> For the complete documentation index, see [llms.txt](https://academy.dnanexus.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://academy.dnanexus.com/mlaccelerator/automlassistant/getting-started-with-automl-assistant.md). # Getting Started with AutoML Assistant Please refer to the [Usage and Limitations](/mlaccelerator/automlassistant/usage-and-limitations.md) section before using AutoML Assistant. A license is required to use AutoML Assistant. For more information, please contact DNAnexus Sales via . ### Initial Setup Requirements Before using the AutoML Assistant, ensure these access and technical requirements are met: You have a license for AutoMl Assistant AND User Access and Permissions: * You have access to the Apollo Dataset. * A configuration file provided in your project. * Large Language Model Access * Claude 4.0 LLM and Amazon Titan Text Embeddings v2 models must be enabled in Amazon Bedrock. ### Open AutoML Assistant Interface AutoML Assistant is integrated directly into the AI/ML Accelerator package to streamline your machine learning workflows. To get started, you must first launch an AI/ML Accelerator job. Depending on your platform subscription (Titan or Apollo), ensure you run the appropriate version of the app: * [AI/ML Accelerator for Titan](https://platform.dnanexus.com/panx/tool/app/ml_jupyterlab_titan): Enables conversational analysis with standard file objects (e.g., CSV, TSV, Parquet, VCF, etc.) * [AI/ML Accelerator for Apollo](https://platform.dnanexus.com/panx/tool/app/ml_jupyterlab_apollo): Enables conversational analysis with both file objects and Apollo cohorts Once the AI/ML Accelerator job is up and running, click the button on the homepage to open the conversational interface.
Figure 1. AutoML Assistant is one solution in the AI/ML Accelerator Package ### Revisit a Conversation or Start a New One After clicking on the AutoML Assistant icon, you will be directed to the Welcome page as below. You can either revisit a former conversation by clicking on one tab on the ‘Conversations’ sidebar or start a new conversation by clicking on the ‘Start a conversation’ button.
Figure 2. In the Welcome page, you can either revisit a conversation or start a new conversation ### Select a Data Format to Begin AutoML Assistant supports multiple data inputs to start your analysis. You can begin with Apollo cohorts or Tabular data files (starting with Parquet format), depending on your use case. Choose the format that best fits your data source to initiate a new conversation and begin building your ML workflow. #### Start a New Conversation with Apollo Cohorts At least one Apollo cohort needs to be selected to initiate a new conversation. This step defines the dataset on which your analysis will be performed. * Select from the list of available Apollo cohorts in your project which you launched the ML JupyterLab job (the cohorts are grouped by the original Apollo dataset). * Confirm your selection and click “Continue” to get started. Note: AutoML Assistant tailors its answers and suggestions based on the structure and content of the selected cohort(s).
#### Start a New Conversation with Tabular Data Files You can also start a conversation by selecting a tabular data file (e.g., Parquet) from your project. Once selected, AutoML Assistant will load and interpret the dataset, allowing you to explore the data, perform statistical analysis, and build ML models through natural language prompts—just as you would with Apollo cohorts. ### Start Your Conversation with AutoML Assistant With the cohort(s) selected, you are now ready to begin the conversation\*. You can start by typing your objective or request or following an AutoML Assistant’s suggestion in natural English language. The created conversations will be persisted and retained on your DNAnexus project. \*AutoML Assistant uses the conversations as the context for the LLM model. DNAnexus does not take your conversation data to train or fine-tune the model.