Is generative AI good for Viewing a concise summary of all the cost generating events on the public cloud and View a high level recommendations to optimize the cloud cost
Welcome to the forum!
What would be the approximate size of the dataset you intend to look at?
Are we taking a couple of pages of text or hundreds?
1 Like
Its hundreds of pages of text. The product is public cloud cost management platform which has 1 PB of data per day with cloud providers dropping bill and asset data 2-3 times a day.
That would make for a very interesting use case, it would need to be split into manageable chunks for the AI to process, the difficulty is keeping contextual links across those chunks and reconstructing the results to also be meaningful.
Non trivial is the quick answer, and it would need time spent on it looking at the data and any potential ways of splitting it.
A hybrid of conventional code with AI management agents… possibly… large project for sure.
Does it make sense to use Generative AI here or typical predictive ai provides better quantitative insights. GenAI models I believe are primarily focused on data generation and capturing patterns , if we are looking for generating human-readable explanations and interactive recommendations then genAI is good.
The GPT models can certainly be instructed to produce concise summations of complex data, I guess the real question is, is it the best at it. I don’t know of anything currently that is better.
When you mentioned 1Pb of data, is that data that needs to be parsed by the analytical system? If so that does tend to lean itself towards conventional deterministic programming solutions able to operate at speed and with near 100% accuracy.
If the AI side is only going to be looking at key metrics from that data and give insightful trend analysis and the detection of anomalies then that is an ideal match for the abilities of GPT level LLM’s