Inspiration
We wanted to explore the Agentforce platform and the capabilities it supports by creating a real-life project. This Hackathon gave us that opportunity to go into the depth of the platform and work on a hands-on development project.
What it does
As a Morgage AI assistant, it helps customers to assist on various tasks related to mortgage:
- Answer customer inquiries related to Home Loan, Personal Loan etc.
- Clarify rules and regulations
- Help in explaining the Loan application process
- Assess a customer’s eligibility for a mortgage
- Compare mortgage rates from various lenders
- Advice customer on the best deals
- Book an appointment with Financial Advisor
- Modify an existing appointment
- Cancel an existing appointment
- Help to create support ticket
- Escalate/transfer to the human agent for further
How we built it
- We have used Agent Builder to create a service agent to perform all the tasks mentioned above.
- An Experinced Cloud site is created and the service agent is embedded in it.
- Omni channel inbound and outbound routing are configured to connect the service agent with the Experience Cloud site user and also to transfer the call to the human agent.
Challenges we ran into
- Tried to setup the Salesforce Scheduler features by configuring the standard objects and importing test data but couldn't make it work fully without the Salesforce Scheduler licence on this org.
- Setting up the different help topics (like loans and appointments) in MortiAI took careful planning.
- Giving MortiAI clear instructions for each action it needed to take was important but sometimes tricky.
- Figuring out how to use the Einstein Data Library to store and find the right information took time.
- Creating test data to make sure MortiAI worked correctly for all situations was a big task.
- Working out the right formulas and steps for calculations, like the EMI, needed careful thought.
- Making sure MortiAI followed the correct logic for things like checking if someone can get a loan was important.
Accomplishments that we're proud of
- We built an AI that can help people find mortgage options and start applications.
- MortiAI can answer general questions and also do specific things like EMI checks.
- We made sure new customers can sign up safely with email checks.
- MortiAI can connect users to real people if things get complicated.
- MortiAI can find the right information quickly from our document library.
What we learned
- Structuring custom topics and actions effectively is key for organizing an AI agent's abilities.
- Clearly defining the instructions that link user requests to specific automated processes is crucial for functionality.
- Understanding how to best organize and retrieve information from the Einstein Data Library significantly impacts response accuracy.
- Generating comprehensive test data is essential for validating the reliability of the AI agent's logic and calculations.
- Designing the flow of logic to handle various user scenarios and decision points is a fundamental aspect of AI agent development.
What's next for MortiAI - a Mortgage Broker AI Agent
- We would like to use this prototype solution to demonstrate to our prospective clients working in Financial, Insurance, Telecommunication domain to see what all opportunities we have to implement agentforce solutions for them.
- If we can engage and collborate with wider teams within the organization and receive good feedbacks then we can extend more features and explore other areas that we haven't done as part of this Hackathon project.
Log in or sign up for Devpost to join the conversation.