Skip to main content
Skip table of contents

AI Agents

image-20240321-090528.png

Zeta's AI features may not be available in all regions. Please reach out to your Account Manager for additional details.

Besides its transformative functionalities, ZOE also has generative AI Agents that evolve from basic functionalities to more specialized, collaborative, and impactful roles across various domains. A ZOE Agent is like a subject matter expert (SME) in a particular ZMP-related topic who acts as your assistant in walking you through any task/questions related to that topic.

For Example

You might have one Agent that has been specifically tuned to write email subject lines, another that can generate images and one more that creates custom campaign reports. Each Agent works independently to help the marketer do their job more effectively. 

ZMP provides a library of pre-built AI agents that you can easily leverage in your day-to-day work.

image-20240206-090741.png

One Size Does Not Fit All

In addition to a library of pre-built AI agents, ZOE also allows you to create your own custom AI Agent that will be tailored to accomplish a specific task for you within ZMP. With this ability for complete customization, you can easily harness the power of Generative AI to create Agents that are tailored to their specific needs and business cases. The Agent Creation follows the PLACE (Part, Landscape, Assignment, Course, and Example) framework guiding you through a systematic process of defining the role, context, tasks, actions, and expected outcomes for your AI Agents.

Step

Action

Description

1

Part

Define the role of the AI Agent within the marketing team.

  • This could be data analysis, content creation, customer interaction, or any other specific marketing function.

The clearer the role, the more effectively the Agent can be programmed to perform its duties. 

2

Landscape

Outline the operating environment or context in which the AI Agent will function.

  • This might include market factors, target audience, competition, or specific products.

  • Information can also be pulled from the Zeta Data Cloud to augment first-party data for more targeted results.

By understanding the landscape, the AI Agent will be better equipped to make informed decisions and adapt to changes in the environment. 

3

Assignment

Specify the particular tasks that the AI Agent will perform.

  • This could range from analyzing customer data, managing social media accounts, conducting market research, to creating and executing full marketing campaigns.

The key is to detail these tasks as clearly as possible so that the AI Agent knows exactly what it needs to do. 

4

Course

Define the actions that the AI Agent will take to accomplish its tasks.

  • This involves setting parameters around what the agent can and should do in certain situations, such as when to access internal or external APIs. 

5

Example

Provide examples of the desired output.

  • This could be in the form of reports, insights, customer engagement metrics, or any other tangible results that the marketer expects from the Agent.

Providing examples helps to fine-tune performance and ensure the Agent delivers on its objectives. 

Creating Agents

Navigate to Data > AI Studio and click on Create Agent.

image-20250625-071741.png
  • Within the Agent Builder, enter your preferred configuration.

image-20250625-140053.png
image-20250625-140103.png

Agent Configuration

Field

Decription

Example

Agent Name

Choose a name that reflects the purpose of your agent.

Churn Predictor Assistant or Template Optimizer Bot.

Model

Select the desired model. The default is OpenAI-GPT 4.1.

Description

Provide a one-line summary of what your agent does.

“Analyzes past campaign data to predict user churn.”

Instructions (Prompt Setup)

Write detailed instructions to guide the agent’s behavior. This acts as the system prompt.

You are a helpful assistant trained to identify churn signals in campaign engagement data.
Always return a list of users most likely to churn in the next 30 days. If no risk is found, provide optimization suggestions for retention.

Reference Documents

Drag and drop any CSV, PDF, or plain text files that your agent needs for context.

“Campaign logs Q1 2025.csv” or “Churn definition handbook.pdf”

Save Agent

Choose who can use this agent.

  • Self: Only you can access it.

  • Account: Share with your team.

Edit Permissions

Decide who can make changes.

  • Self: Only you can edit.

  • Selected Users: Share editing rights.

Enable Capabilities

Select what your agent can do by enabling relevant capabilities:

Web Browsing

Allows the agent to fetch current information online.

Image Generation

Generate images based on text prompts.

Template Information

Accesses message template metadata.

Campaign Information

Pulls data from campaign performance metrics.

Segment Information

Looks up segment-level data like matched lists or LALs.

Data Cloud Insights

Access proprietary insights from Zeta’s Data Cloud.

Json Validator

Validates and parses JSON structures.

Experience Information

Retrieves engagement metrics for customer experiences.

Pick only what’s necessary to avoid unnecessary permissions.

Create Actions (Optional)

If your agent needs to trigger specific tasks, click “Create New Action” and configure the action logic.

Sending a report, triggering an alert

Click the Save Agent button in the top-right corner once everything is set.

  • Your agent will then appear under the AI Studio dashboard and be ready for use or refinement.

Quick Access to Agents

Once you navigate back to ZOE and type in “@” in the text box, it will show a rolling list of all the agents that are created within your account.

  • Once selected, the ZOE chatbox will transform into the chatbox of that specific agent, helping you accomplish the particular actions that the agent was built for.

2025-06-26_10-13-01 (1)-20250626-044707.gif

JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.