Skip to content
Optimizely CertificationOptimizely Certification Optimizely Certification

Learning Path: Opal Agent Builder

⏱ 4-5 hours intermediate
📜CoreOpal

Who this path is for

Section titled “Who this path is for”

You want to build AI agents that automate tasks for your team using Opal. You may be a marketer who wants to codify best practices into reusable assistants, a developer integrating AI into workflows, or an admin responsible for agent governance. By the end of this path, you will understand how Opal works, what agents can do, and how to build one from scratch.

How to use this path

Section titled “How to use this path”

Work through the steps in order. Steps 1-3 build your conceptual understanding of Opal. Steps 4-6 teach you how to configure the building blocks of agents. Steps 7-8 give you hands-on experience building and evaluating a real agent.

Estimated total time: 4-5 hours (spread across multiple sessions is fine).

Step 1: Understand the Opal AI Platform

Section titled “Step 1: Understand the Opal AI Platform”

Read — Opal AI Platform

You will learn:

  • What problems Opal solves for marketing and digital teams
  • How the eight-step processing pipeline handles every request
  • The three types of AI agents: simple assistants, specialized agents, and workflow agents
  • How the credit system governs usage
  • Why Opti ID is required for authentication and access control

Time: 20 minutes

Step 2: Get started with Opal

Section titled “Step 2: Get started with Opal”

Follow — Get Started with Opal

You will learn:

  • How to access Opal through home.optimizely.com
  • How to navigate the Opal interface
  • How to submit your first request in Chat
  • How Opal selects tools and agents to fulfill your request

Time: 15 minutes

Step 3: Explore Chat and Canvas

Section titled “Step 3: Explore Chat and Canvas”

Read — Opal Chat and Canvas

You will learn:

  • How Chat provides a conversational interface for Opal requests
  • How Canvas delivers collaborative workspaces for creating and editing digital assets
  • The relationship between Chat interactions and Canvas outputs
  • When to use Chat vs Canvas for different workflows

Time: 20 minutes

Step 4: Configure instructions

Section titled “Step 4: Configure instructions”

Follow — Configure Instructions

You will learn:

  • How instructions shape Opal’s behavior and output quality
  • How to write instructions that define persona, process, and constraints
  • How organization-level instructions establish brand voice and compliance rules
  • How to test and iterate on instructions for consistent results

Time: 30 minutes

Step 5: Manage tools

Section titled “Step 5: Manage tools”

Follow — Manage Tools

You will learn:

  • What tools are and how agents use them to interact with Optimizely products
  • How to enable and disable tools for specific agents
  • The principle of least privilege applied to agent tool configuration
  • How tools affect agent behavior and credit consumption

Time: 25 minutes

Step 6: Build custom agents

Section titled “Step 6: Build custom agents”

Follow — Build Custom Agents

You will learn:

  • How to create specialized agents with custom instructions, tools, and input variables
  • How to create workflow agents that chain multiple steps together
  • How to browse and install agents from the Agent Directory
  • How to use agents in Opal Chat with mentions
  • How to monitor agent performance through execution logs

Time: 30 minutes

Step 7: Set up AI evaluations

Section titled “Step 7: Set up AI evaluations”

Follow — Set Up AI Evaluations

You will learn:

  • How evaluations automate quality scoring for agent output
  • How to create evaluation sets with example inputs and expected outputs
  • How to run evaluations and interpret quality scores
  • How to use evaluation results to refine agent instructions
  • How to schedule ongoing evaluations for quality monitoring

Time: 30 minutes

Step 8: Build your first AI agent

Section titled “Step 8: Build your first AI agent”

Do — Build Your First AI Agent

You will build:

  • A custom specialized agent with a clear purpose statement
  • Structured instructions covering objective, process, output format, and constraints
  • Input variables that make the agent reusable across different inputs
  • An evaluation set that tests agent quality across multiple scenarios
  • A deployed agent accessible to your organization through Opal Chat

Time: 45 minutes

What to do next

Section titled “What to do next”

After completing this path, you are ready to:

  • Build agents for your team — Apply what you learned to create agents for your specific business processes: content drafting, campaign planning, data analysis, and more
  • Explore RAG — Connect your agents to your content library through Retrieval-Augmented Generation for context-aware responses
  • Build cross-product workflows — Follow the AI-Powered Content Review Pipeline or Automated Experiment Analysis recipes to connect agents across Optimizely products
  • Establish agent governance — Define naming conventions, review processes, and quality standards for agents across your organization
  • Prepare for certification — Visit Optimizely Academy for Opal certification tracks