Michael G.

You could ask them if they care if the get the right answers from that approach; many studies like this make it clear that the kind of process your colleagues are advocating for is only really viable for a use case where being wrong is ok. GenAI did not change the rules of data mgmt. Agents need good data as much or more than their human analyst counterparts. Would your business analysts be happy with some markdown files? I suspect not. They are used to enjoying the advantages a data mgmt and governance program affords them as they perform their job. Treating AI like a second class citizen will yield second class insights at best, and often incorrect ones that the AI is confidently wrong about.

I am not very familiar with either system, but both are new. I've never heard of anyone using them. I'd be willing to bet we're more scalable given we're a much more mature product. Arcade is open source, so its free. so it's probably your cheapest option. puppy will end up being expensive because they charge by CPU and compute resources, so if you want performance, they will break your bank. we charge only by usage.

Stardog is the best option out there, disclaimer, I work for Stardog. We have a free tier for exploration and will be rolling out some other options in the coming months.