EVA - App2Check

EVA is a patent-pending platform that allows non technical experts to create chatbots

in an intuitive way and very low effort compared to the traditional providers
with higher accuracy
and a straightforward maintenance of the chatbot Knowledge Base (KB) like a CMS

EVA platform allows you to create a Chatbot that

is able to instantaneously retrieve the correct and standardized solution to the issues or the right information about product/service
can dramatically improve the overall process efficiency
deliver a faster and better support experience for the customer

EVA Building Features

Learning from a list of possible anwers in an unsupervised way
Turning a FAQ into a chatbot in seconds → Patent Pending
Chatbot Supervision only where needed and with no technical skills required
Learning from unstructured textual documents
Creation of a structured KB from an Unstructured Textual Document → Patent Pending

EVA can answer

is able to instantaneously retrieve the correct and standardized solution to the issues or the right information about product/service
can dramatically improve the overall process efficiency
deliver a faster and better support experience for the customer

Traditional approaches

For each answer/intent to recognize is necessary to provide multiple examples of questions
Top-down, supervised approach, since the project kick-off: it needs to try to imagine many of the customer requests in advance to show them for training the chatbot
Updating the knowledge base or correct the chatbot often requires technical skills to retrain/publish a new version
Given N Answers and M questions associated to each answer, N*M items are needed to start a chatbot project

Example: reset password intent
Answer 1: It is possibile to change your password by clicking here and then filling in all of the fields. You’ll receive a confirmation email.
Question 1: I forgot my password, how can I change it?
Question 2: How can I modify my password?
Question 3: My password is expired or anyway I cannot login the system. What can I do?
Question M: ….

EVA approach

It is possible to start with the list of answers that the Virtual Assistant must know or importing an available FAQ
Updating the knowledge base (adding or modifying answers/questions) does not require any technical skills and is done in a wiki-like or CMS fashion.
Any change in the knowledge base is instantaneously published and ready to the final user
A list of N answers is enough to start a chatbot project with our approach

Example: reset password intent
Answer 1: It is possibile to change your password by clicking here and then filling in all of the fields. You’ll receive a confirmation email.
(Optionally provided Question 1 is already available): how can I change my or renew my password?

Our approach is

Unsupersived at the beginning

It is not necessary to show multiple examples to start
EVA automatically creates a semantic representation of the list of answers
Each question is answered by the most semantically close answer

Supervised through corrections to improve performance

Customer care operators can correct EVA directly while chatting with her and EVA improves along time her performance, trying to generalize the corrections
Corrections can be applied on the base of the past customer requests/tickets
Corrections are focused on the customers’ needs expressed in the real world!

6 good reasons

to choose our Virtual Assistant and CHATBOT solution

1. Innovative

Innovative machine learning-based approach overcoming the usual limitations and high effort required by traditional approaches

2. Accurated

High level of accuracy (ability to retrieve the right answer) close to the 90% on a daily basis, reachable after 1 month from the project kick-off

3. Self-service

EVA Knowledge Base can be easily maintained and updated in a self-service fashion by the user

4. Low effort

EVA Knowledge Base is easier to build and evolve compared to the traditional approaches, as it is mainly based on a list of answers it takes advantage of unsupervised learning techniques, thus reducing setup and training time

5. Integrated Analytics

Thanks to the semantic and sentiment analysis tool, provided with EVA, it is possible to analyze and extract insights from the conversations with EVA to improve services and customer care.

6. Time & money saving

Time and costs needed to run EVA in your company are under the average of the market. Moreover our innovative approach can dramatically improve the overall process efficiency deliver a faster and better support experience for the customer