chore: Add chatbot examples for OpenAI and AWS Bedrock

examples/hello-bedrock/Gemfile

@@ -2,5 +2,7 @@
 
 source 'https://rubygems.org'
 
+ruby '>= 3.0.0'
+
 gem 'aws-sdk-bedrockruntime'
 gem 'launchdarkly-server-sdk-ai', path: '../../..'

examples/hello-bedrock/hello_bedrock.rb

@@ -0,0 +1,144 @@
+# frozen_string_literal: true
+
+require 'aws-sdk-bedrockruntime'
+require 'launchdarkly-server-sdk'
+require 'launchdarkly-server-sdk-ai'
+
+# Set sdk_key to your LaunchDarkly SDK key.
+sdk_key = ENV['LAUNCHDARKLY_SDK_KEY']
+
+# Set config_key to the AI Config key you want to evaluate.
+ai_config_key = ENV['LAUNCHDARKLY_AI_CONFIG_KEY'] || 'sample-ai-config'
+
+# Set aws_access_key_id and aws_secret_access_key for AWS credentials.
+aws_access_key_id = ENV['AWS_ACCESS_KEY_ID']
+aws_secret_access_key = ENV['AWS_SECRET_ACCESS_KEY']
+region = ENV['AWS_REGION'] || 'us-east-1'
+
+if sdk_key.nil? || sdk_key.empty?
+  puts '*** Please set the LAUNCHDARKLY_SDK_KEY env first'
+  exit 1
+end
+
+if aws_access_key_id.nil? || aws_access_key_id.empty?
+  puts '*** Please set the AWS_ACCESS_KEY_ID env variable first'
+  exit 1
+end
+
+if aws_secret_access_key.nil? || aws_secret_access_key.empty?
+  puts '*** Please set the AWS_SECRET_ACCESS_KEY env variable first'
+  exit 1
+end
+
+#
+# Chatbot class that interacts with LaunchDarkly AI and AWS Bedrock
+#
+class BedrockChatbot
+  attr_reader :aiclient, :ai_config_key, :bedrock_client
+
+  DEFAULT_VALUE = LaunchDarkly::Server::AI::AIConfig.new(
+    enabled: true,
+    model: LaunchDarkly::Server::AI::ModelConfig.new(name: 'my-default-model'),
+    messages: [
+      LaunchDarkly::Server::AI::Message.new('system',
+                                            'You are a default unhelpful assistant with the persona of HAL 9000 talking with {{ldctx.name}}'),
+      LaunchDarkly::Server::AI::Message.new('user', '{{user_question}}'),
+    ]
+  )
+
+  def initialize(aiclient, ai_config_key, bedrock_client, context)
+    @aiclient = aiclient
+    @ai_config_key = ai_config_key
+    @bedrock_client = bedrock_client
+    @context = context
+  end
+
+  def ask_agent(question)
+    ai_config = aiclient.config(
+      @ai_config_key,
+      @context,
+      DEFAULT_VALUE,
+      { user_question: question }
+    )
+
+    begin
+      response = ai_config.tracker.track_bedrock_converse_metrics do
+        @bedrock_client.converse(
+          map_converse_arguments(
+            ai_config.model.name,
+            ai_config.messages
+          )
+        )
+      end
+      [response.output.message.content[0].text, ai_config.tracker]
+    rescue StandardError => e
+      ["An error occured: #{e.message}", nil]
+    end
+  end
+
+  def agent_was_helpful(tracker, helpful)
+    kind = helpful ? :positive : :negative
+    tracker.track_feedback(kind: kind)
+  end
+
+  def map_converse_arguments(model_id, messages)
+    args = {
+      model_id: model_id,
+    }
+
+    mapped_messages = []
+    user_messages = messages.select { |msg| msg.role == 'user' }
+    mapped_messages << { role: 'user', content: user_messages.map { |msg| { text: msg.content } } } unless user_messages.empty?
+
+    assistant_messages = messages.select { |msg| msg.role == 'assistant' }
+    mapped_messages << { role: 'assistant', content: assistant_messages.map { |msg| { text: msg.content } } } unless assistant_messages.empty?
+    args[:messages] = mapped_messages unless mapped_messages.empty?
+
+    system_messages = messages.select { |msg| msg.role == 'system' }
+    args[:system] = system_messages.map { |msg| { text: msg.content } } unless system_messages.empty?
+
+    args
+  end
+end
+
+# Initialize the LaunchDarkly client
+ld_client = LaunchDarkly::LDClient.new(sdk_key)
+ai_client = LaunchDarkly::Server::AI::Client.new(ld_client)
+
+unless ld_client.initialized?
+  puts '*** SDK failed to initialize!'
+  exit 1
+end
+
+# Create the LDContext
+context = LaunchDarkly::LDContext.create({
+                                           key: 'user-key',
+                                           kind: 'user',
+                                           name: 'Lucy',
+                                         })
+
+bedrock_client = Aws::BedrockRuntime::Client.new(
+  aws_access_key_id: aws_access_key_id,
+  aws_secret_access_key: aws_secret_access_key,
+  region: region
+)
+chatbot = BedrockChatbot.new(ai_client, ai_config_key, bedrock_client, context)
+
+loop do
+  print "Ask a question: (or type 'exit'): "
+  question = gets&.chomp
+  break if question.nil? || question.strip.downcase == 'exit'
+
+  response, tracker = chatbot.ask_agent(question)
+  puts "AI Response: #{response}"
+
+  next if tracker.nil? # If tracker is nil, skip feedback collection
+
+  print "Was the response helpful? [yes/no] (or type 'exit'): "
+  feedback = gets&.chomp
+  break if feedback.nil? || feedback.strip.downcase == 'exit'
+
+  chatbot.agent_was_helpful(tracker, feedback == 'yes')
+end
+
+ld_client.close

examples/hello-openai/Gemfile

@@ -0,0 +1,8 @@
+# frozen_string_literal: true
+
+source 'https://rubygems.org'
+
+ruby '>= 3.0.0'
+
+gem 'launchdarkly-server-sdk-ai', path: '../../..'
+gem 'openai', '~> 0.7.0'

examples/hello-openai/README.md

@@ -0,0 +1,27 @@
+# LaunchDarkly sample Ruby AI application
+
+We've built a simple console application that demonstrates how LaunchDarkly's Ruby AI SDK works.
+
+Below, you'll find the build procedure. For more comprehensive instructions, you can visit your [Quickstart page](https://docs.launchdarkly.com/home/ai-configs/quickstart) or the [Python reference guide](https://docs.launchdarkly.com/sdk/ai/ruby).
+
+This demo requires Ruby 3.0 or higher.
+
+## Build Instructions
+
+This repository includes examples for `OpenAI` and `Bedrock`. Depending on your preferred provider, you may have to take some additional steps.
+
+### General setup
+
+1. Install the required dependencies with `bundle install`
+1. Set the environment variable `LAUNCHDARKLY_SDK_KEY` to your LaunchDarkly SDK key. If there is an existing an AI Config in your LaunchDarkly project that you want to evaluate, set `LAUNCHDARKLY_AI_CONFIG_KEY` to the flag key; otherwise, an AI Config of `sample-ai-config` will be assumed.
+
+   ```bash
+   export LAUNCHDARKLY_SDK_KEY="1234567890abcdef"
+   export LAUNCHDARKLY_AI_CONFIG_KEY="sample-ai-config"
+   ```
+1. Set the environment variable `OPENAI_API_KEY` to your OpenAI key.
+
+   ```bash
+   export OPEN_API_KEY="0987654321fedcba"
+   ```
+1. Run the program `bundle exec ruby hello_openai.rb`

examples/hello-openai/hello_openai.rb

@@ -0,0 +1,113 @@
+# frozen_string_literal: true
+
+require 'launchdarkly-server-sdk'
+require 'launchdarkly-server-sdk-ai'
+require 'openai'
+
+# Set sdk_key to your LaunchDarkly SDK key.
+sdk_key = ENV['LAUNCHDARKLY_SDK_KEY']
+
+# Set config_key to the AI Config key you want to evaluate.
+ai_config_key = ENV['LAUNCHDARKLY_AI_CONFIG_KEY'] || 'sample-ai-config'
+
+# Set openai_api_key to your OpenAI API key.
+openai_api_key = ENV['OPENAI_API_KEY']
+
+if sdk_key.nil? || sdk_key.empty?
+  puts '*** Please set the LAUNCHDARKLY_SDK_KEY env first'
+  exit 1
+end
+
+if openai_api_key.nil? || openai_api_key.empty?
+  puts '*** Please set the OPENAI_API_KEY env first'
+  exit 1
+end
+
+#
+# Chatbot class that interacts with LaunchDarkly AI and OpenAI
+#
+class Chatbot
+  attr_reader :aiclient, :ai_config_key, :openai_client, :context
+
+  DEFAULT_VALUE = LaunchDarkly::Server::AI::AIConfig.new(
+    enabled: true,
+    model: LaunchDarkly::Server::AI::ModelConfig.new(name: 'my-default-model'),
+    messages: [
+      LaunchDarkly::Server::AI::Message.new('system',
+                                      'You are a default unhelpful assistant with the persona of HAL 9000 talking with {{ldctx.name}}'),
+      LaunchDarkly::Server::AI::Message.new('user', '{{user_question}}'),
+    ]
+  )
+
+  def initialize(aiclient, ai_config_key, openai_client, context)
+    @aiclient = aiclient
+    @ai_config_key = ai_config_key
+    @openai_client = openai_client
+    @context = context
+  end
+
+  def ask_agent(question)
+    ai_config = aiclient.config(
+      @ai_config_key,
+      @context,
+      DEFAULT_VALUE,
+      { user_question: question }
+    )
+
+    begin
+      completion = ai_config.tracker.track_openai_metrics do
+        @openai_client.chat.completions.create(
+          model: ai_config.model.name,
+          messages: ai_config.messages.map(&:to_h)
+        )
+      end
+      [completion[:choices][0][:message][:content], ai_config.tracker]
+    rescue StandardError => e
+      ["An error occurred: #{e.message}", nil]
+    end
+  end
+
+  def agent_was_helpful(tracker, helpful)
+    kind = helpful ? :positive : :negative
+    tracker.track_feedback(kind: kind)
+  end
+end
+
+ld_client = LaunchDarkly::LDClient.new(sdk_key)
+ai_client = LaunchDarkly::Server::AI::Client.new(ld_client)
+
+unless ld_client.initialized?
+  puts '*** SDK failed to initialize!'
+  exit 1
+end
+
+puts '*** SDK successfully initialized'
+
+# Create the LDContext
+context = LaunchDarkly::LDContext.create({
+                                            key: 'user-key',
+                                            kind: 'user',
+                                            name: 'Lucy',
+                                          })
+
+chatbot = Chatbot.new(ai_client, ai_config_key, OpenAI::Client.new(api_key: openai_api_key), context)
+
+loop do
+  print "Ask a question (or type 'exit'): "
+  input = gets&.chomp
+  break if input.nil? || input.strip.downcase == 'exit'
+
+  response, tracker = chatbot.ask_agent(input)
+  puts "AI Response: #{response}"
+
+  next if tracker.nil? # If tracker is nil, skip feedback collection
+
+  print "Was the response helpful? [yes/no] (or type 'exit'): "
+  feedback = gets&.chomp
+  break if feedback.nil? || feedback.strip.downcase == 'exit'
+
+  helpful = feedback.strip.downcase == 'yes'
+  chatbot.agent_was_helpful(tracker, helpful)
+end
+
+ld_client.close