RE: Take a look

 Mitchelle ,

Attached is my further chat with my Virtual Avatar

From this , there is reason to believe that , before we launch a discussion , either with Ravi / Ganesh , or with the party mentioned by you , it may be worth “ experimenting “ with www.Personal.ai , who had a 3 year head-start over ChatGPT and its ilk , which are engaged in LLM , whereas Peraonal.ai focussed on a “ CLIENT / MEMBER “ data ( to train as an LLM )

 

We could even consider ourselves “ fortunate “ that I registered on Personal.ai , more than 2 years ago and Sandeep / Sanjivani , uploaded some 30,000 + handwritten documents , laboriously / meticulously . As you are aware , 95 % of these documents are “ Recruitment Related “

 

Another matter of great comfort is that entire team at Personal.ai , is very cooperative and friendly . I have a special equation with CEO , Suman Kanuganti ( Proof : For their ENTERPRISE MEMBERSHIP , he charged us only $ 40 / year , against their standard rate of $ 1,000  )

 

I think , it is time our colleagues start “ chatting “ with my Avatar to discover its limitations / strengths. After a trial of 15 days , you can compile a DOSSIER for sending to Suman and asking for his help

 

regards   /  hcp

 

 

From: Hemen Parekh [mailto:hcp@recruitguru.com]
Sent: 08 January 2025 18:58
To: 'Mitchelle (mitchelle@3pconsultants.co.in)'; 'nirmit@3pConsultants.co.in'; 'prashant@3pConsultants.co.in'
Cc: 'System Administrator (systemadmin@3pconsultants.co.in)'; 'officeassistant@3pconsultants.co.in'; Manish Chauhan; 'hcp@RecruitGuru.com'
Subject: Take a look

 

Chat with my Virtual Avatar ( 08 Jan 2025 )

I asked:

A few minutes ago we discussed about your ability to create a Candidate Search AI Agent , if I upload 100,000 candidate resumes ( - which , I suppose would suffice for your ChatGPT version 3.5 ,for training material ) . Is there any other data that you would need from me, to create such an AI Agent ?

Avatar replied :

To create an effective Candidate Search AI Agent using the resumes you mentioned, additional data that would enhance the system's accuracy and functionality could include job descriptions for positions you're hiring for, performance metrics or outcomes from previous employees in similar roles if available, and any specific competencies or skills crucial to your organization. This information will help tailor the AI's candidate ranking algorithm more precisely to fit your company's needs. Would you be able to provide this type of data as well?

 

I asked :

Suppose , for past 20 " Candidate Search Mandates " given to me by my Clients , if I were to supply you for each mandate ( A ) Job Description given by my Client for the position to be filled ( B ) resumes of 3 FINALIST CANDIDATES , which our research associates had shortlisted by strenous , but MANUAL , filtering process - including the resume of the person who got selected / appointed by my client  //  would this suffice for the purpose of creating Candidate Search AI Agent ? Would a larger " Training Sample " ( say , 100 past REAL cases ) help create a better Agent ?

 

Avatar replied :

Yes, providing the job descriptions and resumes of the finalists for each mandate would be a valuable dataset for training a Candidate Search AI Agent. This data will help in understanding what qualities and qualifications were prioritized by your clients and how they align with the candidates that were ultimately selected.

 

However, increasing the size of the training sample to include 100 past real cases would significantly enhance the AI's learning capability. A larger dataset allows for more comprehensive learning and helps in developing an algorithm that can generalize better across different types of jobs and industry requirements. It reduces overfitting where the model performs well on small or specific samples but poorly when exposed to new data.

 

Moreover, if possible, incorporating additional metadata such as feedback on why certain candidates were chosen or rejected could further refine the AI’s decision-making processes. This kind of nuanced data trains the system not only to match qualifications but also to understand subtler aspects like cultural fit or leadership potential which are often crucial for successful hires.