The analogy table really helped me understand the relationship between ontology and knowledge graphs. It was hard for me to visualize these concepts before but now I understand how they both fit into graphRAG!
Shared it with my fellow terminologist. We are super interested into exploring this in 2026. Thing is, at times I feel like I don’t necessarily have a visible problem that ontologies/KGs would come to solve for me, nor how I could feed them into our existing, non-LLM workflows. Have you been able to see any improvements in any important business metrics thanks to KGs?
Hi Gonzalo, will DM you and maybe we can jump into a call. I can give better advice once I get to understand your current workflows.
I have personally implemented it for multilingual content creation (see the post about Blackbird.io) as a replacement of traditional translation workflows for certain type of content (blogs and other marketing oriented material).
I have also implemented it in a knowledge chatbot which has saved hundreds of hours for different teams which used to email doubts on market availability and approved translations.
Insightful!
The analogy table really helped me understand the relationship between ontology and knowledge graphs. It was hard for me to visualize these concepts before but now I understand how they both fit into graphRAG!
Shared it with my fellow terminologist. We are super interested into exploring this in 2026. Thing is, at times I feel like I don’t necessarily have a visible problem that ontologies/KGs would come to solve for me, nor how I could feed them into our existing, non-LLM workflows. Have you been able to see any improvements in any important business metrics thanks to KGs?
Hi Gonzalo, will DM you and maybe we can jump into a call. I can give better advice once I get to understand your current workflows.
I have personally implemented it for multilingual content creation (see the post about Blackbird.io) as a replacement of traditional translation workflows for certain type of content (blogs and other marketing oriented material).
I have also implemented it in a knowledge chatbot which has saved hundreds of hours for different teams which used to email doubts on market availability and approved translations.
Great breakdown! Would love to know your point of view on chunking and embeddings in this context 💡