AkasicDB: Demonstrating Omni RAG with a Unified Vector-Graph-Relational DBMS

A native execution framework that jointly performs vector similarity search, graph traversal, and relational filtering for retrieval-augmented generation

Geonho Lee; Jeongho Park; Donghyoung Han; Min-Soo Kim· 2026· DOI 10.48550/arXiv.2608.09214

The core problem

Recent Retrieval-Augmented Generation (RAG) systems increasingly combine vector retrieval with structured knowledge, such as Graph RAG and Filtered vector search. However, existing database architectures struggle to support such complex RAG workflows efficiently, as they rely on out-of-DB pipelines or in-DB non-native integration, leading to high overhead. This demo paper presents AkasicDB, a database system that natively supports such RAG workflows by jointly executing vector similarity search, graph traversal, and relational filtering within a single execution framework. AkasicDB extends our prior work, Chimera, with native vector support to enable such unified execution. Based on AkasicDB, we demonstrate the first native integration of Vector-Graph-Relational RAG, which we refer to as Omni RAG. Through an interactive chat-style demonstration, users execute and visualize Omni RAG queries, directly experiencing its superior retrieval and reasoning over vector-only approaches while observing the practical limitations of existing database architectures in supporting Omni RAG. A demonstration video is available at https://youtu.be/8d09_dtrEIM.

The demonstration shows that AkasicDB achieves superior retrieval and reasoning compared to vector-only approaches. Users can directly experience the benefits of Omni RAG through interactive queries. The system also highlights the practical limitations of existing database architectures in supporting Omni RAG, such as high overhead due to out-of-DB pipelines or non-native integration. The demonstration video (https://youtu.be/8d09_dtrEIM) provides a visual walkthrough of the system's capabilities. Key results include the successful joint execution of vector, graph, and relational operations within a single framework, and the ability to visualize query results in real-time.

Recent Retrieval-Augmented Generation (RAG) systems increasingly combine vector retrieval with structured knowledge, such as Graph RAG and Filtered vector search. However, existing database architectures struggle to support such complex RAG workflows efficiently, as they rely on out-of-DB pipelines or in-DB non-native integration, leading to high overhead. This demo paper presents AkasicDB, a database system that natively supports such RAG workflows by jointly executing vector similarity search, graph traversal, and relational filtering within a single execution framework. AkasicDB extends our prior work, Chimera, with native vector support to enable such unified execution. Based on AkasicDB, we demonstrate the first native integration of Vector-Graph-Relational RAG, which we refer to as Omni RAG. Through an interactive chat-style demonstration, users execute and visualize Omni RAG queries, directly experiencing its superior retrieval and reasoning over vector-only approaches while observing the practical limitations of existing database architectures in supporting Omni RAG. A demonstration video is available at https://youtu.be/8d09_dtrEIM.

AkasicDB extends the Chimera system by incorporating native vector support into a unified execution framework. The core methodology involves jointly executing three types of operations: vector similarity search, graph traversal, and relational filtering. This is achieved through a single execution engine that avoids out-of-DB pipelines or non-native in-DB integration, thereby reducing overhead. The system is designed to support Omni RAG, which integrates vector, graph, and relational data natively. The demonstration uses an interactive chat-style interface where users can execute and visualize Omni RAG queries. The architecture can be represented as follows:

The paper discusses the challenges of existing database architectures in supporting complex RAG workflows. By natively integrating vector, graph, and relational operations, AkasicDB overcomes these challenges and enables efficient Omni RAG. The authors argue that this unified approach is essential for advanced RAG systems that require structured knowledge. The demonstration serves as a proof-of-concept for the first native integration of Vector-Graph-Relational RAG. Future work may involve further optimizations and extensions to other data models. The system's design emphasizes the importance of avoiding out-of-DB pipelines to reduce overhead and improve performance. The interactive chat-style demonstration allows users to directly observe the advantages of Omni RAG over vector-only approaches.

CS practitioners and researchers

Opening member content…