As you know, cost of acquisition is a major concern for solar companies across the spectrum. Whether you are the size of Sunrun or a smaller mom and pop, it costs a lot of money to acquire customers. We thought it'd be great to see if it's possible to qualify a homeowner without talking to a person.
What it does
It uses api.ai's natural language processing to classify and then qualify or disqualify the most critical solar qualification criteria
How we built it
We created a dynamic conversation flow that maps a user through a standard lead qualification flow, while gathering lead information along the way via a customer lead engine. To make the conversation natural and conversational, we used a Natural Language Processing AI library, which we wrote a customer wrapper around.
Challenges we ran into
- Just getting the AI to talk at all was very difficult
- Styling the application with the non-standard small chat window was tricky
Accomplishments that we're proud of
- Classifying and getting the conversational flow to work (and seem natural) without any technical experience.
- Run from concept to proof of concept in 10 hrs.
What we learned
Natural language processing is going to change the world
What's next for QUAL-E