Ever wondered what your technical role would be as a Data 360/AI Admin?
With Salesforce in its Agentic Era, and AI agents being only as smart as the data they can see, Data 360 (formerly Data Cloud) is probably the centre of the Salesforce and other businesses’ universe right now. So, venturing in to the realms of Salesforce AI wouldn’t be complete without looking at Data 360. But like anything AI-related, it can get quite technical, so hold in there.
My name is Jenny Bamber, Head of Salesforce Success at Desynit, a Salesforce Partner based in Bristol. I have taken the plunge into the depths of Salesforce AI and Data 360, which, believe me, is not for the faint-hearted. However, as I wrote in my previous blog…., I believe it’s a necessary step forward for a Salesforce Admin wanting to keep up in this Agentic era.
Come with me as I look at some of the core pillars behind Data 360 and what responsibilities as a Salesforce Data 360/AI Administrator will look like.
Whilst Data Stewards have always been a thing, with the main focus being Data Quality, the Data Stewards of 2026 have changed from a kinda Clean-Up role to Curator of AI Intelligence, with the main priority now being Grounding. More specifically, to Salesforce, Data 360 AI Grounding. Which, in a nutshell, are concepts used to provide a solid data foundation (a single source of truth) for your AI to digest. I like to think of it as a golden record containing all the information about a customer, masterfully curated and ready to be used by your Agents when needed.
Let’s have a very brief look at what it takes to get there.
This is all about managing how data enters Salesforce without having to physically move it. Think Zero Copy Data Management.
This is no longer about importing CSVs, but creating new Data Shares (so that’s reading data from Salesforce) with warehouses like Snowflake, AWS, etc, with no export required. And then Ingestion, allowing Salesforce to read data from other systems. To create Data Streams that allow for real-time flow and processing of data.
Once I have this data. I can create some nice little Data Spaces to organise these streams in logical buckets, e.g. region, gender, etc, to help AI Agents provide responses that are accurate, relevant and secure.
With all this data coming in from different sources, there is a high chance that naming conventions are varied for the same thing, e.g. Email vs Email Address.
Not only will Data 360/AI Admins ensure that the data in my own Salesforce org is clean, but also data flowing into Data 360 streams (think specific information like UK Sales Orders, Website Product Clicks, etc.) is clean too.
This requires some serious Data Mapping of Data Lake Objects ( DLO) from warehouses like Snowflake or AWS, and then mapping that to the standard 360 Data Model Objects (DMO).
During this process my questions may look more like, “is this data safe for AI grounding” (using the Einstein Trust Layer Guardrails)…“is there a consent flag included in this table” (think Governance and Ethics checks)… and “does the external data source have a unique Party ID” (think Harmonising Data to fit into Salesforce).
This is way more than cleaning up a spreadsheet, Salesforce Admins. The goal is to remove hallucinations by providing a verified source of truth ready for use.
This is the fun part of being a Data 360/AI Admin. The sorting of the Jon Doe’s, from the John Doe’s, so that I know that the person in Salesforce is the same person who just opened up that marketing email that’s just been sent.
Remember, I said that duplicate management is still important; well, this is where it plays its magic. By creating and combining Matching Rules (exact match, and Fuzzy Name + email, etc.) and reconciliation rules (deciding which system has the true data), help create that single golden record for every customer.
Right, so the hard work is done. I have raw, clean data that I can actually make useful by turning into Insights.
I’m going to want to know multidimensional and calculated metrics like Customer Lifetime Value (LTV) or Satisfaction Scores (CSAT) across all my sources. I’m going to want to predict future outcomes to help the Sales team with the help of Einstein Discovery, and its ability to use AI to predict patterns, e.g. This customer is 80% likely to upgrade if offered a demo. And then finally, I am going to want to make trends pop by creating dashboards using the likes of Tableau.
Check out this Trailhead module to learn more about churning insights with beautifully curated data.
Once I have my data and it’s giving the right insights in pretty dashboards, it becomes a bit useless if I let it sit here. And after working hard to get to this point, there is no way I am about to let the ball slip here.
Let’s bring back a core skill to the table: Flow Builder. I can use my skills to build a flow that triggers an automated workflow (like a Slack alert or a refund) based on a change in my Golden Unified Record.
But I also want to do more than Flow. I want to build Data Actions that allow me to send a segment of “At-Risk Customers” directly to Marketing Cloud for a targeted email campaign.
The aim is to make sure that every interaction is continuously being monitored and a positive action is created in a bid to keep the business successful and customers happy.
This is scratching the surface of Data 360 and Salesforce AI; however, it’s a great introduction to those embarking on this new AI landscape.
If you like what you read here, then check out my blog: Is this the Death of the Standard Admin: Why Data 360 Is Your New Reality.
Our independent tech team has been servicing enterprise clients for over 15 years from our HQ in Bristol, UK. Let’s see how we can work together and get the most out of your Salesforce implementation.
