Access Vertex AI models from a workflow

Generative AI on Vertex AI (also known as genAI or gen AI) gives you access to Google's generative AI models for multiple modalities (text, code, images, speech). You can test and tune these large language models (LLM), and then deploy them for use in your AI-powered applications. For more information, see the Overview of Generative AI on Vertex AI.

Vertex AI has a variety of generative AI foundation models that are accessible through an API, including the models used in the following examples:

  • Gemini Pro is designed to handle natural language tasks, multiturn text and code chat, and code generation.
  • Gemini Pro Vision supports multimodal prompts. You can include text, images, and video in your prompt requests and get text or code responses.
  • Pathways Language Model 2 (PaLM 2) for text is fine-tuned for language tasks such as classification, summarization, and entity extraction.

Each model is exposed through a publisher endpoint that's specific to your Google Cloud project so there's no need to deploy the foundation model unless you need to tune it for a specific use case. You can send a prompt to the publisher endpoint. A prompt is a natural language request sent to an LLM to elicit a response back.

This tutorial demonstrates workflows that generate responses from Vertex AI models by sending text prompts to the publisher endpoints using either a Workflows connector or an HTTP POST request. For more information, see the Vertex AI API connector overview and Make an HTTP request.

Note that you can deploy and run each workflow independently of each other.

Objectives

In this tutorial, you will do the following:

  1. Enable the Vertex AI and Workflows APIs, and grant the Vertex AI User (roles/aiplatform.user) role to your service account. This role allows access to most Vertex AI capabilities. For more information about setting up Vertex AI, see Get set up on Google Cloud.
  2. Deploy and run a workflow that prompts a Vertex AI model (Gemini Pro Vision) to describe an image that is publicly available through Cloud Storage. For more information, see Make data public.
  3. Deploy and run a workflow that loops through a list of countries in parallel and prompts a Vertex AI model (Gemini Pro) to generate and return the histories of the countries. Using parallel branches allows you to reduce the total execution time by starting the calls to the LLM at the same time and waiting for all of them to complete before combining the results. For more information, see Execute workflow steps in parallel.
  4. Deploy a workflow similar to the preceding one; however, prompt a Vertex AI model (PaLM 2 for text) to generate and return the histories of the countries. For more information about how to choose a model, see Model information.
  5. Deploy a workflow that can summarize a large document. Because there is a limit to the context window which sets how far back the model looks during training (and for forecasts), the workflow divides the document into smaller parts, and then prompts a Vertex AI model (Gemini Pro) to summarize each part in parallel. For more information, see Summarization prompts and Forecast horizon, context window, and forecast window.

Costs

In this document, you use the following billable components of Google Cloud:

To generate a cost estimate based on your projected usage, use the pricing calculator. New Google Cloud users might be eligible for a free trial.

When you finish the tasks that are described in this document, you can avoid continued billing by deleting the resources that you created. For more information, see Clean up.

Before you begin

Before trying out the examples in this tutorial, ensure that you have completed the following.

Console

  1. Sign in to your Google Cloud account. If you're new to Google Cloud, create an account to evaluate how our products perform in real-world scenarios. New customers also get $300 in free credits to run, test, and deploy workloads.

  2. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

    Go to project selector

  3. Make sure that billing is enabled for your Google Cloud project.

  4. Enable the Vertex AI and Workflows APIs.

    Enable the APIs

  5. Create a service account:

    1. In the Google Cloud console, go to the Create service account page.

      Go to Create service account
    2. Select your project.

    3. In the Service account name field, enter a name. The Google Cloud console fills in the Service account ID field based on this name.

      In the Service account description field, enter a description. For example, Service account for quickstart.

    4. Click Create and continue.

    5. Grant the Vertex AI > Vertex AI User role to the service account.

      To grant the role, find the Select a role list, then select Vertex AI > Vertex AI User.

    6. Click Continue.

    7. Click Done to finish creating the service account.

  6. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

    Go to project selector

  7. Make sure that billing is enabled for your Google Cloud project.

  8. Enable the Vertex AI and Workflows APIs.

    Enable the APIs

  9. Create a service account:

    1. In the Google Cloud console, go to the Create service account page.

      Go to Create service account
    2. Select your project.

    3. In the Service account name field, enter a name. The Google Cloud console fills in the Service account ID field based on this name.

      In the Service account description field, enter a description. For example, Service account for quickstart.

    4. Click Create and continue.

    5. Grant the Vertex AI > Vertex AI User role to the service account.

      To grant the role, find the Select a role list, then select Vertex AI > Vertex AI User.

    6. Click Continue.

    7. Click Done to finish creating the service account.

gcloud

  1. Sign in to your Google Cloud account. If you're new to Google Cloud, create an account to evaluate how our products perform in real-world scenarios. New customers also get $300 in free credits to run, test, and deploy workloads.
  2. Install the Google Cloud CLI.

  3. To initialize the gcloud CLI, run the following command: 

    gcloud init

  4. Create or select a Google Cloud project.

    • Create a Google Cloud project:

      gcloud projects create PROJECT_ID

      Replace PROJECT_ID with a name for the Google Cloud project you are creating.

    • Select the Google Cloud project that you created:

      gcloud config set project PROJECT_ID

      Replace PROJECT_ID with your Google Cloud project name.

  5. Make sure that billing is enabled for your Google Cloud project.

  6. Enable the Vertex AI and Workflows APIs: 

    gcloud services enable aiplatform.googleapis.com workflows.googleapis.com

  7. Set up authentication:

    1. Create the service account: 

      gcloud iam service-accounts create SERVICE_ACCOUNT_NAME

      Replace SERVICE_ACCOUNT_NAME with a name for the service account.

    2. Grant the roles/aiplatform.user IAM role to the service account: 

      gcloud projects add-iam-policy-binding PROJECT_ID --member="serviceAccount:SERVICE_ACCOUNT_NAME@PROJECT_ID.iam.gserviceaccount.com" --role=roles/aiplatform.user

      Replace the following:

      • SERVICE_ACCOUNT_NAME: the name of the service account
      • PROJECT_ID: the project ID where you created the service account
    8.
  8. Install the Google Cloud CLI.

  9. To initialize the gcloud CLI, run the following command: 

    gcloud init

  10. Create or select a Google Cloud project.

    • Create a Google Cloud project:

      gcloud projects create PROJECT_ID

      Replace PROJECT_ID with a name for the Google Cloud project you are creating.

    • Select the Google Cloud project that you created:

      gcloud config set project PROJECT_ID

      Replace PROJECT_ID with your Google Cloud project name.

  11. Make sure that billing is enabled for your Google Cloud project.

  12. Enable the Vertex AI and Workflows APIs: 

    gcloud services enable aiplatform.googleapis.com workflows.googleapis.com

  13. Set up authentication:

    1. Create the service account: 

      gcloud iam service-accounts create SERVICE_ACCOUNT_NAME

      Replace SERVICE_ACCOUNT_NAME with a name for the service account.

    2. Grant the roles/aiplatform.user IAM role to the service account: 

      gcloud projects add-iam-policy-binding PROJECT_ID --member="serviceAccount:SERVICE_ACCOUNT_NAME@PROJECT_ID.iam.gserviceaccount.com" --role=roles/aiplatform.user

      Replace the following:

      • SERVICE_ACCOUNT_NAME: the name of the service account
      • PROJECT_ID: the project ID where you created the service account
    14.

Deploy a workflow that describes an image (Gemini Pro Vision)

Deploy a workflow that uses a connector method (generateContent) to make a request to a Gemini Pro Vision publisher endpoint. The method provides support for content generation with multimodal inputs.

The workflow provides a text prompt and the URI of an image that is publicly available in a Cloud Storage bucket. You can view the image and, in the Google Cloud console, you can view the object details.

The workflow returns a description of the image from the model's generated response.

For more information about the HTTP request body parameters used when prompting the LLM, and the response body elements, see the Gemini API reference.

Console

  1. In the Google Cloud console, go to the Workflows page.

    Go to Workflows

  2. Click Create.

  3. Enter a name for the new workflow: describe-image.

  4. In the Region list, select us-central1 (Iowa).

  5. For the Service account, select the service account you previously created.

  6. Click Next.

  7. In the workflow editor, enter the following definition for your workflow:

    main:
    params: [args]
    steps:
    - init:
        assign:
            - project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
            - location: "us-central1"
            - model: "gemini-1.0-pro-vision"
            - text_combined: ""
    - ask_llm:
        call: googleapis.aiplatform.v1.projects.locations.endpoints.generateContent
        args:
            model: ${"projects/" + project + "/locations/" + location + "/publishers/google/models/" + model}
            region: ${location}
            body:
                contents:
                    role: user
                    parts:
                    - fileData:
                        mimeType: image/jpeg
                        fileUri: ${args.image_url}
                    - text: Describe this picture in detail
                generation_config:
                    temperature: 0.4
                    max_output_tokens: 2048
                    top_p: 1
                    top_k: 32
        result: llm_response
    - return_result:
        return:
            image_url: ${args.image_url}
            image_description: ${llm_response.candidates[0].content.parts[0].text}

  8. Click Deploy.

gcloud

  1. Create a source code file for your workflow:

    touch describe-image.yaml

  2. In a text editor, copy the following workflow to your source code file:

    main:
    params: [args]
    steps:
    - init:
        assign:
            - project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
            - location: "us-central1"
            - model: "gemini-1.0-pro-vision"
            - text_combined: ""
    - ask_llm:
        call: googleapis.aiplatform.v1.projects.locations.endpoints.generateContent
        args:
            model: ${"projects/" + project + "/locations/" + location + "/publishers/google/models/" + model}
            region: ${location}
            body:
                contents:
                    role: user
                    parts:
                    - fileData:
                        mimeType: image/jpeg
                        fileUri: ${args.image_url}
                    - text: Describe this picture in detail
                generation_config:
                    temperature: 0.4
                    max_output_tokens: 2048
                    top_p: 1
                    top_k: 32
        result: llm_response
    - return_result:
        return:
            image_url: ${args.image_url}
            image_description: ${llm_response.candidates[0].content.parts[0].text}

  3. Deploy the workflow by entering the following command:

    gcloud workflows deploy describe-image \
    --source=describe-image.yaml \
    --location=us-central1 \
    --service-account=SERVICE_ACCOUNT_NAME@PROJECT_ID.iam.gserviceaccount.com

Execute the workflow

Executing a workflow runs the current workflow definition associated with the workflow.

Console

  1. In the Google Cloud console, go to the Workflows page.

    Go to Workflows

  2. On the Workflows page, select the describe-image workflow to go to its details page.

  3. On the Workflow details page, click Execute.

  4. For the Input, enter the following:

    {"image_url":"gs://generativeai-downloads/images/scones.jpg"}

  5. Click Execute again.

  6. View the results of the workflow in the Output pane.

    The output should be similar to the following:

    {
  "image_description": "There are three pink peony flowers on the right side of the picture[]...]There is a white napkin on the table.",
  "image_url": "gs://generativeai-downloads/images/scones.jpg"
}

gcloud

  1. Open a terminal.

  2. Execute the workflow:

    gcloud workflows run describe-image \
    --data='{"image_url":"gs://generativeai-downloads/images/scones.jpg"}'

    The execution results should be similar to the following:

      Waiting for execution [258b530e-a093-46d7-a4ff-cbf5392273c0] to complete...done.
  argument: '{"image_url":"gs://generativeai-downloads/images/scones.jpg"}'
  createTime: '2024-02-09T13:59:32.166409938Z'
  duration: 4.174708484s
  endTime: '2024-02-09T13:59:36.341118422Z'
  name: projects/1051295516635/locations/us-central1/workflows/describe-image/executions/258b530e-a093-46d7-a4ff-cbf5392273c0
  result: "{\"image_description\":\"The picture shows a rustic table with a white surface,\
    \ on which there are several scones with blueberries, as well as two cups of coffee\
    [...]
    \ on the table. The background of the table is a dark blue color.\",\"image_url\"\
    :\"gs://generativeai-downloads/images/scones.jpg\"}"
  startTime: '2024-02-09T13:59:32.166409938Z'
  state: SUCCEEDED

Deploy a workflow that generates country histories (Gemini Pro)

Deploy a workflow that loops through an input list of countries in parallel and uses a connector method (generateContent) to make a request to a Gemini Pro publisher endpoint. The method provides support for content generation with multimodal inputs.

The workflow returns the country histories generated by the model, combining them in a map.

For more information about the HTTP request body parameters used when prompting the LLM, and the response body elements, see the Gemini API reference.

Console

  1. In the Google Cloud console, go to the Workflows page.

    Go to Workflows

  2. Click Create.

  3. Enter a name for the new workflow: gemini-pro-country-histories.

  4. In the Region list, select us-central1 (Iowa).

  5. For the Service account, select the service account you previously created.

  6. Click Next.

  7. In the workflow editor, enter the following definition for your workflow:

    main:
    params: [args]
    steps:
    - init:
        assign:
            - project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
            - location: "us-central1"
            - model: "gemini-1.0-pro"
            - histories: {}
    - loop_over_countries:
        parallel:
            shared: [histories]
            for:
                value: country
                in: ${args.countries}
                steps:
                    - ask_llm:
                        call: googleapis.aiplatform.v1.projects.locations.endpoints.generateContent
                        args:
                            model: ${"projects/" + project + "/locations/" + location + "/publishers/google/models/" + model}
                            region: ${location}
                            body:
                                contents:
                                    role: "USER"
                                    parts:
                                        text: ${"Can you tell me about the history of " + country}
                                generation_config:
                                    temperature: 0.5
                                    max_output_tokens: 2048
                                    top_p: 0.8
                                    top_k: 40
                        result: llm_response
                    - add_to_histories:
                        assign:
                            - histories[country]: ${llm_response.candidates[0].content.parts[0].text}
    - return_result:
        return: ${histories}

  8. Click Deploy.

gcloud

  1. Create a source code file for your workflow:

    touch gemini-pro-country-histories.yaml

  2. In a text editor, copy the following workflow to your source code file:

    main:
    params: [args]
    steps:
    - init:
        assign:
            - project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
            - location: "us-central1"
            - model: "gemini-1.0-pro"
            - histories: {}
    - loop_over_countries:
        parallel:
            shared: [histories]
            for:
                value: country
                in: ${args.countries}
                steps:
                    - ask_llm:
                        call: googleapis.aiplatform.v1.projects.locations.endpoints.generateContent
                        args:
                            model: ${"projects/" + project + "/locations/" + location + "/publishers/google/models/" + model}
                            region: ${location}
                            body:
                                contents:
                                    role: "USER"
                                    parts:
                                        text: ${"Can you tell me about the history of " + country}
                                generation_config:
                                    temperature: 0.5
                                    max_output_tokens: 2048
                                    top_p: 0.8
                                    top_k: 40
                        result: llm_response
                    - add_to_histories:
                        assign:
                            - histories[country]: ${llm_response.candidates[0].content.parts[0].text}
    - return_result:
        return: ${histories}

  3. Deploy the workflow by entering the following command:

    gcloud workflows deploy gemini-pro-country-histories \
    --source=gemini-pro-country-histories.yaml \
    --location=us-central1 \
    --service-account=SERVICE_ACCOUNT_NAME@PROJECT_ID.iam.gserviceaccount.com

Execute the workflow

Executing a workflow runs the current workflow definition associated with the workflow.

Console

  1. In the Google Cloud console, go to the Workflows page.

    Go to Workflows

  2. On the Workflows page, select the gemini-pro-country-histories workflow to go to its details page.

  3. On the Workflow details page, click Execute.

  4. For the Input, enter the following:

    {"countries":["Argentina", "Bhutan", "Cyprus", "Denmark", "Ethiopia"]}

  5. Click Execute again.

  6. View the results of the workflow in the Output pane.

    The output should be similar to the following:

    {
  "Argentina": "The history of Argentina is a complex and fascinating one, marked by periods of prosperity and decline, political [...]
  "Bhutan": "The history of Bhutan is a rich and fascinating one, dating back to the 7th century AD. Here is a brief overview: [...]
  "Cyprus": "The history of Cyprus is a long and complex one, spanning over 10,000 years. The island has been ruled by a succession [...]
  "Denmark": "1. **Prehistory and Early History (c. 12,000 BC - 800 AD)**\\n   - The earliest evidence of human habitation in Denmark [...]
  "Ethiopia": "The history of Ethiopia is a long and complex one, stretching back to the earliest human civilizations. The country is [...]
}

gcloud

  1. Open a terminal.

  2. Execute the workflow:

    gcloud workflows run gemini-pro-country-histories \
    --data='{"countries":["Argentina", "Bhutan", "Cyprus", "Denmark", "Ethiopia"]}' \
    --location=us-central1

    The execution results should be similar to the following:

      Waiting for execution [7ae1ccf1-29b7-4c2c-99ec-7a12ae289391] to complete...done.
  argument: '{"countries":["Argentina","Bhutan","Cyprus","Denmark","Ethiopia"]}'
  createTime: '2024-02-09T16:25:16.742349156Z'
  duration: 12.075968673s
  endTime: '2024-02-09T16:25:28.818317829Z'
  name: projects/1051295516635/locations/us-central1/workflows/gemini-pro-country-histories/executions/7ae1ccf1-29b7-4c2c-99ec-7a12ae289391
  result: "{\"Argentina\":\"The history of Argentina can be traced back to the arrival\
    [...]
    n* 2015: Argentina elects Mauricio Macri as president.\",\"Bhutan\":\"The history\
    [...]
    \ natural beauty, ancient monasteries, and friendly people.\",\"Cyprus\":\"The history\
    [...]
    ,\"Denmark\":\"The history of Denmark can be traced back to the Stone Age, with\
    [...]
    \ a high standard of living.\",\"Ethiopia\":\"The history of Ethiopia is long and\
    [...]
  startTime: '2024-02-09T16:25:16.742349156Z'
  state: SUCCEEDED

Deploy a workflow that generates country histories (PaLM 2 for text)

You might not want to use Gemini Pro as your model. The following example uses a workflow similar to the preceding one; however, it uses a connector method (predict) to make a request to a PaLM 2 for text publisher endpoint. The method performs an online prediction.

For more information about the HTTP request body parameters used when prompting the LLM, and the response body elements, see the PaLM 2 for text API reference.

Console

  1. In the Google Cloud console, go to the Workflows page.

    Go to Workflows

  2. Click Create.

  3. Enter a name for the new workflow: text-bison-country-histories.

  4. In the Region list, select us-central1 (Iowa).

  5. For the Service account, select the service account you previously created.

  6. Click Next.

  7. In the workflow editor, enter the following definition for your workflow:

    main:
    params: [args]
    steps:
    - init:
        assign:
            - project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
            - location: "us-central1"
            - model: "text-bison"
            - histories: {}
    - loop_over_countries:
        parallel:
            shared: [histories]
            for:
                value: country
                in: ${args.countries}
                steps:
                    - ask_llm:
                        call: googleapis.aiplatform.v1.projects.locations.endpoints.predict
                        args:
                            endpoint: ${"projects/" + project + "/locations/" + location + "/publishers/google/models/" + model }
                            region: ${location}
                            body:
                                instances:
                                    - prompt: '${"Can you  tell me about the history of " + country}'
                                parameters:
                                    temperature: 0.5
                                    maxOutputTokens: 2048
                                    topP: 0.8
                                    topK: 40
                        result: llm_response
                    - add_to_histories:
                        assign:
                            - history: ${llm_response.predictions[0].content}
                            # Remove leading whitespace from start of text
                            - history: ${text.substring(history, 1, len(history))}
                            - histories[country]: ${history}
    - return_result:
        return: ${histories}

    Note that depending on the model used, you might need to remove any unnecessary whitespace from the response.

  8. Click Deploy.

gcloud

  1. Create a source code file for your workflow:

    touch text-bison-country-histories.yaml

  2. In a text editor, copy the following workflow to your source code file:

    main:
    params: [args]
    steps:
    - init:
        assign:
            - project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
            - location: "us-central1"
            - model: "text-bison"
            - histories: {}
    - loop_over_countries:
        parallel:
            shared: [histories]
            for:
                value: country
                in: ${args.countries}
                steps:
                    - ask_llm:
                        call: googleapis.aiplatform.v1.projects.locations.endpoints.predict
                        args:
                            endpoint: ${"projects/" + project + "/locations/" + location + "/publishers/google/models/" + model }
                            region: ${location}
                            body:
                                instances:
                                    - prompt: '${"Can you  tell me about the history of " + country}'
                                parameters:
                                    temperature: 0.5
                                    maxOutputTokens: 2048
                                    topP: 0.8
                                    topK: 40
                        result: llm_response
                    - add_to_histories:
                        assign:
                            - history: ${llm_response.predictions[0].content}
                            # Remove leading whitespace from start of text
                            - history: ${text.substring(history, 1, len(history))}
                            - histories[country]: ${history}
    - return_result:
        return: ${histories}

    Note that depending on the model used, you might need to remove any unnecessary whitespace from the response.

  3. Deploy the workflow by entering the following command:

    gcloud workflows deploy text-bison-country-histories \
    --source=text-bison-country-histories.yaml \
    --location=us-central1 \
    --service-account=SERVICE_ACCOUNT_NAME@PROJECT_ID.iam.gserviceaccount.com

Deploy a workflow that summarizes a large document (Gemini Pro)

Deploy a workflow that divides a large document into smaller parts, making http.post requests to a Gemini Pro publisher endpoint in parallel so that the model can summarize each part simultaneously. The workflow finally combines all the partial summaries into a complete one.

For more information about the HTTP request body parameters used when prompting the LLM, and the response body elements, see the Gemini API reference.

The workflow definition assumes that you have created a Cloud Storage bucket to which you can upload a text file. For more information about the Workflows connector (googleapis.storage.v1.objects.get) used to retrieve objects from the Cloud Storage bucket, see the Connectors reference.

After deploying the workflow, you can execute it by creating an appropriate Eventarc trigger and then by uploading a file to the bucket. For more information, see Route Cloud Storage events to Workflows. Note that additional APIs must be enabled, and additional roles must be granted, including granting to your service account the Storage Object User (roles/storage.objectUser) role that supports using Cloud Storage objects. For more information, see the Prepare to create a trigger section.

Console

  1. In the Google Cloud console, go to the Workflows page.

    Go to Workflows

  2. Click Create.

  3. Enter a name for the new workflow: gemini-pro-summaries.

  4. In the Region list, select us-central1 (Iowa).

  5. For the Service account, select the service account you previously created.

  6. Click Next.

  7. In the workflow editor, enter the following definition for your workflow:

    main:
    params: [input]
    steps:
    - assign_file_vars:
        assign:
            - file_size: ${int(input.data.size)}
            - chunk_size: 64000
            - n_chunks: ${int(file_size / chunk_size)}
            - summaries: []
            - all_summaries_concatenated: ""
    - loop_over_chunks:
        parallel:
            shared: [summaries]
            for:
                value: chunk_idx
                range: ${[0, n_chunks]}
                steps:
                    - assign_bounds:
                        assign:
                            - lower_bound: ${chunk_idx * chunk_size}
                            - upper_bound: ${(chunk_idx + 1) * chunk_size}
                            - summaries: ${list.concat(summaries, "")}
                    - dump_file_content:
                        call: http.get
                        args:
                            url: ${"https://storage.googleapis.com/storage/v1/b/" + input.data.bucket + "/o/" + input.data.name + "?alt=media"}
                            auth:
                                type: OAuth2
                            headers:
                                Range: ${"bytes=" + lower_bound + "-" + upper_bound}
                        result: file_content
                    - assign_chunk:
                        assign:
                            - chunk: ${file_content.body}
                    - generate_chunk_summary:
                        call: ask_gemini_for_summary
                        args:
                            textToSummarize: ${chunk}
                        result: summary
                    - assign_summary:
                        assign:
                            - summaries[chunk_idx]: ${summary}
    - concat_summaries:
        for:
            value: summary
            in: ${summaries}
            steps:
                - append_summaries:
                    assign:
                        - all_summaries_concatenated: ${all_summaries_concatenated + "\n" + summary}
    - reduce_summary:
        call: ask_gemini_for_summary
        args:
            textToSummarize: ${all_summaries_concatenated}
        result: final_summary
    - return_result:
        return:
            - summaries: ${summaries}
            - final_summary: ${final_summary}

ask_gemini_for_summary:
    params: [textToSummarize]
    steps:
        - init:
            assign:
                - project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
                - location: "us-central1"
                - model: "gemini-pro"
                - summary: ""
        - call_gemini:
            call: http.post
            args:
                url: ${"https://" + location + "-aiplatform.googleapis.com" + "/v1/projects/" + project + "/locations/" + location + "/publishers/google/models/" + model + ":generateContent"}
                auth:
                    type: OAuth2
                body:
                    contents:
                        role: user
                        parts:
                            - text: '${"Make a summary of the following text:\n\n" + textToSummarize}'
                    generation_config:
                        temperature: 0.2
                        maxOutputTokens: 2000
                        topK: 10
                        topP: 0.9
            result: gemini_response
        # Sometimes, there's no text, for example, due to safety settings
        - check_text_exists:
            switch:
            - condition: ${not("parts" in gemini_response.body.candidates[0].content)}
              next: return_summary
        - extract_text:
            assign:
                - summary: ${gemini_response.body.candidates[0].content.parts[0].text}
        - return_summary:
            return: ${summary}

  8. Click Deploy.

gcloud

  1. Create a source code file for your workflow:

    touch gemini-pro-summaries.yaml

  2. In a text editor, copy the following workflow to your source code file:

    main:
    params: [input]
    steps:
    - assign_file_vars:
        assign:
            - file_size: ${int(input.data.size)}
            - chunk_size: 64000
            - n_chunks: ${int(file_size / chunk_size)}
            - summaries: []
            - all_summaries_concatenated: ""
    - loop_over_chunks:
        parallel:
            shared: [summaries]
            for:
                value: chunk_idx
                range: ${[0, n_chunks]}
                steps:
                    - assign_bounds:
                        assign:
                            - lower_bound: ${chunk_idx * chunk_size}
                            - upper_bound: ${(chunk_idx + 1) * chunk_size}
                            - summaries: ${list.concat(summaries, "")}
                    - dump_file_content:
                        call: http.get
                        args:
                            url: ${"https://storage.googleapis.com/storage/v1/b/" + input.data.bucket + "/o/" + input.data.name + "?alt=media"}
                            auth:
                                type: OAuth2
                            headers:
                                Range: ${"bytes=" + lower_bound + "-" + upper_bound}
                        result: file_content
                    - assign_chunk:
                        assign:
                            - chunk: ${file_content.body}
                    - generate_chunk_summary:
                        call: ask_gemini_for_summary
                        args:
                            textToSummarize: ${chunk}
                        result: summary
                    - assign_summary:
                        assign:
                            - summaries[chunk_idx]: ${summary}
    - concat_summaries:
        for:
            value: summary
            in: ${summaries}
            steps:
                - append_summaries:
                    assign:
                        - all_summaries_concatenated: ${all_summaries_concatenated + "\n" + summary}
    - reduce_summary:
        call: ask_gemini_for_summary
        args:
            textToSummarize: ${all_summaries_concatenated}
        result: final_summary
    - return_result:
        return:
            - summaries: ${summaries}
            - final_summary: ${final_summary}

ask_gemini_for_summary:
    params: [textToSummarize]
    steps:
        - init:
            assign:
                - project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
                - location: "us-central1"
                - model: "gemini-pro"
                - summary: ""
        - call_gemini:
            call: http.post
            args:
                url: ${"https://" + location + "-aiplatform.googleapis.com" + "/v1/projects/" + project + "/locations/" + location + "/publishers/google/models/" + model + ":generateContent"}
                auth:
                    type: OAuth2
                body:
                    contents:
                        role: user
                        parts:
                            - text: '${"Make a summary of the following text:\n\n" + textToSummarize}'
                    generation_config:
                        temperature: 0.2
                        maxOutputTokens: 2000
                        topK: 10
                        topP: 0.9
            result: gemini_response
        # Sometimes, there's no text, for example, due to safety settings
        - check_text_exists:
            switch:
            - condition: ${not("parts" in gemini_response.body.candidates[0].content)}
              next: return_summary
        - extract_text:
            assign:
                - summary: ${gemini_response.body.candidates[0].content.parts[0].text}
        - return_summary:
            return: ${summary}

  3. Deploy the workflow by entering the following command:

    gcloud workflows deploy gemini-pro-summaries \
    --source=gemini-pro-summaries.yaml \
    --location=us-central1 \
    --service-account=SERVICE_ACCOUNT_NAME@PROJECT_ID.iam.gserviceaccount.com

Clean up

To avoid incurring charges to your Google Cloud account for the resources used in this tutorial, either delete the project that contains the resources, or keep the project and delete the individual resources.

Delete the project

Console

  1. In the Google Cloud console, go to the Manage resources page.

    Go to Manage resources

  2. In the project list, select the project that you want to delete, and then click Delete.
  3. In the dialog, type the project ID, and then click Shut down to delete the project.

gcloud

Delete a Google Cloud project:

gcloud projects delete PROJECT_ID

Delete individual resources

Delete the workflows that you created in this tutorial.

What's next