Examples

This page provides practical examples of using Wallaroo AI endpoints in your DB2 for IBM i applications.

Note: Before running these examples, make sure to configure your Wallaroo endpoint settings. See the Usage page for configuration instructions.

Basic Examples

Simple Text Generation

-- Generate text using Wallaroo AI
SELECT dbsdk_v1.wallaroo_generate('Why are armadillos so cute?') as response
FROM sysibm.sysdummy1;

-- Generate with custom parameters
SELECT dbsdk_v1.wallaroo_generate(
  'Explain quantum computing in simple terms',
  '{
    "max_tokens": 200,
    "temperature": 0.7,
    "top_p": 0.9
  }'
) as response FROM sysibm.sysdummy1;

JSON Response Generation

Get complete JSON response from Wallaroo AI:

-- Get full JSON response
SELECT dbsdk_v1.wallaroo_generate_json(
  'Write a haiku about databases',
  '{"max_tokens": 100, "temperature": 0.8}'
) as json_response
FROM sysibm.sysdummy1;

-- Parse specific fields from JSON response
SELECT
  JSON_VALUE(response, '$.choices[0].text') as generated_text,
  JSON_VALUE(response, '$.usage.total_tokens') as total_tokens,
  JSON_VALUE(response, '$.model') as model_used
FROM (
  SELECT dbsdk_v1.wallaroo_generate_json(
    'Write a haiku about databases',
    '{
      "max_tokens": 100,
      "temperature": 0.8
    }'
  ) as response
  FROM sysibm.sysdummy1
);

Advanced Examples

Customer Feedback Analysis

Analyze customer feedback using Wallaroo AI:

-- Create a sample table with customer feedback
CREATE TABLE sample.customer_feedback (
  feedback_id INT PRIMARY KEY,
  customer_name VARCHAR(100),
  product VARCHAR(100),
  comment VARCHAR(1000)
);

INSERT INTO sample.customer_feedback VALUES
  (1, 'John Smith', 'Database Suite', 'The performance is excellent and documentation is clear!'),
  (2, 'Maria Garcia', 'Analytics Tool', 'Good product but needs better error messages'),
  (3, 'Robert Chen', 'API Gateway', 'Terrible experience, nothing works as expected'),
  (4, 'Sarah Johnson', 'Database Suite', 'Love the new features in the latest release');

-- Analyze sentiment and extract insights
SELECT
  f.feedback_id,
  f.customer_name,
  f.product,
  f.comment,
  dbsdk_v1.wallaroo_generate(
    'Analyze this customer feedback and provide: 1) Sentiment (Positive/Neutral/Negative), 2) Key topics, 3) Action items. Feedback: ' || f.comment,
    '{"max_tokens": 200, "temperature": 0.3}'
  ) as analysis
FROM sample.customer_feedback f;

Dynamic SQL Query Generation

Use Wallaroo AI to generate SQL queries from natural language:

-- Set up a sample table for demonstration
CREATE TABLE sample.employees (
  id INT PRIMARY KEY,
  name VARCHAR(100),
  department VARCHAR(50),
  salary DECIMAL(10,2)
);

-- Insert sample data
INSERT INTO sample.employees VALUES
  (1, 'John Smith', 'Engineering', 85000.00),
  (2, 'Maria Garcia', 'Engineering', 92000.00),
  (3, 'Robert Chen', 'Marketing', 78500.00),
  (4, 'Sarah Johnson', 'Marketing', 81200.00),
  (5, 'James Wilson', 'Finance', 95000.00),
  (6, 'Aisha Patel', 'Finance', 98500.00),
  (7, 'David Kim', 'Engineering', 78000.00);

-- Generate SQL query from natural language request
SELECT dbsdk_v1.wallaroo_generate(
  'Write an SQL query to find the average salary by department for the sample.employees table with columns: id, name, department, salary',
  '{"max_tokens": 150, "temperature": 0.1}'
) as generated_query
FROM sysibm.sysdummy1;

Data Summarization

Summarize data from a query:

-- Create sales data table
CREATE TABLE sample.sales_data (
  region VARCHAR(50),
  q1_sales INT,
  q2_sales INT,
  q3_sales INT,
  q4_sales INT
);

INSERT INTO sample.sales_data VALUES
  ('North', 150000, 180000, 120000, 200000),
  ('South', 120000, 110000, 140000, 130000),
  ('East', 160000, 170000, 180000, 190000),
  ('West', 130000, 140000, 150000, 160000);

-- Generate a summary
SELECT dbsdk_v1.wallaroo_generate(
  'Analyze the following sales data and provide insights: ' ||
  'Region: North, Q1: 150000, Q2: 180000, Q3: 120000, Q4: 200000. ' ||
  'Region: South, Q1: 120000, Q2: 110000, Q3: 140000, Q4: 130000. ' ||
  'Region: East, Q1: 160000, Q2: 170000, Q3: 180000, Q4: 190000. ' ||
  'Region: West, Q1: 130000, Q2: 140000, Q3: 150000, Q4: 160000.',
  '{"max_tokens": 500, "temperature": 0.7}'
) as sales_summary
FROM sysibm.sysdummy1;

Integration Patterns

Storing AI-Generated Content

-- Create a table to store generated content
CREATE TABLE sample.ai_generated_content (
  content_id INT GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
  prompt_text VARCHAR(1000),
  generated_content VARCHAR(5000),
  created_timestamp TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

-- Generate and store content
INSERT INTO sample.ai_generated_content (prompt_text, generated_content)
VALUES (
  'Explain the benefits of DB2 for IBM i',
  dbsdk_v1.wallaroo_generate(
    'Explain the benefits of DB2 for IBM i',
    '{"max_tokens": 300, "temperature": 0.7}'
  )
);

-- Retrieve stored content
SELECT * FROM sample.ai_generated_content
ORDER BY created_timestamp DESC;

Custom Base URL Support

If your Wallaroo endpoint uses a different URL structure, you can specify a custom base URL:

-- Use custom base URL for a specific endpoint
SELECT dbsdk_v1.wallaroo_generate(
  'What is machine learning?',
  '{"max_tokens": 150, "temperature": 0.5}',
  'https://custom.wallaroo.io/v1/api/pipelines/infer/my-pipeline/openai/v1/completions'
) as response
FROM sysibm.sysdummy1;

These examples demonstrate how to leverage Wallaroo AI endpoints for various use cases in your IBM i applications, from simple text generation to complex business analytics and reporting workflows.