SFDC Penguin-Salesforce AI Day Cover

Salesforce AI Day: Three things you need to know

On the 12th of June 2023, Salesforce hosted ‘Salesforce AI Day’. The keynote is available on Salesforce+. Not had a chance to watch it? No problem! Here are the three key take-home messages you need to know.


Before getting started, it is worth having a basic awareness of generative AI (artificial intelligence), its terminology and pros and cons. If this is new to you, check out the section below, otherwise, feel free to skip to the key take-homes.

Salesforce already does AI. What’s new?

AI is not new. In fact, Salesforce have been making strides into AI since 2014. In 2016, Salesforce introduced the Einstein’ character to be the face of all things AI in Salesforce.

Salesforce Einstein Image
Image of the Salesforce Einstein character. Property of Salesforce.

Up until recently, ‘predictive AI’ has been the focus. This uses ‘machine learning’ to analyse data, identify patterns and make predictions. Salesforce examples include ‘Einstein Lead Scoring’ and ‘Einstein Opportunity Scoring’. In fact, through Salesforce’s existing AI tools, ~1 trillion predictions are made per week!

Tools such as this allow teams to focus efforts wisely. However, we’re moving into a new era of AI.

Generative AI: A new era of AI

Generative AI solutions, such as ChatGPT, Bard and DALL-E have been making headlines. They are causing a mix of excitement and concern.

Generative AI: Reasons for excitement and concern
Generative AI: Reasons for excitement and concern

The above is not definitive, but there is no doubt generative AI represents a potential paradigm shift. As with all new technologies, there are challenges, but also significant benefits to be realised. 

Understanding the terminology

AI day focused on Generative AI. Here are some key terms to be aware of:

  • Bias: Unintentional outputs from an AI, which unfairly favour certain results. Often caused by data used to train a model in the machine learning process. Click here for more info.
  • Data Masking: Keep personally identifiable information (PII) data secure by modifying it prior to sharing with other systems (e.g., AI models). Click here for more info.
  • Generative AI: Based on a user’s prompt, AI uses the data it has consumed to create something new, such as text, images, or code. Click here for more info.
  • Hallucinations: Where generative AI analyses data provided but produces an invalid or unrealistic output. Click here for more info.
  • Large Language Model (LLM): A type of generative AI; typically trained using text to generate text outputs based on a user’s prompt. Click here for more info.
  • Predictive AI: An AI model designed to analyse prior data to predict an outcome to help decision-making. Typically a simpler model than generative AI. Click here for more info.
  • Prompt: A question and/or instruction provided to an AI tool to generate an output (e.g., providing instructions to a LLM to create a specific, personalised email). Click here for more info.
  • Toxicity. Output from AI which is defined as a “rude, disrespectful, or unreasonable response”.  Click here for more info.
  • Zero Retention: Ensuring that no data is stored outside of Salesforce. Click here for more info.
For a glossary of other generative AI terms, check out this article from SalesforceNow we’ve explored some of the background, lets look at the three key take-homes from AI Day

Take-home 1: Trust & responsibility

Trust featured throughout the talk. Given the challenges posed by Generative AI (e.g., security, privacy, hallucinations, bias, and toxicity, etc), this is not surprising. LLMs consume large amounts of data. However, what if data is unvetted, biased or outdated? It leads to poor quality outputs. 

Trust is a core value of Salesforce. To overcome these challenges, Salesforce has unveiled ‘AI Cloud’, which is built on the Einstein GPT Trust layer.

The Einstein GPT Trust Layer creates separation between all of your corporate enterprise data stored in your CRM, in databases where we can apply access controls and it allows you to responsibly ground all of your prompts in that data without that data ever leaving Salesforce.

The diagram below shows the key aspects of the Einstein GPT Trust layer. Check out Srini Tallapragada’s (President & Chief Engineering Officer, Salesforce) walkthrough here.

Take-home 2: Product releases

Salesforce clearly has ambitious goals for Einstein GPT, as reflected by recent releases:

More is yet to come. Currently, Apex GPT will be available as a pilot later this month, with Flow following in October 2023. For more information on Salesforce’s ‘Summer of AI’, check out the slide below, or watch Clara Shih, (CEO, Salesforce AI), walkthrough some of the upcoming features.


Take-home 3: Training

Lastly, if this has whetted your appetite to learn more, Trailhead has 35 badges available covering various AI topics. Alongside this, a new Generative AI Certification will be launched at Dreamforce!

If you are interested in exploring the Trailhead modules available, check out this article for more information.


This post has provided a brief summary of the announcements from the AI Day keynote. The keynote covered a wide range of topics (and case studies) to highlight how Salesforce will help its customers get the best out of Generative AI. Whilst this is still an emerging technology, there is plenty for businesses and individuals to be excited about. 

If you want to get ahead of the curve, be sure to keep an eye out for further announcements and check-out the Trailhead modules!

Until next time, thanks for reading! 🐧

Bonus Penguin Fact

This month is Pride Month and in the UK, today is also Father’s day! It seems only fitting for this penguin fact to cover both!

Did you know there are plenty of examples of same-sex penguin couples? Check out this video about Sphen and Magic, two male Gentoo Penguins at Sea Life in Sydney, Australia. These two caring penguins have successfully adopted multiple eggs. Check out this video for more info.  

Want More?

Found this article useful? Why not share!


Or check out some of our other articles:

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.