Optimizely Mark overview
Mark (Optimizely Agent Platform) is an agent orchestration platform that helps you work smarter across Optimizely One. Whether you create content, manage experiments, or analyze data, you can use Mark to automate tasks, surface insights, and guide decision-making. Mark uses a flexible Mark credits system of AI agents. Agents are software components that understand your intent and help you reach your goal.
See Optimizely Opal Essentials in Optimizely Academy for information.
AI background
AI refers to software that performs tasks typically requiring human reasoning, such as learning from data, making decisions, or solving problems. These systems identify patterns, generate content, and offer insights. They use large language models (LLMs) and other algorithms trained on vast datasets.
Advances in computing power, data availability, and model design have changed how people use AI across industries. Optimizely embeds AI into the platform to help you move faster, automate repetitive work, and make more informed decisions.
AI agents
AI agents are software systems that act on your behalf. Agents analyze information, interact with the platform or other tools, and adapt their behavior based on feedback. Unlike traditional tools that only respond to direct input, agents manage goals and take steps to accomplish them.
Optimizely classifies AI agents into the following three broad types:
- Simple assistants – Help with straightforward tasks such as generating content, formatting it into a structure, or offering ideas based on a prompt.
- Specialized agents – Focus on a domain or tool. These agents use expert-level context to analyze data, predict outcomes, or recommend next steps.
- Workflow and autonomous agents – Coordinate and perform multi-step processes, often by integrating with other tools. These agents handle tasks across systems and adapt workflows over time.
These agents improve productivity by reducing manual work, increasing accuracy, and providing tailored experiences. They also support scalability, helping teams achieve more without growing headcount.
Mark
Mark works across Optimizely One, where it responds to natural language questions, generates insights, and helps you complete tasks faster.
Mark uses Mark credits.
What you can do with Mark
Most interactions happen through Mark chat, which is available across the Optimizely One platform in the global navigation bar and on mark.optimizely.com.
Whether you are launching a campaign or fine-tuning an experiment, Mark adapts to your role and needs. Use natural language to ask questions, generate content, or request help. Mark responds in real-time and remembers threads so you can pick up where you left off.
The following are some examples of what Mark can help with:
- Use tools – Ask Mark to complete tasks using tools, such as creating a campaign, drafting content, analyzing images, or suggesting flag variations. See Tools overview.
- Analyze files – Upload a document, image, or spreadsheet and ask questions about it. See Mark chat.
- Answer questions – Ask Mark about Optimizely features, capabilities, or concepts. Mark explains terms, walks you through configuration steps, and points you to relevant documentation.
See Mark and AI features for a list of Mark features.
How Mark works
Behind the scenes, Mark follows this sequence each time you send a prompt:
- Your prompt – You enter a request in natural language. Your prompt can be a simple question, a complex task, or a creative brief. Mark analyzes your intent, extracting key information to guide its next steps.
- Skills – These foundational guidelines define Mark's persona, approach, tool usage, evaluation criteria, and response formatting, influencing every step of its operation.
- Contextual intelligence – Mark adds context using content knowledge, Optimizely platform data, and your workspace context. This ensures Mark's responses are informed by your specific environment and goals.
- Strategic tool selection – Mark selects the right tools for your task based on your request and the enriched context. Tools let Mark perform a wide range of tasks from data analysis to content generation.
- Agent input generation and validation – Mark collects the specific inputs (parameters) it needs to run a tool or agent. For example, a search_query for a web search, a start_date and end_date for a report, or recipient_email_addresses for an email. Mark validates these inputs before continuing:
- If the input is valid – Mark proceeds to the LLM interaction.
- If the input is invalid – Mark goes directly to step 7 (Execute and refine) and asks you for clarification. This prevents unnecessary LLM engagement.
- LLM interaction – Mark uses advanced AI models to process complex information, generate creative content, and derive actionable insights. This is where raw data transforms into intelligent solutions. Mark uses foundational LLMs through business accounts. Your data is never used to train the model or shared across customers.
- Execute and refine – Mark executes the planned actions (calling agents and tools as needed), and iterates. If the initial output is not perfect, Mark re-evaluates. This continuous feedback loop ensures optimal results and measurable impact. If clarification is needed, Mark asks you for more information.
- Tailored response – Mark delivers a clear, concise, and actionable response. Mark often presents the response in a canvas or in the Mark UI, where you can interact with and manage generated content or executed actions.
Benefits of Mark and AI agents
Mark helps you with the following:
- Work faster – Eliminate repetitive tasks and reduce manual configuration.
- Scale content and insights – Get more done with fewer handoffs.
- Improve personalization – Tailor output based on your brand, audience, and goals.
- Make informed decisions – Use data-backed recommendations without digging through dashboards.
- Simplify complexity – Let Mark manage workflows so you can focus on strategy.
See Mark and AI features for a list of all available Mark, generative AI, and machine learning features.
Data residency
Optimizely hosts Mark in US and EU regions. You select your region on your order form when you contract with Optimizely, and Optimizely provisions your data to the corresponding regional instance.
The app URL is the same for all customers (mark.optimizely.com). Optimizely routes you to the correct regional instance behind the scenes, so you do not need to configure anything.
Where is data stored?
Customer data — both inputs and outputs — is stored long-term in either North America or the EU (Belgium) on Optimizely-managed infrastructure. Customers choose their preferred region at the time of purchase. EU storage covers the region as a whole; country-level selection is not available.
How does data processing work?
When a prompt is submitted, Mark routes it to the underlying LLM — such as Gemini or Claude — for processing. This happens through global endpoints for compute availability and performance, so a prompt may be processed in any region. The LLM does not retain your data. The output is returned directly to your Optimizely-managed infrastructure, in the region you selected.
See Default model provider for information on selecting an LLM for your organization.
Next steps
- Get started with Mark for administrators – Manage and configure users, skills, tools, agents, and connected product instances in the Mark app.
- Get started with Mark for agent builders – Create skills, build agents, and access the Agent Directory.
- Get started with Mark for users – Discover how to access and use Mark.
- Mark credits – Learn how credits work and how Mark calculates usage.
- Mark chat – See how to interact with Mark, upload files, and return to past threads.
- Mark and AI features – Review a full list of Mark and included AI features for specific products.
Note
If you use Opti ID, administrators can turn off generative AI in the Opti ID Admin Center. See Turn generative AI off across Optimizely applications.