Insight · Plateer AI Labs

Article series

LLM Inside

The core ideas you need to make sense of an LLM, taken one at a time.

4 articles

The GS Certification Diary

Three weeks of reading 80 defect reports and rebuilding the code, the errors, and the manual

5 articles

The Harness Diary

Designing the execution environment an AI agent works in, from the ground up

10 articles

The Ontology Diary

Starting from the questions, building a knowledge graph, then working on quality and search

13 articles

All articles

XGEN Preview2026.06.21

MCP Apps — build once, run anywhere

Past wrapping an engine, XGEN's SDK compiles workflows and policies into code and exports them as a standalone MCP server.

PPlateer AI Labs

Tech NoteThe Ontology Diary2026.06.16

Most of the nodes we painted red were not evidence (Part 7)

With retrieval quality raised, verifying on screen showed far more nodes highlighted than there was evidence for. What the highlight really proves.

김진수

Tech NoteThe GS Certification Diary2026.06.16

The manual was part of the product: what three weeks of GS certification left behind (Part 5)

Much of the 30 usability defects in GS certification were documentation, not code. Moving the PDF manual to an MkDocs wiki, and fixing Korean IME input.

유지수

Tech NoteThe Harness Diary2026.06.13

The 30x gap wasn't the model (Part 7)

Two models on the same workflow gave 2,488 and 84 seconds. Matching iteration ceilings and output limits turned 2,488 into about 100.

김진수

Tech NoteThe GS Certification Diary2026.06.09

If changing the setting changes nothing, that is a defect (Part 4)

Most of the 36 defects in the third GS certification report were node options. Repairing the broken ones and making the invisible ones observable.

유지수

InsightInsight from our research and the field

The tools our agent built for itself were not canvas nodes