InsightInsight from our research and the field

All articles

Tech Note2026.10.06

The tools our agent built for itself were not canvas nodes

Geny agents build their own tools mid-chat, yet users could not see them. Making those tools visible overturned three of our assumptions along the way.

박태준

Tech NoteLLM Inside2026.10.01

Does longer reasoning make an answer better? (Part 4)

Reasoning helps decompose problems; it is not proof. Separating hallucination types and task-specific checks makes model results explainable.

김해수

Tech NoteThe Ontology Diary2026.09.30

The ontology build and search improvement journey, Part 3 — What to decide before adoption

The current limits and open choices—delete and update, typed values, provenance, namesakes, and evaluation—turned into B2B adoption criteria.

김진수

Tech Note2026.09.29

Sharper service observability, down to each AI agent node

Extending service-level tracing to AI agent nodes, then connecting execution evidence to Grafana Tempo dashboards and alerts.

Insoo Jeon

Tech NoteThe Ontology Diary2026.09.23

The ontology build and search improvement journey, Part 2 — Eight search turns became one fusion

How we replaced repeated LLM exploration with one graph-vector fusion and made the canonical store and incremental refresh operational.

김진수