Generative AI is a feature of Social that allows end users to generate social media post descriptions using artificial intelligence. The user can choose the tone and length, and then either select specific words or phrases to include in the description or write a prompt in order to generate the most relevant and helpful text. The user can then edit the text, or add images and links to it, before publishing.
Note: Currently, only the prompt method works with non-English languages.
To enable Generative AI, see Enabling Generative AI.
To see the user workflow for the Prompt method, see Creating a Generative AI Post by Writing a Prompt.
To see the user workflow for the Words and Phrases method, see Generating an AI Post Description Using Key Words and Phrases
To generate alternative descriptions using AI, see Generate Alternative Descriptions Using AI.
To configure brand style guidelines, see Enable and Configure Brand Style for Generative AI.
How does Generative AI find suggested words and phrases?
When an end user generates a post with AI using the words and phrases method, we suggest a list of words and phrases to use in their description. In order to create this list, Yext extracts relevant and timely words and phrases using a ranking algorithm from the past three weeks of publishing history within a hierarchy. If there isn’t enough publishing history for the past three weeks within a hierarchy, the entire organization’s publishing history is used to generate the suggested words and phrases list.
Yext stores 30 words and phrases at a time for each hierarchy, and each workspace is given 10 of them to choose from each time they click New Post with AI. We monitor publishing history daily and update the list regularly to ensure there is new content to choose from and content remains relevant to trending topics.
Suggested words and phrases drop down
How does Generative AI create post descriptions?
Yext uses the OpenAI Large Language Model (LLM) to create post descriptions through automation. The LLM operates within the Yext AWS Bedrock environment, meaning no user or organizational data is sent externally. After a user either writes a prompt or selects the tone, length, and words and phrases to be included in a post description, Yext automatically sends the request through the LLM, and then displays the resulting description to the user.
Example
- The user chooses the following parameters for their description:
- Tone: Casual
- Length: Short
- Words and phrases: get organized, financial health, complete form, beneficiary information
- Yext asks the LLM it to create a post description for social media based on the parameters.
- The LLM sends back the following post description, which Yext displays to the user:
Getting organized is a key step toward achieving financial health. Completing a beneficiary information form is an important part of this process, ensuring that your assets are distributed according to your wishes. Don't delay, take control of your financial future today!
Supervision and Reporting
In Supervision, posts that were created with Generative AI include an internal note with details of the user's request. They are supervised as normal posts, and any lexicon violations are highlighted on the Review and Publish page at the end of the workflow.
The post approval process helps build confidence in using AI by mitigating risk and ensuring those posts meet compliance requirements. Previously, the only way to require approval for all AI posts before they’re published was to require approval for all AI and non-AI posts, which is not feasible or necessary for many organizations.
A new setting allows organizations with Generative AI enabled to require approval for all AI-generated posts, independent of other post approval configurations.
The setting is disabled by default. To enable the new setting in the Administration view:
- In the side menu, click in Settings, then click the Organizational Settings tab.
- On the AI tab, change the Require approval for all posts created with AI to Enabled.
The Post All Data report has a GenAI supported column with true/false values that indicate whether a post was created using Generative AI.
The Workspace Directory report has a GenAI enabled column with true/false values, which indicate whether Generative AI is enabled for a workspace.