What exactly did Toyobo change to improve accuracy?

It explicitly paired keyword search with semantic search for a 0.57 score, then widened the number of retrieved sources from 5 to 60 to reach 0.86.

Why did semantic search alone fail?

Factory terms like "B1A3" are abbreviations with no general meaning, so meaning-based search alone could not capture them.

What platform runs the agent?

The core is Amazon Bedrock AgentCore, with services split across agent runtime, user authentication, answer generation, dialogue memory, term-dictionary storage, document search, and log analysis.

Large Language ModelsTop Companies' AI MovesAI Business & IndustryTop Companies AI6d agoOct 10, 2026, 06:30 JST

Toyobo lifts AI answer accuracy to 0.86 with agentic RAG

Send on LINE

3 Key Points

What happened
Toyobo's team scored 0.57 on accuracy after pairing keyword search with semantic search, then reached 0.86 by raising retrieved sources from 5 to 60, and built the agentic RAG on Amazon Bedrock AgentCore.
Why it matters
Because factory shorthand like "B1A3" has no general meaning, semantic search alone missed it, so the improvement suggests AI quality depends on how well a system learns a site's own vocabulary.

WHO IT HITSManufacturing and plant IT teams running internal document search now have a concrete case that domain jargon — not model choice — was the bottleneck, and that data-entry discipline drives AI answer quality.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

FAQ

What exactly did Toyobo change to improve accuracy?: It explicitly paired keyword search with semantic search for a 0.57 score, then widened the number of retrieved sources from 5 to 60 to reach 0.86.
Why did semantic search alone fail?: Factory terms like "B1A3" are abbreviations with no general meaning, so meaning-based search alone could not capture them.
What platform runs the agent?: The core is Amazon Bedrock AgentCore, with services split across agent runtime, user authentication, answer generation, dialogue memory, term-dictionary storage, document search, and log analysis.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.