by Zoe Fisher 
In the Spring’24 release, Advanced and Premium Account Engagement customers were given the gift of generative AI in the form of Einstein Assistant. Let’s take a closer look at this tool…
Co-create with Einstein Assistant
Struggling to put words together? We’ve all been there! Einstein Assistant helps users to generate content based on prompts provided. The assistant is available for forms, landing pages and list emails, in the Account Engagement Lightning App (not the lightning experience for emails and landing pages) and on pi.pardot.com. You can use the assistant for the following elements:
Forms
- Form Name
- Tags
- Headlines
- Paragraphs
Landing Pages
- Landing Page Name
- Tags
- Landing Page Title
- Landing Page Description
- Headlines
- Paragraphs
List Emails
- Headlines
- Paragraphs
- Subject Lines
Writing prompts that generate useful output
The output of natural language AI is only as good as the input. To ensure the content that Einstein generates for you is useful, write clear prompts with content and specific instruction.
Prompt is a natural language text that you craft to communicate with an AI. The more specific you are the more useful and accurate the results will be. To ensure the content generated by Einstein assistant is useful and relevant, make sure you write clear prompts with specific instruction.
We recommend following the below structure when writing prompts:
- Introduce the situation
- Give content to help Einstein understand what you are looking for
- Give clear instructions on what the outcome should be
[Situation] + [Background / context] + [Instruction]
Example:
[Situation] I work for an event company called PURSUIT, we run an event aimed at professionals in the beauty industry. [Context] At our event, we connect beauty pioneers with suppliers, media partners and brand experts who share insights on how to stand out against competition. The event occurs on 12-13 July 2024 in London at the Olympia building. [Instruction] Instruction: Write an email promoting the event, to increase registrations.
The assistant will generate a response for a much shorter and vague input, however this is more likely to be generic and non-specific for your audience.
How secure is the data in Einstein Assistant?
Data input to the Einstein Assistant is retained on an individual asset and tab basis, meaning that if you access a different asset or switch tab on an asset, the chat history will be unique.
Salesforce’s Einstein Trust Layer contains data security guardrails such as data masking, TLS in-flight encryption, and Zero Data Retention with Large Language Models. This is applied to the generative AI features baked into the Salesforce platform:
Do you want some support to get started with AI? Get in touch with us today!