Datanexus

9. The Public Benchmark Returned 56%: Nine Experiments and What Got Ruled Out

I hit 80% on my own 30-question benchmark, but only 56% on BIRD Mini-Dev’s 50 public questions. Nine experiments later, I had ruled out the multi-candidate hypothesis from three different angles. What’s left is schema understanding and methodology.

8. From 66% to 80% NL2SQL Accuracy: Four Measure-and-Fix Loops

After wiring up the router, I ran a 30-question benchmark and pushed NL2SQL EX (Execution Accuracy) from 66.67% to 80%. Here’s what I fixed across four cycles and where things broke.

7. When a Question Comes In, Who Decides the Routing?

The term definitions are done. But when a user asks a question, who decides whether to search the graph, write SQL, or run a vector search? Things I ran into while designing the router.

6. When You Don't Have to Build Agent Infra Yourself, Harnesses Become Obsolete. What About the Ontology?

Shortly after the Conway leak, Anthropic officially launched Claude Managed Agents. As agent infrastructure gets absorbed into platforms, here’s why DataNexus’s ontology layer remains safe.

5. Automating Metadata Maintenance: Karpathy's LLM Wiki Architecture

RAG starts from scratch every time. Karpathy proposes having the LLM maintain a wiki directly so knowledge accumulates. DataNexus’s ontology catalog needs the same principle to avoid abandonment.

Legal Data Gets a Brain: The Synergy of korean-law-mcp and DataNexus

korean-law-mcp streamlines legal data access while DataNexus’s knowledge-graph-based ontology layer explicitly maps complex statutory connections, significantly enhancing AI’s ability to interpret law accurately.

4. Why We Added a SKOS Compatibility Layer

Why we chose SKOS to connect the DataNexus ontology with external systems. Designing a compatibility layer between LPG and RDF – two different graph models.

3. Can DataHub's Glossary Work as an Ontology?

We tried using DataHub’s Business Glossary as an ontology store. What worked, what didn’t, and how we worked around it.

2. How We Chose These 4 Open-Source Tools

How we decided on DataHub + Vanna + ApeRAG + DozerDB for DataNexus. What got eliminated from the candidate list, and why.

1. Why We're Building DataNexus

“What’s Your VIP Criteria?” This happened during a BI Agent project for a retail company. A business user was testing the Agent and asked, “Show me last month’s VIP customer revenue.” The system spit out a number, but the user didn’t look happy. “Something’s off. I think the VIP criteria are different from what our team uses.” Marketing’s VIP and CRM’s VIP were different. Same with revenue. Depending on whether you meant net revenue (순매출) or gross revenue (총매출), the difference could be hundreds of millions of won. ...