Sponsored: Reltio | From Fragmented Records to Trusted Context: Unified 360 Profiles for Databricks

Summary

Reltio's context intelligent platform unifies fragmented customer, product, and supplier data from multiple source systems into real-time trusted 360 profiles that AI agents and downstream applications can rely on for accurate decisions.
Reltio integrates with the Databricks Data and AI platform through Delta Sharing, delivering live governed access to unified entities, relationships, interactions, and calculated attributes directly in the lakehouse without moving data.
Reltio Agent Flow introduces an agentic platform for data quality tasks such as address enrichment with confidence scoring and marketing segment creation, enabling business teams to focus on generating insights instead of manually maintaining data.

Chapters

FAQs

Why do AI agents fail without unified customer context?

When customer, product, and supplier data is fragmented across multiple systems, AI agents receive incomplete or conflicting information and cannot produce reliable answers. Reltio addresses this by unifying and cleansing data from transactional, operational, legacy, and unstructured sources into a single trusted context layer that agents can query with confidence.

How does Reltio integrate with Databricks?

Reltio integrates with the Databricks Data and AI platform through Delta Sharing, enabling bidirectional data exchange that gives teams live governed access to unified 360 profiles — including entities, relationships, and calculated attributes — directly within Databricks without copying or duplicating data.

What is Reltio Agent Flow?

Reltio Agent Flow is an agentic platform for data quality that uses AI agents to automate tasks such as address enrichment with confidence scoring and marketing segment creation. It is designed to let business teams generate outcomes without spending time on the manual data cleaning and preparation that previously required dedicated data steward resources.

How does Reltio handle duplicate record resolution?

Reltio uses AI-powered identity resolution to automatically match and consolidate duplicate records across source systems, replacing the slow manual process of data stewards reviewing matches one by one. This automation removes the expensive bottleneck that prevents organizations from keeping data quality in step with the speed demanded by AI workloads.

Full transcript

[00:08] Hi everyone. Thanks for joining us today. My name is Shankar Gupta and I'm part of the product management team at Reltio. So throughout the keynote, we have been hearing about the importance of the context in the data and AI world. And we know that your AI solutions are
[00:20] only as good as the context they have access to. But at the same time and the problem is clear. If your customer, product, supplier, and the organization data is fragmented across multiple systems, your
[00:33] AI solution and the AI agents are not going to give you the solution which you are the answers that you are looking for or expecting for. And that's where Reltio comes in. Uh Reltio helps organizations uh
[00:47] unify this data across your systems and transform it into the real-time context and the trusted context that your AI agents and your AI applications can rely on.
[01:00] So what are the challenges that organizations are facing today? What we observe is that many, many organizations today are investing a lot of their time and money in managing their data complexity.
[01:13] The data stewards in the organization are investing their time in reviewing and the in reviewing the match and the duplicate records manually. Similarly, uh the analyst in the system
[01:27] are spending their time in cleaning and preparing the data instead of generating insights. And if you need to change any kind of data model in the system, it might take you months. Now these are the kind of
[01:39] problems which are very expensive and slow and they are not fundamentally designed to cater to the AI world which is changing at a very, very rapid speed.
[01:54] So what is Reltio's goal? Our goal is simple. We want to unlock the 10x efficiencies and enable the organizations to be 10x efficient in managing and
[02:06] leveraging the data. We make use of AI to enable the business teams to focus more on generating the business outcomes and not worry about maintaining the data.
[02:18] So, we all know that Databricks enable the the foundation for data and AI. But, to get the context on top of it, that's where Reltio helps you out.
[02:30] So, what Reltio does is it enables it provides you the context intelligent platform where we take the data from multitude of sources. Like, you can have your data in
[02:43] the transactional systems, you can have your data in your operational application, or it could be there in your legacy systems or the unstructured data. Reltio unifies this data, it cleanses it, it's enriches it,
[02:56] and it makes it available to your downstream applications. It could be your automated workflows, it could be your analytics application, it could be your operational application and the AI agents. But, the key point to notice here is
[03:09] it's not just the batch operations or the batch analytical process, it's a real-time operational applications which businesses can use to do any kind of operational workloads on top of it.
[03:22] Yeah? Now, uh as more and more AI agents are introduced into the business and they are um taking more and more decisions by themselves, it becomes more and more
[03:34] important for the context to be there, so that all the decisions which agents are taking for you, they are more and more confident to do so.
[03:47] And this value is generalized is realized today across the industries where Reltio is present, right from pharmaceutical to health care to consumer goods uh to retail and the financial services and many more.
[03:59] So, what these customers what we are looking at here, these are the Reltio customers. What they have in common is they are able to reduce the manual efforts, increase or improve their data quality, increase the efficiency of making this
[04:12] trusted context available to their teams, and also generate the measurable business insights. Plus, on top of it, they are not able to just transform their analytics, but also
[04:24] cater to their operational business operations in near real time. Now, how does uh Reltio adds a context layer on top of the foundation of the data in AI which is provided to you by
[04:36] Databricks? So, Reltio offers a bidirectional data sharing solution, which is a approved built-on solution with Databricks. So, here imagine that you have a data which is there in your Databricks uh
[04:49] environment. Without any data movement, Reltio can ensure that, hey, it adds a context intelligence on top of it, and using the inbound data sharing, that data is available to you in Reltio. It
[05:01] basically takes you through the entire processes of enriching, cleaning, unifying, and so on. And then, using the outbound data sharing, the same data is available to you in your Databricks environment. So, no movement at all. Your data stays where it is. Plus, you
[05:15] get to get the best of the context intelligence on top of your data. So, uh combining the two these two worlds together, what do you get as the you get to transform your agents not
[05:27] just from the agents, but to the super agents who have the access to the right context at the right time. Let's see it in action. So, the first demo which I'm going to show you is is the bidirectional data
[05:39] share which I was talking about. And here, what we will be looking at that you have some data in Databricks. How with a simple click, with one-click operation, you can make it available in Reltio, take it through the unification, and then have it back in Databricks.
[06:01] So, first, what I'll do is I will log in into Reltio and we'll be looking at the individual records here. So, the very first thing you will see is that there are no records, like individual records are not present in Reltio at present, but I switch to
[06:13] Databricks and I can see that you have the individual table where there are certain records present. Now, I'm going to set up the inbound data share from your Databricks account to the Reltio tenant. It's a simple one-click operation. You don't have to
[06:25] worry about the entire integration. Reltio will take care of all the integration between the two systems. Just when the inbound data share is set up, all the data which was there in Databricks is made available into your
[06:37] Reltio tenant, which you can see here. Now, all I'm doing is I'm just filtering the records which have potential matches. I go to one of the records and I can see that it has two records. Similarly, I can go to Databricks and do
[06:49] the query and I can see that yes, there are two records with the same name. So, I come back to Reltio and I can manually merge them. If needed, I can configure the system to do the automated merge as well. And now, this data, which is contextual
[07:02] rich now, I can just set up the outbound data share now to make this data available back in my Databricks account. So, now after the data outbound data share is set up, I can go back to Databricks and I can see my data. So,
[07:14] this time, if I do the same query which I did earlier, I see only one record, which were two records earlier. Now, this makes my agents, you know, way more confident in taking any decision, which was which earlier was not the case
[07:27] because of the data duplicacy. So, moving on, let's look at the other solutions. So, Reltio not only provides you with the context intelligent platform which I spoke of, it also provides you with the
[07:40] agentic platform, which we call it as the Reltio Agent Flow. Now, this is this solution provides you with a pre-built set of agents, which are shipped by Reltio, plus you can also create your custom agents using the
[07:53] Reltio's MC MCP server. And agent uh the uh the Agent Flow platform enables organizations to achieve the 10x efficiency, which we spoke of at the beginning. So, what I'm going to do is uh today in
[08:05] the session, we are going to focus on two such agents. We'll look at the demo for them. I'll start with the address um enricher agent. So, think of the solution like, for example, just like we looked at the individual records in
[08:18] Databricks, think of it like you have the address data as well of those records. But, for some reason, your address data is not complete, which is affecting your communication and shipment to your customers. So, how you
[08:30] can make use of Reltio's inbound data sharing, which we spoke of earlier, to solve that problem. And plus, top of it, you can make use of Agent Flow to make it more and more efficient.
[08:43] So, here I'm logged in into the home page of the Reltio Agent Flow. Here, I can see all the agents which are available to me and I can make use of. So, the first thing which I'll do is I will launch the address enricher.
[08:55] And here, I'm going to just simply provide a text prompt that, "Hey, go and look for my top 100 customers and tell me all those organization where basically my address is not verified." So, it goes back, it looks for uh the
[09:08] data, and it comes back with the top eight organization, where I can see the status as partially verified. Now, it gives me the reason that, "Hey, why does it think so?" And then the agent tells me the next steps as well.
[09:21] So, I just respond that, "Yes, let's move ahead and process all of this organizations one by one." Now, we can check it in Reltio as well that once I open one of the organization which was suggested to me by the agent, I can go it is I can check its address.
[09:35] When I double-click on its location, what will I get to see is I'll get to see the original source of this data. So here I can see that this address came from two sources. One was Oracle and another source was the Reltio data
[09:48] cleanser. Now if I go back again to my agent in the agent flow, what we are going to do is we are going to take the next steps as suggested by the agent itself. So it is telling me that hey, it is going to look for these
[10:01] credible sources to really enrich and augment my address information which is missing today. And it is suggesting me what are the option which I can take. So there are I think A, B, and C option
[10:13] which are provided to me. So I'll choose an option B that yeah, let's go ahead with the option B, please, and enrich my data of the eight organization which you told me. So it takes the required steps and it shows me the summary of the sources
[10:26] which it is taking care of or relying on to enrich the information. And plus gives me the summary and the exact steps which is which it is taking. So once it has enriched the address of
[10:39] one of the organization, it gives me the entire summary that these are the steps I have completed and this is the final how the address of the location of that particular organization looks like. And then I can scroll down further to look
[10:51] at all the other organizations also because if you just take a step back, you would recollect that it initially told me that out of the top 100 customers, there were around eight organizations where the address was not completely verified.
[11:04] So I can just scroll and I can look at all the information if I need to. Now once I'm convinced convinced that yes, this information all looks good, I can go back to Reltio and again re-verify
[11:16] the location address which I checked earlier. So this time when I go to that address, I will not see not only the two sources which I saw earlier. This time I'm seeing at three sources. One is Oracle and the Reltio data cleanser, but we see
[11:30] the address enricher as well, which is nothing but the agent through which we have been able to enrich this address. And on the left, I can see all the address attributes which have been enriched by the agent. Yeah? So, this is
[11:43] how um you can make use of the best of the Relatio platform, on top of it the Relatio agent platform, to be more and more efficient and do the operation which might have taken you maybe
[11:55] 1 hour, 2 hours, or maybe even more to do, now you can do it in a couple of minutes. Now, moving on to the next agent. So, this is about the segmenter.
[12:07] Again, consider a scenario where you have to reach out to a segment of your customers uh for some marketing campaign and how easily it could be for you using
[12:19] the agent flow. So, again, I'm logged in into the agent flow, the screen which we saw earlier. So, here first of all, I'm just asking my agent, "Hey, show me all the segments which are available in Relatio today."
[12:33] So, it gives me the list of the segments which are already configured. Now, what I can do is I can just start a new chat with my agent and uh it and I can create a new segment.
[12:45] So, for this, I'll go to the new chat. And here, I'm just giving a simple prompt that, "Hey, create a segment where there are certain set of health care professionals who have prescribed a certain medication."
[12:57] So, since it is connected to the Relatio tenant, it goes and queries and gets the required data. And then, it basically asks my confirmation that, "Hey, I'm going to take this step. Do you confirm to do so?" And I say, "Yeah, looks good. Go
[13:10] ahead and create the segment for me." So, once I do the confirmation, it goes ahead and calls the required um APIs and everything behind the scenes and creates a segment for me and gives gives me a successful message that I've
[13:23] been able to create the segment. But then, it is intelligent enough to realize that, "Hey, this is a sort of a batch operational segment which requires the update at a regular interval." So,
[13:35] it asked me, "Hey, would you like to schedule it?" So, I say, "Yeah, fine. Let's go ahead and schedule it for every Friday at 10:00 p.m." And then, since I didn't specify the time zone, it asked me specifically that, "Hey, would you
[13:47] like to schedule it for a certain for a particular time zone?" So, I say, "Yeah, let's go ahead for the PST time." And it goes in and takes all these steps. So, like we see, uh with just a general prompt and uh
[14:00] providing this information, I was able to do so. Now, switching back to Reality again, I can see the segment which we just created through the agent flow. And here, we can see the schedule as well that, "Hey, uh it has scheduled my
[14:12] segment for every 1 week at Friday at 5:00 p.m." So, combining all of this together, like if you just take a step back of all the information which we have discussed in the last 15 minutes,
[14:25] you use Databricks to create the foundation of your data and AI, but on top of it to add the layer of the golden context which is available to you in near real time every time,
[14:38] and making it combine together. So, basically, your agents have agents AI application, your analytics, or any downstream processes have the same view, same context, and confident enough to
[14:50] take the right decisions at all the time. Yeah? So, that's what I had to showcase today. And we are just right there at booth 540. So, if you have any subsequent question, please feel free to reach out.
[15:03] I'm happy to discuss it further. Thank you so much.

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.