SHC app

A Voice Agentic Solution

The Client.

NewPartners is a consulting service that helps various organizations with campaigns and public outreach, using methods such as teleservices, fundraising, and polling.

Challenge Image

The Challenge.

The client sought an AI solution to help streamline their telephone fundraising campaigns. Much of their campaigns involve routine phone calls to past donors or campaign participants, where live agents call customers and follow a specific script to give information about the campaign and then ask for a donation. The agents at NewPartners were seeing an inefficient use of their time, and were limited in the amount of calls they could make in a day.

The Solution.

We developed an AI calling agent using BlandAI for the client to use in one of their ongoing fundraisers. We modeled the conversational pathway of the AI agent by using the real script that the client used for the campaign. We identified 3 main customer profiles, and specified the conversational pathways for each based on the most common types of user responses for each customer profile. Each conversational pathway is integrated with DialedIn, the client’s call center software, in the case where a customer needs to connect with a live agent.

The Impact.

1. Saved the client 1000 man hours a month from calling the members.
2. AI Agent provided a cost saving of $25K a month.

Tech Stack


BlandAI

BlandAI


DialedIn

DialedIn


OpenAI

OpenAI

AI Agent Features.

Customer Profile Pathways

Customer Profile Pathways.

  • Customized calling pathways based on the type of customer
  • Pathway 1 for Previous Donors follows the script used for donors that have previously given to the current campaign
  • Pathway 2&3 for Potential Donors follows the script for users that may have an interest in the campaign or are completely new to the call list

Batch Upload the Call List

Batch Upload the Call List.

  • User can upload a spreadsheet with metadata about the customers
  • Metadata is inserted directly into the conversation pathway, so the AI Agent is able to address the customer by their first name
  • AI Agent is able to make multiple calls to all users in the list, much quicker than a human agent

Custom Nodes & Conditions



Custom Nodes & Conditions.

  • Custom nodes in each conversational pathway determine the dialogue of the AI Agent
  • Common customer responses were defined and help direct the pathway from node to node, based on the real-time customer response

Improvisation & Unknown Variables



Improvisation & Unknown Variables.

  • Conversational pathway built with expected and unknown variables based on the customer responses
  • Customized improvisational capabilities of the AI Agent
Connecting to Live Agents



Connecting to Live Agents.

  • Custom pathways designed in case the customer needs to connect to a live agent
  • The AI Agent can identify when it is appropriate to connect to a live agent based on customer responses
  • The AI Agent can end the session and transfer the customer to a live agent via DialedIn
  • The AI Agent can send a report to the live agent with analytics from the call
Analytics



Analytics.

  • User can view call analytics to help improve AI Agent performance
  • User can review the real conversation between the AI Agent and customer
  • User can can see the AI Agent’s logic when deciding which node to switch to based on the customer response