@Anirudha @PravinJ Congratulations on Artham. Raise Financial unveils Artham, India’s first SLM for capital markets
Having used Fuzz for sometime now, apart from much needed UI enhancements the responses from Fuzz look clean.
3 Likes
Hi @thisisbanerjee , thanks for the detailed feedback. This is much appreciated from early users like you as we try to scale fuzz.
For this specific question, since SKF India Industrial is not listed yet, we were unable to get enough verified sources to answer this to the best possible manner. I ran the same query on the other product highlighted in your screenshot; and the sources were random (essentially reaching a state where the user has to “accept” the answer or re-verify the sources one by one). fuzz will bias towards verified sources always.
For example, on your Dabur question (great question btw!), it gave the answer directly from the annual report (suggest you use desktop version for the best experience for opening source-based citations in-app). Sharing the screenshot here as well. The other apps mentioned could not.
For the other areas, I’d be happy to get on a call to understand specifically why you were disappointed and give you clarity on where we are on our way to solve them!
@thisisbanerjee You are burning up Fuzz’s GPUs 
. As I support an AI assistant, Fuzz works well in assisting human assessment e.g:- instrument choices, technical analysis given a trend and market-scenario-based impact. Considering that Fuzz is still new and Dhan is leveraging the Small Language model, it will get better as time goes by.
3 Likes
You’re absolutely right! Both of you are valued members of this Community and I’d take all the feedback I can get!
Let me get back to you on the following @thisisbanerjee :
- How we can capture other documents beyond our current universe
- Calculation methodologies
- Why it missed the toll-road impact in this specific instance
Just like we started with Dhan, we’ve made a dedicated feedback@askfuzz.ai email as well for all of this. Let me continue the thread there based on all of your inputs!
1 Like
@thisisbanerjee I am not criticising your points 
. I just mentioned the technical details. It was interesting to know that Dhan created their own model and they went with a SLM which has very low parameters. They created their own model Artham with a very low parameter set so that the responses are lightning fast and its not just prompt but they have pre trained the model as the base which is very interesting. Hence I mentioned it will take time as this model will just get better with the iterative approach only and with regular improvements.
1 Like
@Anirudha @Dhan Has your Team forgotten about the UI Enhancements needed in Fuzz? There has been no improvement on the UI and even now I cannot delete chats which are just piling up.
Hi @nitishbangera , all your UI feedback has already been incorporated:
- Delete/pinned chats is live.
- Clicking on an old chat does not close the side panel any longer.
- Clicking on an old chat now jumps to the end of the message instead of the last line.
Can you please refresh and check these?
Neatly done and delete works fine. Thanks for incorporating the changes.