--- title: Agent overview slug: agent-platform/docs/agent-overview docTags: createdAt: 2026-08-12T15:37:05.000Z --- Agents are like intelligent assistants within Agent Platform. They use natural language prompts and tools to complete tasks on your behalf. Each agent has a specific purpose, whether that is generating content, analyzing data, or automating workflows. Agents combine [tools](docId:-j4zNs-l3kIKD0n58bVPt) (actions they can take) and [skills](docId:1HTjxW2J-ZOkYEZMwnasU) (guidelines that define their behavior) to complete tasks without requiring you to manage the technical details. Think of agents as skilled team members. With agents, you can make simple requests like, *Draft a social post for our product launch* or *Summarize experiment results*. Mark selects the right tools and follows the right skills to complete the task. For a visual walkthrough of agents, watch the [video overview](https://files.marketing.cmp.optimizely.com/download/assets/Agents+Overview_Nikita.mp4/fcba2a6e89ed11f09ddf6e823a9dee38?attachment=false). ## Why agents matter Agents make Mark flexible and extensible. Describe what you want in natural language. The agent selects the right tools and handles the technical details. The agent does the rest, ensuring tasks are completed consistently and accurately. Without agents, you would need to complete the following tasks manually: - Configure tools for each task. - Write detailed prompts every time you make a request. - Understand the underlying systems or APIs. With agents, you get repeatable workflows that you can customize to your organization's needs. ## Types of agents ### Agents in the Agent Directory Agents in the [Agent Directory](docId\:Xlo4rbSLRP_fxF5NlMLgG) are pre-built by Optimizely and come ready to use. These agents handle common tasks like reviewing content for tone and style, generating competitive insights, or providing customer support responses. Browse and add default agents from the **Agent Directory**. Each listing includes the following: - **Description** – What the agent does and when to use it. - **Tools included** – The built-in tools the agent can call. - **Configuration options** – Settings you can adjust for your organization. Agents from the Agent Directory are the fastest way to get started because they are ready to run without custom development. A curated subset of these agents also displays in the [Agent Library](docId\:zCG_0PQFrNkKNv4sZ_Gou), where end users browse and launch them. See [Agents in the Agent Directory](docId\:mCCl_y-hNkhT-uKIXvReF) for a list of available agents. ### Specialized agents [Specialized agents](docId\:WyA1gsk6tWr84Hum7a1Nx) are custom-built for your organization. They are designed to execute specific, well-defined tasks using your own tools, data, and skills. Specialized agents support the following capabilities: - Build tailored workflows for unique business processes. - Control which tools the agent can access. - Fine-tune inputs, outputs, and behavior using prompt templates. Specialized agents are ideal when you need precise control or want to automate tasks that default agents do not cover. See [Create specialized agents](docId\:tZPJWDyE1GSB-epLsUh5S) for details on how to design and configure your own. ### Workflow agents :::hint{type="info"} **Note** [Workflow agents](docId:2HD0fUsIsfdek516VtN9l) are in private GA and not available in all Mark instances. ::: Workflow agents are designed to process natural language input to accomplish complex workflows. They can automate and streamline multi-step processes within the Optimizely platform. For example, a workflow agent named "Industry Marketer Workflow" can coordinate marketing efforts by integrating with specialized agents such as a case study agent. See [Create a workflow agent](docId\:KWaI2TBwD38sN7p74exoy) for details on how to design and configure your workflow agent. ## Where agents are used Agents work behind the scenes in several areas of Mark. Most collaborators experience agents through [Mark chat](docId\:dtNBEttEGVvkhNkwNypLM) without needing to know which agents are active. ### Mark chat In Mark chat, talk to agents directly using natural language. Mark automatically selects the right default agent based on your request. If multiple agents could apply, Mark asks clarifying questions to choose the best one. Example requests include the following: - *Summarize the results of our last experiment* – Uses a default results analysis agent. - *Create a product announcement post* – Uses a default content generation agent. - *Check if there are any broken links on our website* – Uses a specialized agent with a custom web-check tool. See [Mark prompts](docId\:dqpUGpG_keS_Xab2LcYLM) for information. ### Agent Library The Agent Library is a browsable catalog of agents on the [Mark chat](docId\:dtNBEttEGVvkhNkwNypLM) page. Browse agents by category, open an agent to see what it does, and launch it into your chat. The library displays a curated subset of the agents in the Agent Directory. Find the right agent and run it without installation or configuration. See [Agent Library](docId\:zCG_0PQFrNkKNv4sZ_Gou). ### Automated workflows Workflow agents can run automatically when certain [triggers](docId\:pYeMYR_1QCBBfc6rGEwlR) are met. The following examples show agents that trigger on a condition: - An agent that tags uploaded files with metadata when they are added to a repository. - An agent that generates draft help responses whenever a new support ticket is created. ## Next steps - [Browse the Agent Directory](docId\:Xlo4rbSLRP_fxF5NlMLgG) to see available default agents. - [Create a specialized agent](docId\:tZPJWDyE1GSB-epLsUh5S) or [Create a workflow agent](docId\:KWaI2TBwD38sN7p74exoy). - [Manage specialized agents](docId\:cFB_b7jzgV6k0Ry_cVXSS) or [Manage workflow agents](docId\:MdEwF6vrQCh0zC-A5XqX2) to update the configuration of your agents. :::hint{type="info"} **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](docId\:X5U0kKD2azV5cIdNqwGNj). :::