8 Comments
User's avatar
Amit Agarwal's avatar

Love the breakdown

Moiz Ali's avatar

Love it. Are there any courses you would recommend that explains how to build AI ready data?

Pradeep Sharma's avatar

What is one example of AI-ready data? A specific sample that you have come across in the real-world implementation

Aaliyah Hicks's avatar

This is such a good read. The explanation was so digestible! Thank you for that.

Gildas Trébuchet's avatar

Hello,

Thank you for your article, which is impressive and enlightening. Thank you.

However, I have a question! There is increasing talk of LLM being integrated into BI dashboards, which would support users in making decisions based on BI (analytics) and reinforced by LLM queries.

In these cases, how can the two worlds be reconciled? Does the LLM query the AI part only? How can the junctions between the AI part and the analytics part pipelines/semantic layers be managed?

Do we need to find a third layer, which would be the ‘semantics of analytics’, to allow simple language models to query analytics only? And keep an AI-ready semantics model geared solely towards AI for in-depth analysis cases?

I'm not sure if I'm making myself clear...

Irina Malkova's avatar

I wrote a similar post a bit ago: https://substack.com/@irinamalkova/note/p-174646404

2x2 matrix is 🔥!

The other big difference between the two is what is measured. I find that most analytics focuses on $$, whereas AI focuses on moving some workflow forward and needs behavioral events / telemetry and user attributes for personalization.

Ashish's avatar

Loved it! The definition of ready is often misunderstood.

User's avatar
Comment removed
Jan 6
Comment removed
Animesh Kumar's avatar

Absolutely, thanks for the insights!