No wrapper class exists for 'com.pgvector.PGvector'

[carolosf]

Describe the feature

PGvector is useful for AI workloads but the lack of a wrapper class makes the AWS JDBC wrapper unusable for this kind of usecase.

Use Case

Retrieval-augmented generation support

Proposed Solution

No response

Other Information

No response

Acknowledgements

• I may be able to implement this feature request
• This feature might incur a breaking change

The AWS Advanced JDBC Driver version used

software.amazon.jdbc:aws-advanced-jdbc-wrapper:2.5.6

JDK version used

openjdk 22.0.1

Operating System and version

linux

[sergiyvamz]

Hi @carolosf

Thank you reaching out with this issue. Would you mind to provide a concrete example of Java code that you'd like to be supported by AWS Wrapper?

[carolosf]

Thanks for the response.

The use case I’m looking at is this:

https://github.com/quarkiverse/quarkus-langchain4j/tree/main/samples/movie-similarity-search

It should be possible to reproduce locally by just using the jdbc wrapper and setting jdbc properties.

[onestardao]

Hey @carlosf — totally agree with you on this.
We ran into the exact same limitation when trying to connect our pgvector-backed Postgres instance using AWS JDBC in a production-scale RAG system.

What's going on
The lack of a proper PGvector wrapper (or fallback serializer) breaks compatibility with any typed JDBC integration — it throws errors or defaults to Object, depending on implementation. That’s a blocker for teams using strongly-typed query builders, schema-first libraries, or real-time pipelines.

---

Workarounds we considered:

1. Fallback to raw SQL and JSONB for vectors (clunky, hard to scale)
2. Custom patch to wrap PGvector using PGobject — hacky and fragile
3. Switch to HTTP-based embedding store like Qdrant/Weaviate, bypassing JDBC entirely (but adds complexity)

None of these are ideal.

---

+1 on enhancement — adding PGvector support in AWS JDBC wrapper would seriously open doors for AI workloads, especially for retrieval-heavy applications like RAG, semantic search, and hybrid ranking.

If this gets added, it’d be a major unlock for downstream users building LLM infra on AWS.
Happy to help test or contribute if it moves forward.

Cheers!