Skip to Content

Ollama Prompt activity in Jitterbit Studio

Introduction

An Ollama Prompt activity, using its Ollama connection, sends a prompt to the specified Ollama model and is intended to be used as a target to consume data in an operation.

Each prompt and model response is stored in the conversation history for the chat ID in memory on the agent, so later prompts that use the same chat ID include the earlier conversation as context. The conversation history is cleared when the agent restarts.

Create an Ollama Prompt activity

An instance of an Ollama Prompt activity is created from an Ollama connection using its Prompt activity type.

To create an instance of an activity, drag the activity type to the design canvas or copy the activity type and paste it on the design canvas. For details, see Creating an activity instance in Component reuse.

An existing Ollama Prompt activity can be edited from these locations:

Configure an Ollama Prompt activity

Follow these steps to configure an Ollama Prompt activity:

Step 1: Enter a name and specify settings

In this step, provide a name for the activity, select a model, and specify a prompt and optional settings. Each user interface element of this step is described below.

Tip

Fields with a variable icon support using global variables, project variables, and Jitterbit variables. Begin either by typing an open square bracket [ into the field or by clicking the variable icon to display a menu listing existing variables to choose from.

  • Endpoint menu: If you have multiple endpoints of the same connector type configured, a menu at the top of the screen displays the current endpoint name. Click the menu to switch to a different endpoint. For more information, see Change the assigned endpoint in Configuration screens.

    • Edit endpoint: Appears when you hover over the current endpoint name. Click to edit the currently selected endpoint's connection configuration.

  • Name: Enter a name to identify the activity. The name must be unique for each Ollama Prompt activity and must not contain forward slashes / or colons :.

  • Chat ID: Enter a chat ID to retrieve context from a previous conversation. If empty and not provided in the request transformation, a new chat ID is generated and provided in the activity response.

  • Choose option: Specify the model selection method, either List all models or Enter model identifier.

  • List all models: Select this option to choose from a list of models retrieved from the Ollama endpoint:

    • Select a model: This section displays models available in the Ollama endpoint. Only models that support text completion are listed. DeepSeek models are excluded from the list.

      • Selected model: After a model is selected, it is listed here.

      • Search: Enter any column's value into the search box to filter the list of models. The search is not case-sensitive. If models are already displayed within the table, the table results are filtered in real time with each keystroke. To reload models from the endpoint when searching, enter search criteria and then refresh, as described below.

      • Refresh: Click the refresh icon or the word Refresh to reload models from the Ollama endpoint. This may be useful if models have been pulled to a self-hosted Ollama server. This action refreshes all metadata used to build the table of models displayed in the configuration.

      • Selecting a model: Within the table, click anywhere on a row to select a model. Only one model can be selected. The information available for each model is fetched from the Ollama endpoint:

        • Display Name: The name of the model.

        • ID: The ID of the model used in API requests, including its tag (for example, llama3.2:latest).

        • Modified: The date and time the model was last modified.

      Tip

      If the table does not populate with available models, the Ollama connection may not be successful. Ensure you are connected by reopening the connection and retesting the credentials. For a self-hosted Ollama server, also confirm that at least one model has been pulled to the server.

      Alternatively, the model can be specified in the Model ID field or provided in the request transformation. If a model is specified in multiple places, this order of precedence is followed:

      1. A model provided in the request transformation overrides all fields below.
      2. A model specified in the Model ID field overrides the remaining field below, even if List all models is selected.
      3. A model selected in the Select a model list has the least precedence.

  • Enter model identifier: Select this option to manually enter the ID of a model:

    • Model ID: Enter the model ID (for example, llama3.2). For Ollama Cloud, use a model name listed in Ollama's cloud models. For a self-hosted server, the model must already be pulled to the server.

  • Prompt: Enter the text to prompt the model with.

  • Optional settings: Click to expand additional optional settings:

    • Enable thinking (reasoning models): Select to enable thinking, which allows a reasoning model to reason through a problem before generating a response. The model's reasoning is returned in the thinking field of the response. This option is disabled by default. When this option is cleared, thinking is turned off, including for models that otherwise reason by default.

      Note

      Thinking is supported only by reasoning models. Refer to the Ollama thinking documentation to confirm whether the selected model supports thinking.

    • Keep model loaded: Select how long the model stays loaded in memory after the request completes (see How do I keep a model loaded in memory or make it unload immediately?). Keeping a model loaded avoids load time on subsequent requests but continues to use memory on the Ollama server:

      • 5 minutes: (Default.) Keeps the model loaded for 5 minutes.
      • 30 minutes: Keeps the model loaded for 30 minutes.
      • 1 hour: Keeps the model loaded for 1 hour.
      • Keep loaded indefinitely: Keeps the model loaded until the Ollama server is restarted or the model is unloaded.
      • Unload after each call: Unloads the model immediately after the response is returned.

    • Randomness and diversity:

      • Temperature: Enter a temperature parameter. Higher values produce more varied responses, and lower values produce more deterministic responses. This parameter's value can range between 0 and 2, for example: 0.8.
      • Top P: Enter a Top P parameter for nucleus sampling, which restricts token selection to the smallest set of tokens whose cumulative probability reaches this value. This parameter's value can range between 0 and 1, for example: 0.9.
      • Top K: Enter a Top K parameter, which restricts token selection to the specified number of highest-probability tokens, for example: 40.
      • Seed: Enter a seed parameter. Using the same seed value with the same prompt reproduces the same output. Leave this field empty for nondeterministic output.

      When a field in this group is empty, the model's default value is used.

    • Context and length:

      • Context Window: Enter the size of the context window in tokens, which limits the combined length of the prompt, the conversation history, and the generated output. The default is 4096. The maximum value depends on the selected model and the memory available on the Ollama server (see How can I specify the context window size?).
      • Max Output Tokens: Enter the maximum number of tokens to generate in the response. The default is 2000. Enter -1 to generate without a limit.

    • Continue on error: Select to continue the activity execution if an error is encountered for a dataset in a batch request. If any errors are encountered, they are written to the operation log and returned in the errors array of the activity response.

  • Save & Exit: If enabled, click to save the configuration for this step and close the activity configuration.

  • Next: Click to temporarily store the configuration for this step and continue to the next step. The configuration will not be saved until you click the Finished button on the last step.

  • Discard changes: After making changes, click to close the configuration without saving changes made to any step. A message asks you to confirm that you want to discard changes.

Step 2: Review the data schemas

Any request or response schemas are displayed. Each user interface element of this step is described below.

  • Data schema: These data schemas are inherited by adjacent transformations and are displayed again during transformation mapping.

    Note

    Data supplied in a transformation takes precedence over the activity configuration.

    The request schema accepts one or more records in the schema array, and each record is sent to the model as a separate request. In addition to overriding activity configuration settings such as chatId, modelId, prompt, and the optional generation settings, a record can provide these fields:

    • system: A system instruction for the conversation. A system instruction provided in the request replaces any system instruction stored for the chat ID. When system is empty, the stored system instruction continues to be used.
    • tool_calls and tool_response: The tool calls returned by a previous model response and the results of running those tools, used to return tool results to the model in a tool-calling workflow.

    The Ollama connector uses the Ollama API. Refer to the chat endpoint documentation for information on the schema nodes and fields.

  • Refresh: Click the refresh icon or the word Refresh to regenerate schemas from the Ollama endpoint. This action also regenerates a schema in other locations throughout the project where the same schema is referenced, such as in an adjacent transformation.

  • Back: Click to temporarily store the configuration for this step and return to the previous step.

  • Finished: Click to save the configuration for all steps and close the activity configuration.

  • Discard changes: After making changes, click to close the configuration without saving changes made to any step. A message asks you to confirm that you want to discard changes.

Next steps

After configuring an Ollama Prompt activity, complete the configuration of the operation by adding and configuring other activities, transformations, or scripts as operation steps. You can also configure the operation settings, which include the ability to chain operations together that are in the same or different workflows.

Menu actions for an activity are accessible from the project pane and the design canvas. For details, see Activity actions menu in Connector basics.

Ollama Prompt activities can be used as a target with these operation patterns:

To use the activity with scripting functions, write the data to a temporary location and then use that temporary location in the scripting function.

When ready, deploy and run the operation and validate behavior by checking the operation logs.