Main content

TrailblazerDX 2024 - in review

Rebecca Wettemann Profile picture for user Rebecca Wettemann March 29, 2024
Casting an end-of-month eye over the announcements from the Salesforce developer event.

At the recent TrailblazerDX gathering in San Francisco, it was no surprise that Salesforce was focused on Artificial Intelligence (AI), and in particular its generative AI capabilities. Salesforce’s positioning is around the combination of generative AI, CRM, and Data Cloud – and the Einstein Trust Layer – to deliver trusted AI and automation.

At the event, Salesforce announced Einstein 1 Studio, a set of low-code tools to enable Salesforce admins and developers to customize Einstein Copilot, Salesforce’s conversation AI assistant. Einstein 1 Studio includes:

  • Copilot builder to enable organizations to configure and customize Einsten Copilot.
  • Prompt builder to enable admins and developers to create custom reusable AI prompts without coding so they can be embedded in CRM.
  • Model builder to enable organizations to choose a Large Language Model (LLM) from Salesforce or other providers or bring their own model, and train or fine-tune the selected models on Data Cloud.

These announcements signaled both how Salesforce’s AI strategy is evolving and its recognition that customers’ level of readiness and willingness to adopt AI is dependent on how easy Salesforce can make it.

Although Bring Your Own Model (BYOM) and training and fine tuning models are a key part of Salesforce’s high-level messaging, it recognizes that – at least in the short term – most organizations don’t have the resources, expertise, or risk tolerance to evaluate, build, or train different LLMs or other AI models and are likely to use what Salesforce or partners recommend.

It also recognizes that that these models are evolving very rapidly, and the current 'best' ones are likely to be obsolete in the next 12-18 months – so the ability to swap them out easily is going to be important. 

Although Salesforce isn’t stepping away from its own LLM research, there’s a recognition that it may not be the best use of Salesforce’s resources. Instead, a key part of its strategy is emphasizing openness, and that LLMs, no matter how good they are, will need CRM data for grounding and the privacy and security measures found in the Einstein Trust Layer. 

Salesforce also recognizes that organizations (and departments) are at very different levels of maturity when it comes to Data Cloud and AI:

  • For relative newbies, the prepackaged Einstein Copilot capabilities will enable them to automate some basic activities with generative AI – such as writing case summaries or generating e-mail responses – with no heavy lifting required.
  • For those looking to support more specific or custom AI-driven automation, Copilot builder will let admins and developers go a level deeper while using tools they know – such as Apex, Flow, and MuleSoft APIs – to support more complex automations that span multiple clouds and applications beyond Salesforce.
  • Admins and developers that are more advanced will be able to use prompt builder to design and build prompts without code that can then be embedded for reuse by a broad population of users. 

My take

Wherever its customers are on the maturity curve, Salesforce recognizes that it has to meet them where they are and help them see a rapid path to benefit from AI if it expects them to grow their investments in Einstein and Data Cloud.  

A grey colored placeholder image