FW: LLM Demo - Feedback

 Kishan ,

 

This is a very valuable feedback . It will help you to perfect “ Executive Zero-in “ tool , with which , BRIHAS would be able to exceed the expectations of your customers

 

Good luck

 

Hcp

 

From: Mitchelle Shetty [mailto:mitchelle@3pconsultants.co.in]
Sent: 03 November 2025 16:20
To: Hemen Parekh <hcp@recruitguru.com>; Nirmit Parekh <nirmit@3pconsultants.co.in>
Cc: Kishan Kokal <kokalkishan.official@gmail.com>; Mahendra Tarkhad <mahendra@3pconsultants.co.in>; Prashant Sharma <prashant@3pconsultants.co.in>; Manish Chauhan <accounts@3pconsultants.co.in>
Subject: LLM Demo - Feedback

 

Dear Sir,

 

We had a demo discussion with Kishan today. Below are the observations we shared with him:

 

• While creating an assignment, the JD that needs to be attached can only be uploaded as a PDF document by the LLM. Word option could be explored.
• Company website to be auto picked
• Target list view to be in table format
• The workflow initially identifies target companies based on the JD; however, the list of target companies that appears is limited and not comprehensive.
• The LLM then identifies candidates only from these target companies, which restricts the search results to a few executives.
• During a few test runs, the system identified two candidates for BSH who matched our recommendations. However, for other test cases such as Blue Star/Manufacturing and Japfa – Head D2C, none of the recommended executives appeared in the search results.
• When you run parallel searches; the time taken is long. Speed will have to be better else the Recruiter will lose interest
• When we did a test run for Blue Star; the LLM created target list of companies of durables however the client was keen to identify talent from automotive. Automotive companies did show up in Tier 2 companies, however it could not identify the candidates recommended.
• We have suggested to Kishan that there should be a box where we can add inputs, free text, what to include or exclude like the earlier User interface shared by Prashant as each search will have some nuances and thus the input logic being hard coded needs to ape the recruiters mind for more relevant and comprehensive results.

 

Kishan is reworking on the same basis these inputs and will target to show us a revised demo by Monday, 10^th November

 

Thanks,

 

Mitchelle