NVIDIA open-sources Auto Ontology, a semantic layer to stop AI agents querying the wrong data
DrDatta_AIIMS · x · 2026-11-03
NVIDIA has open-sourced Auto Ontology, a Python framework (868 stars on GitHub) that acts as a semantic/context layer between AI agents and databases, targeting the failure mode where an agent answers fast but uses the wrong column — the README puts it bluntly: a revenue metric answered faster is worth nothing if the number is wrong.
The workflow has four steps:
- Discover: reads database metadata and drafts an ontology mapping business terms like "active customer" to fields
- Ground: agents query with certified business terms instead of guessing from raw table names
- Validate: checks generated SQL against certified definitions
- Expose: exposes the layer to LLM/agent workflows via MCP and skills
It enables deterministic, graph-based query generation, better context retrieval, and more explainable AI, under the Apache-2.0 license.
More from coding & agent
- Replit CEO Backs Call to Run AI Agents in Managed Cloud Environments by Default — amasad · 2026-11-03
- Pipecat's Voice Reader hears tone, sarcasm and energy in under 50ms on CPU — solyarisoftware · 2026-11-03
- SGLang HiCache extends KV reuse across parallelism; iFLYTEK hit rate 30%→95% — ying11231 · 2026-11-03
- Model leaderboard inside Copilot's agent harness shows success rate vs cost with error bars — 0xkarasy · 2026-11-03
- A Quick Prompt That Cuts Your Tests by 40% — RishabJainK · 2026-11-03
- Tool success is not evidence: the quiet agent failure mode nobody checks — Key-Register-1969 · 2026-11-03