Data entry ops · Production

ApertureDB builds JSON-first AQL for multimodal AI with RAG, GraphRAG, and agentic MCP interfaces

The problem

SQL and graph query languages could not handle multimodal AI queries—involving vector search, cross-modal joins, and media preprocessing—without compounding JOIN complexity, performance penalties, and language-extension sprawl that slowed development and broke compatibility.

First attempt

SQL required ever-growing JOIN chains and syntax extensions for each new multimodal data type, hurting development speed while breaking SQL compatibility; graph query languages lacked native vector search and data preprocessing support.

Workflow diagram · grounded in source
1
Natural language query submitted
trigger
“User typed a natural language query asking for highlights, pictures, video snippets, and a list of performers from a Taylor Swift concert attended by more than 20,000 people in midwest America.”
2
RAG/GraphRAG processes query
ai_action
“We've built retrieval-augmented generation (RAG) and GraphRAG chatbots that serve as natural language interfaces to ApertureDB”
3
MCP injects memory and context
integration
“Our MCP server plugin enables agents and chatbots to inject structured memory and multimodal context directly into queries, grounding natural language interactions in precise, queryable semantics”
4
AI Workflows execute data operations
ai_action
“They are structured sequences of operations designed to solve common AI/ML tasks like multimodal data ingestion, search, data correlation, or metadata filtering, using graph, vector, and multimodal-native capabilities offered by ApertureDB”
5
Multimodal results delivered via UI
output
“You can do no-code searches for your images, videos, PDFs, run semantic searches across these data types”
Reported outcome

ApertureDB built AQL, a JSON-first query language natively supporting multimodal data types, vector search, graph traversal, and preprocessing, with layered interfaces including RAG/GraphRAG chatbots and an MCP server for agentic systems.

Reported stack
ApertureDBAQLRAGGraphRAGMCPOpenCVFFMPEGLabelStudioGrafanaMetaBasePixeltablePandasSQLAlchemySPARQL
Source
https://mlops.community/blog/beyond-sql-the-query-language-multimodal-ai-really-needs
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

ApertureDB built AQL, a JSON-first query language natively supporting multimodal data types, vector search, graph traversal, and preprocessing, with layered interfaces including RAG/GraphRAG chatbots and an MCP server…

What tools did this team use?

ApertureDB, AQL, RAG, GraphRAG, MCP, OpenCV, FFMPEG, LabelStudio, Grafana, MetaBase.

What failed first in this deployment?

SQL required ever-growing JOIN chains and syntax extensions for each new multimodal data type, hurting development speed while breaking SQL compatibility; graph query languages lacked native vector search and data pre…

How is this data entry ops AI workflow structured?

Natural language query submitted → RAG/GraphRAG processes query → MCP injects memory and context → AI Workflows execute data operations → Multimodal results delivered via UI.

Related data entry ops cases