
[05:51] bring this context into other places that you want to build agents, you can leverage our Glean MCP for that. And so anywhere that you want to build apps or build AI projects, you can leverage Glean.
[06:17] capabilities. We have something called the enterprise knowledge graph. And essentially what this does is it crawls across all of your apps and it builds entity and permission graph across all of the people in your organization, all
[06:31] of the communication that happens within employees in your communic in in your organization and basically leverage those as signals and context in all of the answers and the way that you interact with AI. Um the second thing is
[06:44] permissions aware retrieval. every answer actually respects the source systems um uh ACL's your unity catalog whatever that system is we respect all of those permissions so
[06:57] let's say you have an agent to answer questions uh within data bricks if you don't have permissions to actually read that table then you will probably not be able to read that table and it will tell you you don't have access to it
[07:10] um we also have cross app context like I mentioned earlier we're combining structure data data bricks results with the unstructured signals that are often not in a cloud data warehouse um so that you can get an answer that has
[07:23] all of the context. So on the left there you see data bricks brings the metric and glean really brings the additional context around the metric to give you a business ready answer.
[07:42] Here's another graph of our enterprise context. And there's really two things I want to point out here. Um, is really our enterprise graph and our personal graph. And what our enterprise graph is, like I mentioned earlier, is really around understanding how your company
[07:55] works. And so this is the context that a lot of companies don't have is around what are the business processes, who's within each team, um, and who are the customers, what are the data around all
[08:07] of this. And on the right side, we have your personal graph. And so this is a graph about you as an individual user, your own goals, what are you trying to do every day at work. We also learn your
[08:19] writing style, your response style, and so if you ask lean to help you draft up a Slack message um around, you know, what you learned at Data Bricks conference, it's going to be able to help you do that in your own tone of voice. So these are really the core
[08:32] elements that really make us different from everyone else in the industry. So here's the fun part. These are the four different integrations that we have um with with data bricks. And because
[08:46] this is a very short section, I'm going to try to go through them. I'm happy to answer questions later on. The first one on the very right, data bricks genie and glean assistant. This is our newest integration. It is currently in public beta. And what this does is that it
[09:01] allows Glean as a front door for all of your knowledge workers to ask data questions in natural language. And so it will hit data bricks underneath the hood. You do not need to build anything for this. It just works out of the box
[09:14] once you're connected and you have all the right permissions. Um the second one is direct SQL in in clean assistant and clean agents. So not only can you use natural language and clean assistant uh
[09:26] to get answers to your to to your business questions, you can also write SQL and it will translate that all the way to data bricks. And not only is it in clean assistant, you can also do this in clean agents. So let's say it's not
[09:39] an ad hoc question, you actually want to build weekly reports or quarterly reports. You can build an agent that actually creates these analytical workflows for you and it will hit directly um within data bricks to get
[09:51] you those answers. The third one is u our data bricks connector in glean. And what this does is that it indexes data bricks AI and vi dashboards and genie spaces so that all
[10:04] of your knowledge workers are able to search for these different dashboards as needed. So like I've mentioned before searchability is a problem. You sometimes have built many dashboards where people are still slacking you
[10:16] asking do you have a dashboard for XYZ? they can now use Glean to do that instead of bothering you for
[10:30] your agent workflow if that's something that you want to do. The last path is um data bricks genie and genie code can call glean mcp. So depending on what
[10:42] your use case is, if you are typically within data bricks and you want to ask a quick question, you're able to just leverage Gleam's context directly. So you can still have the same accurate answer no matter which interface you're
[10:55] using. Um, Genie Code, we had a session earlier yesterday around this. And so if you're looking to build dashboards, uh, data pipelines, and you want to bring in Glean's context to make that more
[11:08] accurate, you're able to do that with Glean MCP. So these are the four different integration paths that we have today. Um we have customers using I think almost all of these and I'll talk
[11:20] a little bit more about um what those are later. So overall Glean delivers unified intelligence across your data teams and your business teams. oftent times a lot
[11:34] of your data teams are in data bras and they're writing SQL but your DTM teams your knowledge workers they are not in data bricks they don't have access and so what this does is it is really bringing in data you know governed
[11:47] structured data whatever you have in data bricks along with all of these other unstructured data sets like your gong calls your jera tickets that maybe you don't want to pipe into data bras and so we're able to bring those two
[12:00] together so that you have an accurate answer across multiple interfaces. The second thing that you really unlock is self-service analytics, right? So again, giving your uh knowledge workers
[12:12] different interfaces to choose from uh but allowing them to use natural language to ask questions. Um Gleam routes, you know, text to SQL through data brick stream to bring the answer back to Glean. The outcome is faster
[12:26] decisions, fewer analyst tickets, more time for you guys to work on things that actually matter. The last one is secure discoverability. So surfacing dashboards to so that everyone is is self-service. Um and all
[12:40] of this honors data bricks and uni catalog permissions so that people aren't seeing things that they're not supposed to see or able to query. And so you're able to reuse insights without weakening governance.
[12:58] And so I'm going to wrap with um two examples. We have a very very large manufacturing company that use both fle and data bricks. Um and they and their design and research engineers actually use glean to query data bricks genie
[13:12] directly. And this is because they're not data folks. They want to look at all of the data around you know building the things that need that they need to build but they might have PDFs of um the
[13:24] diagrams of of the things that they are building. And so being able to use Glean to query uh data bricks Genie and then merging that with things like SharePoint, email and Slack is really
[13:36] driving a ton of value for them uh with a projected ROI of 15x um and almost an over $300 million in annual savings. So that's massively huge and a lot a lot of
[13:49] uh use cases for folks who are not just on the data team but across the entire company. The other example I have is a real estate company. Um, and what they do is
[14:01] they use Glean and Data Bricks to
[14:14] support tickets with being data bricks. Um, and so they've used agents to do that. And now they have data bricks genie to really help summarize some of the most recent changes. um and also pull metrics from their data warehouse
[14:28] uh reducing the back and forth between teams. And so this is a very very quick overview of um how Glean and data bricks work together. I'm happy to take questions. I'm happy to chat afterwards,
[14:42] but thank you for joining the session today.

Learn more about the Databricks Data and AI platform.

The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.