It’s official: Salesforce 3.0 is here, and the Cloudwerx team has never been so excited. The new Salesforce combines Data Cloud, Generative AI, Retrieval Augmented Generation (RAG), Einstein One and now AgentForce to produce a set of capabilities beyond anything that’s ever been possible in Salesforce before.
At our recent Werxfest, our CTO, Chris Baldock and CEO of Lightfold (A Cloudwerx Company) John (JC) Cosgrove sat down to discuss these changes, and what they’ll mean for Salesforce customers. In this post, we’re going to dive into the new changes, and how you can take full advantage of these new capabilities today.
Right now, we trust AI to generate. Soon, we’ll be trusting it to act.
Now, if you’ve been keeping up with Salesforce for some time, you’ll be familiar with their AI innovations over the past few years. From the game-changing introduction of Einstein Discovery to their customer churn and propensity-to-buy prediction models to their record summary SKUs for generative AI — with each new development, Salesforce customers have been granted access to a wealth of new functionality.
But the new AI capabilities? They’ve completely shifted the paradigm.
Salesforce AI is no longer just generating text; it can actually do things, like update a record or even send an email on behalf of your company. It creates a plan of attack based on the language you use in the query, deciphering context from the language you use. By picking up on what you’re asking, the new AI is able to determine the necessary next steps. You’re not required to build a fully fleshed out process—the Salesforce 3.0 is smarter than that. You provide the tools, and the generative AI has the ability to pick the right tools for the job.
This is where the next wave of innovation is heading—moving from generation to action.
As John points out:
“In addition to the intuitive understanding that more context equals a better response, it’s mathematically deeper. The context you provide changes the weights and attention of the model’s execution. This is why having it built into the Salesforce architecture from day one is market-beating.”
The way we store and access records has changed, for good
Have you ever used a ‘vector database’? It’s all about storing information in a way that’s incredibly efficient and meaningful for generative models. By capturing the meaning and context of your information in a vector database, you ‘ground’ your generative AI’s responses in your world.
Now, Data Cloud has a vector database. This is fantastic news for users and administrators because it supports semantic search — where responses to prompts or query are more powerful because they also include contextual information, allowing for the fast extraction of meaning.
Additionally, Data Cloud can now store both structured and unstructured data. This is a huge development, as users are no longer limited to just storing records, like rows on a table. Now, you can store PDFs, videos, audio snippets, images and more — and Data Cloud is able to extract and search content from those different formats. In particular, we at Cloudwerx believe that these updates will completely change the way contact centres operate, taking entire libraries of knowledge and making them instantly available via Copilot or generative AI.
Salesforce will continue to innovate and build additional capabilities in trust
Finally, Salesforce has been developing additional trust capabilities that are designed specifically to support generative AI. As we know, trust has always been a crucial component of the Salesforce platform, and so it’s essential that whenever a user engages with generative AI, the query will go through several steps before producing an output. These processes include:
Secure data retrieval: This ensures that when you ask Salesforce bot to perform a task, it can only access data based on the user or bot’s privileges. In short, it prevents access to any data that you shouldn’t be able to see.
Dynamic grounding: This feature ensures that the AI doesn’t just rely on the data used to train the underlying language model (for example, GPT). Instead, it prioritises your data, so that the response you get is relevant, accurate and most importantly, reliable.
Data masking: This feature is vital for organisations that handle a large amount of personal and sensitive information. Salesforce now offers tools that allow you to classify sensitive information within your organisation and industry, which helps their AI dynamically identify and maks personal or sensitive data to ensure it’s never shared inappropriately.
Toxicity detection: This layer ensures that the content generated is always filtered to avoid toxic responses — from biases, profanity, racism and other unwanted content. The result? The responses Salesforce generates are always appropriate and safe to share.
Zero data retention: Salesforce acts as a gateway to the language model, rather than the model itself. They have a strict policy with their partners, including language models like OpenAI, to ensure that any data used in generating responses is not stored or retained afterwards. The data is used only in the context of generating the output, but it’s not persisted or saved.
Getting on board with Salesforce 3.0
Thankfully, Salesforce has made it really easy to get up to speed and start implementing its new features. One of the best ways that you can begin to interact with Salesforce 3.0 is through their new Einstein for Sales and Einstein for Service licenses. They give you all the core capabilities of Sales and Service, along with traditional predictive AI and new generative AI capabilities. You’ll also have access to Copilot, Einstein platform capabilities and pre-loaded capacity for Einstein AI requests, Data Cloud storage and Data Cloud credits.
At Cloudwerx, we have multiple streams available to support our clients, including a dedicated AI advisory team and a blueprint for guiding any organisation along their journey to implementing autonomous AI. So if you like what you’ve read, and want to start taking advantage of these new capabilities, we’d love to hear from you.
To get started, reach out to your local Cloudwerx rep and start the conversation today!
Interested in working with Cloudwerx?
Please reach out to hello@cloudwerx.com. We offer free consultations, and would love to hear about your business.