Zero Copy 101: Building Open Data Ecosystems with Data Cloud | Dreamforce 2025

You.

In this session, we're going to show you how Data360 can power open ecosystems. As you've heard throughout today's Dreamforce presentation, we are in the middle of an agentic revolution. And for AI agents to be able to work successfully, they need access to all kinds of data, both structured and unstructured. And winning companies actually need to be able to turn that trapped enterprise data into intelligent action. So we're all familiar with all of the AI experiences that we see today across all the touch points where AI can reach us. Sometimes they're at the right time, providing the right context, but all of this doesn't really happen on its own. For data to be able to be used by your AI agents, and for you to be able to activate on that data, you need to bring it all together. But today, data remains fragmented, so data is scattered across multiple systems. So you can have your sales service, marketing, third party applications. All these systems often don't even talk to each other. And second, even when you bring all this data together without connecting it back to a single source of truth, or most importantly your customer, you really don't have any reliable insights. And then you have poor actionability, meaning you have the data. You really don't have enough to drive context, to have the right types of actions with your customers. This is where Data360 comes in. It's the solution that's built to solve all of the challenges that you need around fragmented data, around having the right contextual data for your agents and for your customer.360. So with Data 360, you can activate trusted data everywhere. Imagine this. All of your enterprise data scattered across all of your multiple systems, all available to you instantly without moving it in some cases. Then every workflow, every decision and every interaction is working off of this one single source of customer truth. So your business is actually moving in sync with the customer. Agents don't just need raw data, they need data that is meaningful and contextual. And that's exactly what Data 360 does. It brings context and insight to your data, so agent force can put it to work. So how does Data360 work? First, we connect all of your data with data cloud so you have multiple ways in which you can do this. We have 270 plus out of the box connectors where you can bring your data from your disparate systems. Whether it's CRM orgs, whether it's third party systems, whether it's different sorts of apps, you can bring that data into Data Cloud And Vijay is going to talk about this more. But with our groundbreaking and innovative Zero Copy technology, you don't even have to move that data over. You can actually connect to data in your hyperscalers like Snowflake or Bigquery or Databricks. And then you have the advantage of no data duplication, no brittle ETL pipelines. And most importantly, that data becomes immediately available to you to be able to act on it. Once you connect your data, you actually do what we call harmonization. Harmonization is where you take all your disparate sources of data and you connect it all back into one single canonical model that's defined by Data Cloud. So whether you have different definitions of the same object in different systems, you can map it all back into one canonical model. In Data Cloud, think of it as like a Rosetta Stone. You're mapping it all back to one object. Once we harmonize it, Data360 allows you to set up identification rules and you can unify your profiles. Once you've unified them and you've created this single source of truth, you can actually activate your unified profile across any workflow or across agent 4s. I'm sure you've heard about governance a lot. We now have governance within Data Cloud. So you can apply governance and set up policies on how you want your data to be accessible across all of your workflows using Data360. And another thing to keep in mind is while Data360 works very well to connect and unify data, the real power comes when you're actually able to act on that data. And that's exactly what the Data 360 does. It allows you one, to create rich context around your customers so you know exactly who your customers are and what they're looking for. So with segmentations, your segments are precise, your agents have the right context, and you're able to activate all of this data anywhere with real time. So to summarize, Data360 gets all of your data to work for you. So any data, whether it's structured, unstructured, coming in via our 270 plus out of the box connectors, coming in via zero copy, it comes into this unified metadata framework and you can use it to trigger any action across your customer 360 or agent force. So think of it as routing leads to with product lead, qualification, marketing, nurture journeys, hyper personalized campaigns, and for AgentForce, you're able to ground every interaction with the rich, complete view of your customer. So we have multiple ways to bring data into Data Cloud. Into Data 360, we have batch and streaming, which is plant transfers. You drop your files on hyperscaler, we can import that data in, we have unstructured, where we have our SDKs to bring in data into Data360. And we have our real time framework as well where we can actually ingest the data into data360 in sub seconds. But for today, what we're gonna focus on is zero Copy. And Vijay is gonna talk to us about how you can virtualize your data without moving it.

Thank you, Gina. All right, so as Gina mentioned, Zero Copy is our next game changer. You have the ability to access your data from your source without ever having to move it. You don't have to make copies of it, no ETL pipelines. All you gotta do is anytime you wanna use the data in Data Cloud, you're directly able to access it from your Source. So with Data360 0 copy, it's built in open extensible iceberg platform. You're able to connect your data with clicks, not code. Right? I mean, you don't have to worry about complex ETL pipelines, don't have to worry about refreshing your data. Anytime you're looking to use the data, you directly go to the source and bring the data. Not only that, we also have the ability to push down filters. So when you're reaching out to the source, not bringing the whole data set back, you're bringing back the pertinent data set that's based off your use case in Data Cloud. Secondly, this is bidirectional, meaning you can use data that's available in sources without copying it into Data Cloud. And similarly, if you want to use data that's available in Data Cloud, in your external systems, you could use the same technology without having to worrying or worry about copying or keeping multiple versions of it. Third, you have the ability to unlock the full value of your data. You don't have to, there's no heavy lift at all, right? All you have to do is set up a connection and then the rest it flows automatically. So over this year, we launched File Federation with Snowflake Databricks, AWS and IBM. What is File Federation? We have two modes which we'll talk about it in a second. There's Query Federation and File Federation. I'll dive deep into that in a second. But with File Federation, you do not utilize your compute at all. With Query Federation, we run the queries on your data warehouse. With File Federation, we directly access the data from your storage layer, run it on salesforce Compute, which means you do not have a compute. You do need a compute layer, you do not have a compute cost associated with it. Secondly, we have our Zero Copy framework where we have now started building new connectors based off the Apache Iceberg generic connector that we have. So we have the basic connector. Anyone who has parquet files, an iceberg rest catalog and iceberg tables are able to leverage this and build a connector for their own system. Thirdly, we are launching 108 new zero copy sources. So lots of goodies there. Azure SQL Server, Azure Synapse, Postgres, MySQL, MongoDB, Enterprise DB so lots of new connectors coming in where now you'll have the ability to bring data via Zero Copy into Data Cloud from multitude of sources. So so let's talk a bit about Zero Copy, right? Zero Copy is a new blueprint for data collaboration. We spoke about bidirectional Zero Copy. We spoke about Data Sharing which is sharing data from Data Cloud to external sources. We spoke about Federation which is accessing external data in Data Cloud. We also launched recently a new feature, DC to DC Sharing where with Federation and sharing it's from an external or to an external source. With DC to DC sharing you're able to use Zero Copy between two DC tenants. So if you have two DC tenants and you want to use Zero copy, you can go ahead and use it. We have Clean Rooms which is built on DC to DC sharing, where again we use Zero Copy, you're not moving any of the data. And lastly we are working on Personal Spaces that's for enabling self service scenarios for analytics. As you can see we are moving towards Zero Copy being a foundational piece where now you will not need to copy your data. You can use your data from where it is as it is and still leverage the power of it in Data Cloud. So let's look at this from an end to end perspective. I spoke about in Data In I spoke about Query Federation and File Federation. Query Federation is when you push down the queries that emanate from Data Cloud directly down to your data warehouse. So let's take an example. If you're looking to segment customers who have made a purchase in North America over the last 60 days, you got a bunch of filters, you got regional filters, North America, you have a date filter which is last 60 days, we have the ability to push down those filters to your source. Let's say in this case if you have a Snowflake data warehouse, we're able to push it down to Snowflake Run the query and just bring back the pertinent data set that others to all the filters that you set in Data Cloud. Similarly based off the service principle that's used to create the connection, we honor any rls, fls, ols, anything that's set in Snowflake. So all your governance that's set in Snowflake is automatically honored and therefore the data that's coming in is pertinent to that user. It's just not the filters. It also has the governance that's set in Snowflake. Now with File Federation, like I said, we bypass the compute. We directly hit your storage layer and we're able to access the files. When you have a scenario where your data cloud tenant and your storage layer are in the same region, you will see near native performance. In fact, we've been seeing blazing fast performance with billions of rows of data with File federation again prerequisite for file federation, Iceberg rest catalog, Iceberg tables and parquet files. It's based off the Apache Iceberg system. Once you do that, you would create external data lake objects. Think of them as standalone tables that represent the data in the source. You would map them to a unified metadata model and then you resolve customer identities by resolution. And now your data is ready for analysis. Within Data Cloud you can light up multiple use cases. It could be agent forced, it could be segmentation and marketing, it could be analytics. You can do all that with that dmo and once you're done with that, if you want to share this data, let's say with external systems, it could be sharing this data with your technical marketing team for them to join it with native data and run queries. Or it could be, let's say to run purchase propensity models where your data scientists want to run purchase propensity and determine your next cohort of customers that you want to reach out to. You can do all that with data sharing based. We currently support all the hyperscalers when it comes to Data Cloud. Sorry, Data 360 and we have many more coming. Okay, so major use cases that light up with zero copy today. AgentForce, you have analytics where there's tableau dashboarding, you have segmentation marketing and segmentation and then customer360 where you can bring all your data together and then create a unified profile of the customer. We also have our Zero Copy partner network. We currently have our premium technology partners that we are working with. AWS databricks, Snowflake, Google Cloud. We support Zero Copy, a lot of them recently launched IBM support and we are working towards Microsoft Azure and Workday. I also spoke about the 108connectors. So some great sources you got Aurora, RDS, SQL, DB, Synapse, Postgres, Hive. So some you know, irrespective of where your data is, we are expanding our suite of connections to ensure that you have the ability to access them by zero copy. In Data Cloud we are working with a rich partner set for the data ecosystem. It's got Moody's Zoom info and also we got a great set of implementation ready SI partners who are helping our customers make all these use cases successful. With that being said, what's next in Zero copy we're going to have later this year we're going to have an Azure Synapse connector as a part of the 108. We are working on the Microsoft Fabric File Federation Systems workday connectors, DC to DC sharing which I'd spoken about. We are currently in beta and in October we will go GA. It's October, so end of October, zero copy connectors, 108 connectors in number, will launch the beta program and then we'll start rolling out the GA over the next year and then the Zero copy framework where you will be able to leverage the iceberg connector, build on it and use it with different sources. Okay, with that being said, I'm gonna do a quick demo. You don't have to trust anything I said you can see the demo and we'll see how it all works. So we're gonna look at how a company, Tamuno Robotics, enhance customer Service by using Data360 and agents. Right. So here I'm gonna go. Okay, there we go. All right. So this is Tamono Robotics. Why Tamono? Don't ask me. That's a tough question. But this is Tamono's page. They have an account, Omega Inc. That they're working with. The orders here they're able. This is the page where they're able to see everything about Omega Inc. So this table on the bottom, it's a customer orders table where they're able to see all the orders that the customer made. Now if you look at the first row, we have an order where the order status is pending. The order priority is low for shipping. So the customer calls Tamono, Omega calls Tomono. They wanna know if they can get expedited shipping. Give me a second. I'm better when I use the most with this. Oh actually you know what, nevermind. So while the, while the agent, while the customer representative is talking to the agent, he quickly uses the data Cloud agent to determine what this customer's order set is. Sorry. So let's start. He's going to start off by summarizing the account while he's eyeing the conversation. And they're going to look at what has this customer done, what is their revenue, where are they? So give it a minute. Once it pops up. Okay. While he's doing that, we can do something. Right. We believe in multitasking. So the data comes in from Snowflake. Right. So this is where we're looking at the customer orders table. We said it's all flowing in from Snowflake. So we're going to first run a query, see what's the current customer status, pull up the results. Okay. So there you can see the order status is pending and the order priority was low. This is exactly what we saw on the page. The customer page. So the agent gets the customer summary. The agent, quicksilver. Okay, I'll add another one. The human agent gets the information and decides to see if this customer is eligible for expedited shipping. Shipping. So it's going to look. It's going to say yes based on the current order value, they are eligible for shipping. Note that this value, the current order value is coming from Snowflake, from the table we saw. And then. So now the. And I'll skip this part. You know. I know, I know. I promised Gina I'll give her the last two minutes. I've taken 30 minutes. So now the agent, human agent can go change the status of the shipping to in progress. Auto status being high. It's updated here. Let's go back. Look at this table. Let's see what happened there. So if you look at the first row, it's in process. The order priority has gone too high. And as they've gone through expedited shipping, the agent asks, what products can I recommend to this customer based on the purchase. Customer. And it's going to give you the set of products that the agent can then recommend to the customer for future purchases. So you can see, as you can see, this whole thing is powered by Zero Copy here. Gina.

Sorry, 30 seconds. Okay, thank you. Back to the slide. So I just want to quickly announce. Thank you, Vijay, for that great session on Zero Copy and the great demo. We are running out of time, but I just want to share that we are actually having. We now have pro code capabilities with data cloud. So we are all about zero code. But now, now we have AI, CTOs and CIOs asking for more capabilities. So we now have pro code capabilities with data cloud, and that's in GA shortly. But I want to thank you for your time. And if you have any questions, Vijay and I will be here, and we'll be more than happy to answer them. Thank you, everyone.