Contact Center Assistant

GIF
Northwell Agent Assist

Inspiration

The inspiration for my project comes from the complex selection process residency and fellowship applicants face when applying through ERAS and MyERAS. With multiple steps and eligibility criteria, the journey to securing a spot in a program is often intricate and time-consuming. I see immense potential in leveraging Agentforce to streamline this process, ensuring a more accurate match between applicants and programs while reducing manual effort. By integrating AI-driven insights, we can make the selection process more efficient, fair, and candidate-friendly.

In my use-case I have tried to emulate a similar scenario where Pre-med Applicants apply to a list of summer health/educational programs available on an institution called Northwell Medical Organization's website. Applicants apply to those programs by submitting a case via email. Contact center agents are responsible for handling these requests and add interested students/contacts based on eligibility criteria for each program. The way they achieve this is by leveraging the power of Agentforce to make informed decisions of why a particular applicant will fit the elected program.

What it does

Contact center agents working in Service Console receive cases from applicants. Contact center agents use the internal employee agent to fulfill the request by making informed decisions through natural language processing. Contact center agents request a summary on the applicant and using Agentforce's power to search internal knowledge articles, reading contact's interested programs and making sure they have correctly applied to the program. Once agentforce deems the applicant is eligible, contact center agent can request Agentforce to add the contact to the respective program and inform the student that they are registered and proceed with payment instructions. Agentforce also helps making informed decision for the agent to effectively categorise a case, lookup knowledge articles to resolve other types of inquiries.

How I built it

Leveraged Agentforce employee agent, Knowledge article integration,DataCloud, Prompts, Instructions, email-to-case, and Autolaunch Flow to lookup contacts by email and add a contact to a Program. Agentforce uses topics such as General CRM inquiries, FAQs and custom agent actions.

Challenges I ran into

I noticed if I used too many topic actions for a particular topic agentforce's actions were not precisely selecting the correct action. In order to make sure the correct topic action was being selected I only added actions that were specific to the request. During testing of my agentforce agent I ran into some challenges with the output generated from the agent's responses if the agent wasn't able to produce a desired next step. Debugging the flow and using the agentforce testing center really helped in pin-pointing how to write the output instructions for the agent action.

Accomplishments that I am proud of

I am proud of the simplicity of natural language communication that agentforce uses to communicate thoughtful responses and how effectively it can navigate through different topics to pick the right action based on the conversation.

What we learned

Testing is very crucial in ensuring an agent will be able to process the conversation or request. Overloading the agent with too many topics could produce incorrect outcomes.

What's next for Contact Center assistant

A customer facing agent will be able to assist applicants with their interests via Chat and be able to guide them on how to apply and if they fit the criteria for being selected into a program. The agent will be able to suggest other programs to applicants if they are not eligible for a particular program that they are trying to apply.