AgentForce: Smart Interview Assistant

Inspiration:

Interviewing is a critical yet time-consuming part of student selection process, especially in academic and early-career settings where large volumes of candidates need to be evaluated quickly. We wanted to bridge the gap between intelligent automation and human decision-making by empowering interviewers with AI-generated, personalized interview questions. AgentForce: Smart Interview Assistant came from the desire to create a more objective, efficient, and personalized interview process. By leveraging AI, we aim to streamline the interview process, make it more insightful, and reduce the cognitive load on interviewers, ultimately leading to better student selection decisions for academic courses.

What it does:

Agentforce - Smart Interview Assistant is an intelligent, AI-powered platform designed to revolutionize the way interviews are conducted. It analyses a candidate's profile and generates tailored, insightful questions to help interviewers engage more effectively. When interviewers (academic counsellors) are reviewing candidates on the Salesforce Experience site, they can click a "Generate Questions" button, triggering a screen flow that calls an LLM (Large Language Model) to dynamically create a set of personalized questions based on the candidate's background. Additionally, there’s an option to generate a PDF of the questions, which is then stored in the student's Notes and Attachments section in Salesforce for easy access and record-keeping. This ensures the interviewer have a documented set of questions, enhancing the interview process’s transparency and organization.

How we built it :

We built AgentForce - Smart Interview Assistant using the following key components:

Salesforce Lightning Web Components (LWC): The front-end was developed using Salesforce’s LWC framework to provide a smooth, interactive user interface.

*Salesforce Flow: * We created a custom Screen Flow to trigger AI-based question generation when users interact with the "Generate Questions" button.

Prompt Template & LLM: For the question-generation logic, we created Prompt Template that uses an LLM to process the candidate’s profile and generate relevant, personalized interview questions based on the contextual data.

PDF Generation: We leveraged Salesforce’s PDF generation capabilities (VisualForce rendering) to allow counsellors to save the generated questions as a PDF, which is then uploaded to the candidate’s Notes and Attachments section for future reference

Challenges we ran into:

*Personalization of Questions: * The AI had to generate questions that were not only relevant but also contextually aligned with the candidate’s profile. Ensuring high accuracy was a complex task. we ensured that we have achieved the same using finetuned prompt instructions. *PDF Generation: * The ability to format and export the AI-generated questions into a structured PDF format that was easily understandable took some trial and error.

Accomplishments that we're proud of:

Successful integration of Salesforce’s Agentforce AI to generate personalized, context-aware interview questions based on candidate profiles.

*Efficient Workflow: * We created a streamlined workflow that saves time and ensures consistency in the interview process.

*Innovation in Edu Tech: * By blending AI with Salesforce, we’ve created an innovative tool that transforms student selection process, making it more efficient and less biased.

*User-Centric Design: * We’re proud that the platform is intuitive and easy to use, making it accessible for both tech-savvy and non-technical users alike.

What we learned:

We deepened our understanding of Salesforce's AI capabilities and prompt engineering and how it can be integrated into custom applications. Gained insights into creating intelligent workflows and how AI can enhance traditional processes.

*AI’s Role in Personalization: * We gained deeper insights into how AI can be used not just for automation, but for creating personalized experiences based on data. *Problem-Solving: * The process of troubleshooting integration challenges, particularly with AI and Salesforce, deepened our technical skills and helped us better understand how to marry the two technologies seamlessly.

What's next for AgentForce: Smart Interview Assistant

We’re excited to take AgentForce to the next level. Our upcoming features include: *Candidate Evaluation: * We plan to capture and store candidate responses to the generated questions, allowing interviewers to evaluate the answers more effectively. *Answer Probability Scoring: * We aim to integrate an evaluation mechanism that calculates the probability of a candidate’s answer being correct or aligned with the expected response, providing deeper insights for interviewers. *Advanced Analytics: * Future versions will offer analytics on interview performance, helping academic counsellors to track trends and identify top performers across multiple candidates. We envision Agentforce evolving into a comprehensive AI-driven tool that not only helps interviewers create personalized questions but also assists in evaluating and scoring candidates more accurately, making the student selection process faster and smarter.

Built With

• agentforce
• ai
• apex
• flows
• lwc
• promptbuilder
• vfpage