AI ML - ValueLabs

End-To-End Case Study

Geo location

US (South)

Engagement

ETM

Service

AI&ML

Challenges

The market research industry relies on content written by trained writers. Our client, one of the largest in the US, enjoys a hard-won reputation for innovation and accuracy. They wanted to explore how they could introduce AI/ML to a process that is traditionally reliant on human creativity and, as a result, is often time-consuming and expensive. The client wanted to introduce a machine learning and a natural language processing (NLP) solution in this part of the project, but they needed someone to provide machine learning as a service.

Slow manual content creation

Inefficient use of resource

Updating traditional processes

Quality of market research

Survey writers were expected to write 40 questions per shift
Questions could take up to 10 minutes to write, and a significant amount of allotted time was spent generating incorrect answers
The client wanted to integrate NLP services into their process
The client needed AI and ML solutions to match the quality of human writers

Our Solution

We are an experienced, full-service machine learning service provider, and we quickly saw how our solutions could speed up and simplify the process of generating questions. We built a custom program that automatically generated incorrect survey answers and integrated a video player and Google Docs into the program. Writers could view research videos, take notes, write the question and correct answer, generate and select incorrect answers, and move to the final distribution step within one product.

Solution Impact

Created significant savings

In the first quarter, with expected increases in subsequent quarters

Sped up answer generation

Made new writers more productive & less reliant on experienced writers

New opportunities for automated AI

The client wished to explore other areas of our artificial intelligence consulting services

Our Approach

Bringing AI/ML technology solutions to market research

Artificial intelligence and machine learning services is one of our core strengths. As always, we began the project by understanding the client’s needs. That way, we can provide the right advice and software for the project from the beginning. We gathered a team with vast experience and researched the most appropriate tools. We built a cloud-based app that used machine learning to automatically generate false answers to survey questions and included a video player and Google Docs. It allowed writers to write surveys more quickly without switching between apps and windows.

Defined support delivery processes and required cloud machine learning services
Used Terraform scripting to configure automation and generate unique incorrect options
Included active learning component to improve the program while being used
Integrated the solution with a video viewer and Google Docs to create a single product

Handling a 6x increase in inquiries during COVID with an intelligent chatbot

Read Blog

Created a one-stop survey generation shop through AI/ML development

Survey writers had a complex job. They watched market research videos, took notes on what they saw, wrote the questions and correct answers, thought of incorrect answers, and then uploaded all of that for publication. We built the answer-generation program and integrated their video viewer and Google Docs into it to save them from bouncing from one program to another. Now, they can generate survey questions from raw video into the finished product without wasting time opening and closing several different programs. In just a few clicks, they can create the quality market research results on which our client has built their reputation.

Added active learning component that improves with use

We wanted our AI and ML solution to constantly learn and improve on its own and hence took content generation a step further. Our solution analyzes the language the writers use to create incorrect answers and allows the writer to rate the responses it generates. Writers can create more options, which helps the AI/ML-driven program to learn what kind of answers don’t work. The writers can also edit the options, showing the app improvement opportunities. Finally, writers can select options, which informs the program that the choice was right. Our solution then uses that information to improve options in the future. That way, the program is constantly learning and improving, creating even better choices and more productivity.

Business Impact

We brought the full power of our AI and ML services to bear on this part of the project, and the results speak for themselves. Our natural language processing (NLP) solution has allowed our client to write survey questions seamlessly. It reduces newer writers’ dependency on experienced writers and increases their productivity. As productivity increased, so did the cost savings.