feat(dotAI): Dot AI LangChain4J - Amazon Bedrock

[ihoffmann-dot]

Summary

Adds AWS Bedrock as a supported provider. Bedrock is a managed platform that
proxies multiple model families (Anthropic, Amazon Titan, Cohere, Meta, etc.)
via a unified Converse API — a single integration covers all of them.

• Add langchain4j-bedrock dependency
• Add bedrock case to LangChain4jModelFactory switch
• Implement buildBedrockChatModel using BedrockRuntimeClient with explicit or IAM role credentials
• Implement buildBedrockEmbeddingModel with automatic Titan/Cohere dispatch by model ID prefix
• Add embeddingInputType field to ProviderConfig (Cohere-specific; default: search_document)
• buildBedrockImageModel throws UnsupportedOperationException (no LangChain4J support)
• Add 4 unit tests in LangChain4jModelFactoryTest

Configuration

{
  "chat": {
    "provider": "bedrock",
    "region": "us-east-1",
    "accessKeyId": "...",
    "secretAccessKey": "...",
    "model": "anthropic.claude-3-5-sonnet-20241022-v2:0",
    "maxTokens": 16384,
    "temperature": 1.0
  },
  "embeddings": {
    "provider": "bedrock",
    "region": "us-east-1",
    "accessKeyId": "...",
    "secretAccessKey": "...",
    "model": "amazon.titan-embed-text-v2:0"
  }
}

Notes

• If accessKeyId / secretAccessKey are omitted, credentials resolve via DefaultCredentialsProvider (IAM role, environment, ~/.aws/credentials).
• Embedding dispatch: model IDs starting with cohere. → BedrockCohereEmbeddingModel; all others → BedrockTitanEmbeddingModel.
• embeddingInputType is Cohere-only. Use search_document when indexing content, search_query when embedding a search query. Titan silently ignores this field.
• Image generation via Bedrock is not available through LangChain4J. Attempting it throws UnsupportedOperationException.

Related Issue

This PR fixes #35334
EPIC: dotAI Multi-Provider Support #33970

[github-actions]

❌ Codex Review failed — openai.gpt-5.5

The review job failed before producing output. See the run for details.

_{Run: #27446303016}

[fabrizzio-dotCMS]

same here

[github-actions]

🤖 Bedrock Review — deepseek.v3.2

[🔴 Critical] dotCMS/pom.xml:525 — Adding a new dependency (langchain4j-bedrock) without specifying a version violates dotCMS conventions. Dependency versions must be defined in bom/application/pom.xml only, never in dotCMS/pom.xml. This will cause a build failure.

[🟡 Medium] dotCMS/src/main/java/com/dotcms/ai/client/langchain4j/BedrockModelProviderStrategy.java:139-142 — Logging timeout and maxRetries as debug when unsupported for embedding models is appropriate, but there's no warning to the user that these configurations are silently ignored, which could lead to confusion. Consider adding a Logger.warn for at least one of these cases.

[🟡 Medium] dotCMS/src/main/java/com/dotcms/ai/client/langchain4j/BedrockModelProviderStrategy.java:154 — The config.embeddingInputType() is passed directly to BedrockCohereEmbeddingModel.builder().inputType(...). If the provided value is invalid (not search_document or search_query), the error will come from the LangChain4J library at runtime, not during validation. This is acceptable but could be documented.

[🟡 Medium] dotCMS/src/main/java/com/dotcms/ai/client/langchain4j/BedrockModelProviderStrategy.java:156 — The condition modelLower.startsWith("amazon.titan-") may not match all Titan embedding model IDs (e.g., amazon.titan-embed-g1-text-02). The check should be modelLower.contains("titan") or a more flexible pattern to avoid missing future Titan variants.

[🟡 Medium] dotCMS/src/main/java/com/dotcms/ai/client/langchain4j/BedrockModelProviderStrategy.java:159 — Setting dimensions on BedrockTitanEmbeddingModel may not be supported for all Titan models; passing null could cause an error. The LangChain4J library should handle this, but it's a potential runtime issue.

[🟡 Medium] dotCMS/src/main/java/com/dotcms/ai/client/langchain4j/BedrockModelProviderStrategy.java:239 — The overrideConfiguration method uses AwsRetryStrategy.standardRetryStrategy().toBuilder(). If maxRetries is set to a large number, it could lead to excessive retries and prolonged failures. Ensure the value is reasonable (maybe cap it), though this is currently left to the user.

[🟡 Medium] dotCMS/src/test/java/com/dotcms/ai/client/langchain4j/BedrockModelProviderStrategyTest.java:465 — The test helper bedrockConfig uses hardcoded AWS credentials ("test-access-key", "test-secret-key"). While this is safe for unit tests, ensure no real secrets are ever committed. This is acceptable for test code.

[🟡 Medium] dotCMS/src/test/java/com/dotcms/ai/client/langchain4j/BedrockModelProviderStrategyTest.java:156 — Test test_buildEmbeddingModel_titanMixedCase_routesToTitan uses "AMAZON.TITAN-embed-text-v2:0". The model ID casing may not be preserved by AWS; the test should verify that the routing logic is case-insensitive as intended.

[🟡 Medium] dotCMS/src/test/java/com/dotcms/ai/client/langchain4j/LangChain4jModelFactoryTest.java:350 — Test test_buildEmbeddingModel_bedrock_cohereUppercase_routesToCohere uses "Cohere.embed-english-v3" (mixed case). This is good for verifying case-insensitive routing, but ensure the actual AWS model ID is case-sensitive; the test should reflect that the strategy lowercases the model for routing only.

---

_{Run: #27647750313 · tokens: in: 12404 · out: 863 · total: 13267}