Kidong Lee·Jul 23Beyond RAG: Giving AI Agents a Typed Ontology and Governed Write-Back — an Open Alternative to…Most “AI + data” stories stop at retrieval. You wire up a vector store, do RAG, and your agent can answer questions. That’s the easy half…
Kidong Lee·Jun 28The Ontul Lance Connector — Multimodal Vector & Full‑Text Search for AgentsLance is a “data format for AI/ML workloads.” It treats the things classic analytics formats (Parquet/Iceberg) don’t do well — random…
Kidong Lee·Jun 25Realtime Iceberg Serving — Query Iceberg Like a DatabaseNo data movement. PostgreSQL-compatible. Governance built in.
Kidong Lee·Jun 12From Kafka to Iceberg with No Code — Introducing ItdaStream StreamingProduce a message to a topic, and it lands in an Iceberg table — without writing a single line of code.
Kidong Lee·Jun 2ShannonStore: an S3-Compatible Distributed Object Store Built on HRW-Based Elastic ScalingA look at the distributed object store that keeps up with petabyte-scale growth — with no downtime — for on-prem and private clouds.
Kidong Lee·Jun 2Meet kiok — a workflow orchestrator that ships with its own brainIf you have ever stood up a workflow scheduler in production, the shopping list is depressingly familiar:
Kidong Lee·May 31Give Your AI Agent a Semantic Layer, Not a Schema DumpText-to-SQL agents have a dirty secret: they’re confidently wrong. Hand a large language model your raw schema and ask for “revenue by…A response icon1A response icon1
Kidong Lee·May 10One Engine, Not Three: Vector + Graph + Full-Text in a Single Database for Agent RetrievalThe retrieval shape modern agents converge toA response icon2A response icon2
Kidong Lee·May 8One database, not two: vector + graph in a single engine for ontology agentsIf you’re building an ontology-driven agent, you’ve almost certainly run into the same architectural pattern:A response icon1A response icon1