A day in the life of an Optimizely OMVP - Optimizely Opal: Specialized Agents, Workflows, and Tools Explained

The AI landscape in digital experience platforms has shifted dramatically. At Opticon 2025, Optimizely unveiled the next evolution of Optimizely Opal – positioning it not merely as an AI feature bolted onto their DXP, but as a fully independent agent orchestration platform. As Kevin Li, SVP of Product at Optimizely, put it: "We're after 20X improvements. We need autonomous processes to run."

This isn't just marketing speak. Optimizely has committed a dedicated team of 30-40 engineers, product managers, and designers solely focused on building Opal as a standalone platform. Let's break down what this means for marketing and digital teams.

What Is Optimizely Opal?

Opal is Optimizely's AI-powered agent orchestration layer that sits atop Optimizely One. Built on Google's Gemini models, it combines three core elements to deliver what Optimizely calls "infinite scale" for marketing teams:

•      Industry-leading context windows – Opal intelligently combines diverse data sources including your corporate knowledge, existing assets, campaign performance, and experimentation results

•      Purpose-built tools – A library of system and custom tools that connect to your data and take action on your behalf

•      Agentic workflows – The ability to orchestrate multiple AI agents working together on complex, multi-step processes

Understanding Specialized Agents

Agents in Opal are intelligent assistants that use natural language instructions and tools to complete tasks. Rather than requiring users to understand underlying systems or APIs, you simply describe what you want, and the agent handles the rest.

Default Agents

Optimizely provides a library of pre-built default agents through the Agent Directory. These handle common tasks out-of-the-box such as:

•      Content model creation agent – Analyzes URLs or images to build best-practice CMS (SaaS) content models, referencing existing models to prevent duplicates

•      GEO recommendations agent – Audits webpages for LLM discoverability and delivers actionable recommendations to optimize content for generative search

•      Heatmap analysis agent – Transforms heatmap images and URLs into precise, tailored test ideas

•      AI variation development agent – Helps modify website elements and generate enhancement suggestions while maintaining brand consistency

•      Accessibility-check agent – Evaluates pages against WCAG standards without manual review

•      Keyword research agent – Takes a topic and returns relevant keywords using data from Optimizely Idea Lab and Google Search

Specialized Agents

Where default agents provide immediate value, specialized agents let you build custom AI capabilities tailored to your organization. These operate in a "single-shot" model – you provide inputs, the agent runs once, and returns a result.

Key characteristics of specialized agents:

•      Controlled context – Define exactly what inputs, tools, and instructions the agent can access

•      Modular design – Reuse agents as components within larger workflow agents

•      Independent testing – Test, tune, and debug each agent independently for faster iteration

•      Fine-grained optimization – Experiment with inference settings or prompt variants per agent without affecting the rest of your workflows

Workflow Agents: The Orchestration Layer

This is where Opal truly differentiates itself. Workflow agents allow you to combine multiple specialized agents into complex, automated processes. The new drag-and-drop visual interface makes it possible for non-technical users to build sophisticated automations.

Workflow agents are built from three main components:

•      Triggers – Define the events or conditions that start the workflow (schedule-based, event-based, or manual)

•      Logic – Configure advanced decision-making, including loops and branching through conditional rules

•      Agents – Add and arrange specialized agents to create a complete workflow with intelligent hand-offs

Real-World Workflow Example

Shafqat Islam, Optimizely's CMO, illustrated the power of workflow agents with this example: "Imagine a workflow that runs autonomously every day or every week that can look at the goal you have set, comes up with test ideas, builds two or three experiments, runs them, picks the winner, and pushes that to production."

That's a six or seven-step workflow that Opal can run autonomously versus a marketer manually pushing buttons. The same applies to CMS tasks – automatically checking accessibility across all pages and pushing updates live.

Tools: The Foundation of Agent Capabilities

Tools are what enable Opal to take action. Each tool performs a specific task – creating a campaign, uploading files, generating images, or pulling data from connected systems. You can have up to 128 tools enabled across an Opal instance.

System Tools

Built and maintained by Optimizely, system tools provide core functionality across the platform:

•      Create campaigns, tasks, and content in CMP

•      Search the web and generate images

•      Analyze data from Google Analytics (GA4)

•      Process Microsoft Office files up to 50MB using smart chunking

Custom Tools

The real power comes from extending Opal with custom tools. Using the Opal SDK (available in Python, JavaScript/TypeScript, and C#), organizations can connect Opal to their internal systems or third-party services.

Custom tools can be self-hosted or deployed through Optimizely Connect Platform (OCP), giving you flexibility in how you extend the platform.

Third-Party Integrations

Opal makes it simple to extend agent orchestration beyond the platform itself. You can connect to existing data sources, plug in third-party tools, and even access Opal directly through Slack – bringing AI into the conversations where work already happens.

What This Means for Marketing Teams

The shift from AI as a feature to AI as an orchestration platform represents a fundamental change in how marketing teams can operate. Rather than incremental efficiency gains, Optimizely is positioning Opal as a path to exponential scale.

Key benefits include:

•      Faster AI deployment at lower cost – Pre-built agents and no-code workflow builders reduce time to value

•      Brand consistency – Instructions and context ensure AI outputs align with your brand voice and guidelines

•      True automation – Workflows can run autonomously, freeing teams to focus on strategy rather than execution

•      Unified platform experience – Opal is embedded across Optimizely One, providing consistent AI assistance wherever you work

Looking Ahead

For organizations already invested in the Optimizely ecosystem, the path to leveraging these capabilities is clear. For those evaluating options, Opal represents a compelling vision of what AI-powered marketing operations could look like – not as a collection of point solutions, but as an integrated orchestration layer that can genuinely transform how work gets done.

The question isn't whether AI will transform marketing operations – it's whether your organization is ready to move from incremental improvements to exponential scale.