I think its just a matter of audience? Of course report generation involves many things, including RAG/retrieval. These are implementation details.
But the end value being communicated is the generated report, which is what non-technical people would care about
yes but people get swayed easily nowadays especially in this space
plus it was obviously a marketing technique to get people to sign up for his RAG improvement email thing
For sure. But I don't think that takes away from the fact that RAG on its own is small peanuts if its not being used to solve an actual problem/use-case
RAG is part of your infrastructure. just like a cloud s. iThe compute vm you rent s never your end product.. that doesn't make it irrelevant and ahve 0 value in the value chain of activities
a report is just a deliverable . it's not the process
that's like saying a car's chassis is peanuts compared to its engine. Well it isn't . It's an important part of a the car's foundation and it won't work without it
and you're talking to a guy who worked in corporate alongside c-suite folks for 12 years so i know how it goes really well
not really preaching you out of my ass 😄
The wording was wrong from the perspective where i come from.
(1) He's not comparing apple to apple but that's totall fine.
(2) The fact that he used it as a bait/hook to promote his RAG email course, is an overused digital ad technique they teach smbs. But in corpos this is frowned upon. It's an overused technique that isn't really looked at positively.
The reason i'm highlighting this , is that fake experts who are all over the place and trying to market themselves abuse these same techniques. He doesn't wanna mistakingly start getting perceived in a similar manner. I know he's a very knowledgeable guy and knows his shit. that's why the tweet jumped at me.
and I hate the word RAG . It undervalues the role retrieval plays in the pipeline. Information retrieval is a major problem (even if solved) at companies and it grows larger with growth of organizational size.
again it's about hte process not the deliverable. The problem corporations have and that you're solving is not in the format.
The report is just the final deliverable and is only 10% of the effort required to solve the problem.
That's why comparing reporting to RAG is not a proper comparison.
just my 2 cents from someone who reported a bit too much in his past corpo life 😄
and I apologize if I came on too strong. I just appreciate the work you guys are doing.
Yeah I see where you're coming from. Preparing a report likely requires context, and that context needs to be retrieved from somewhere, so in that sense it doesn't seem like a comparison of RAG vs Report gen.
I think this commentary was more intended to be about LLM use cases: "move over chat, here's report gen (and data extraction) as other LLM use cases to think about"
yes chat vs report. I totally agree
upper management don't want chatbots
but that's not a new discovery if you've been working in that space
I mean 90% of clients i have want a reporting format
and I still use semantic search especially where the data is scattered and is mostly unstructured
which is the case more often than not
ok more liek hybrid... but yes
albeit chatbots still useful for people conducting searches etc..
but anything which goes on a c-suite desk is more often than not a report
