JD for structure, OKF for content

The Johnny.Decimal system System expansion
1

(Way down the rabbithole of RAG, knowledge graphs, embedding - I get this pop-up in one of my feeds with this article).

Google Cloud recently proposed this - OKF ( Open Knowledge Format)

As I’m learning more about AI (what works and what does not) - I’ve already tipped over to the persuasion that companies who will succeed long-term in using AI to augment their work and products without breaking the bank, are those who heavily structure and keep their data in dual “agentic” and “human” readable format, preferably vendor-neutral.

(You can swap the brain (model/LLM), the harness, the infrastructure can grow and evolve - but the company needs to be able to describe how the company is (supposed) to work. It is the contents (words) that matter.)

I sat down, read the spec for OKF - and I see a relationship to JD.

I imagine:

  • JD to organize and structure
  • Content (under JD ID’s) as OKF
1 Like
Manage an Index in Obsidian with Stashpad
2

Interesting, will read over the next couple of days.

I have a task to spend some ‘deep work’ time thinking about the separation of ‘JDex’ – metadata – from ‘notes’ – freeform text. The more I use my own system the more it’s something I crave. This might help?