House of Dentistry Inbound AI Agent
Key insights about the house of dentistry project
Definition and Analyzation:
In the project'sfirst step, we defined tasks the AI Agent should take over. We watched TheHouse of Dentistry business in detail. From how they work to how appointmentsare booked. After we watched their business in detail for two weeks andanalyzed everything, we started to develop tasks that AI could replace.
We cameup with the following tasks the AI can do:
Development of the Agent:
After wefinished analyzing all parts of the business, it was time to start collectingdata and information start with the development of the AI Agent. We needed allthe information to train the AI Agent and learn the ins and outs of House ofDentistry.
We developed the first version of the AI Agent. It took a couple of days todevelop the agent, but we were convinced that we can deliver amazing results.
Here you can find our first voice recordings after completly setting up our first version of the Agent:
Setup and testing:
Thefirst version of the AI Agent was already a good start, but there were still lotsof things to improve. We tried out various voices, changed the scripts athousand times, and started live testing the AI bot.
See the voice recordings below:
Fine-tuning:
Afterthe first tests we noticed that there are still a lot of things to improve. Sowe did more tests, improved all aspects of the AI Agent until it could handle anytelephone conversations without problems.
Of course we tested it several times to make sure there aren’t any situations ourAI Agent can’t handle.
Checkout the voice recordings below:
Results and expectations for the future
Exspectations and Results:
The AI Agent was in the Beta Testing Phase for The House of Dentistry throughout March and April. Checkout our following results for March and April and our expectations for the nextmonths:
As shownin the pie chart above, the AI Agent cost only around one third of a humanassistant. While a human costs the company around 4000 USD per month, themaintenance and fees for the AI Agent are 1300 USD per month.
The linechart above shows the time-effectiveness of a Human compared to the AiBell AI-Agent.
We discoveredthat a human only works 40% of the time effectively. This has many reasons, themain reason for this being distractions like social media.
We knewour AI-Agent could top 40%, as it can’t be distracted. It may not do as well atfirst as a human with years of experience, but we knew that AI had a muchfaster learning curve that we could scale infinitely.
As youcan see, especially in March and April (Months 1 and 2), the AiBell AI-Agentwas barely effective. This was due to gathering all necessary information, and developing,and testing the AI. It only worked about 15% efficiently.
It yetcouldn’t compete with a human that had years of experience. We knew there wasstill a lot of work to do until it could outwork any human.
In Monththree (May), the effectiveness of the AiBell AI-Agent increased rapidly. As westarted to test the AI-Agent more and more and adapted the scripts andinformation, we saw massive increase in effectiveness. From 15% in the firsttwo months, it more than doubled to 35% and almost passed a human ineffectiveness.
Thecrazy thing about this is that we tested the AiBell AI-Agent for 2 hours daily.We knew if we increased the daily testing hours to 4 and more, we could pass 40%effectiveness easily. And we did just that.
In thefourth month, we tested the AI-Agent 4 hours per day and immediately increased effectivenessagain to 55%. It was time that the AiBell AI-Agent was integrated completelyinto the business.
For the rest of Q2, we expect the AI Agent toreach an effectiveness of 65%. In Q3 & Q4, it is expected to reach up to90% effectiveness.
Thechart above shows the business growth of The House of Dentistry since westarted implementing our AiBell AI-Agent in their business.
Similarto the effectiveness chart, in the first two months The House of Dentistrydidn’t see any huge difference in growth with the AiBell AI-Agent.
Anyhow in the third month The House of Dentistry grew 10%. They took on 10% moreappointments, which led to an increase in revenue of 15%, as most of theappointments were higher paying clients.
In month4 (June), as we started to implement our AiBell AI-Agent further into thebusiness, The House of Dentistry grew 30% in appointments and 45% in revenue,which was far beyond our expectations. They even had to decline appointments asthey are fully booked out by now.
This isonly the beginning. We knowthat the possibility of growing The House of Dentistry 200% this year is givenwith our AiBell AI-Agent. Though, we are going to create a detailed roadmap forQ2-Q4, which will lay out the business’s growth plan for this year.
