Viable is attempting to quantify qualitative customer feedback thanks to AI

Viable is attempting to quantify qualitative customer feedback thanks to AI ...

Most analytics solutions have an implicit assumption: the results analyzed and the findings derived, are almost exclusively quantitative. They include number of customers, sales, and other information.

When it comes to customer feedback, perhaps the most important data is qualitative: text contained in sources such as feedback forms and surveys, tickets, chat, and email messages. The problem with that data is that they require domain experts and a lot of time to read through and classify. Or, at least, this was up until now.

Viable is contemplating addressing this problem. Today, the company, which is renowned as the finest qualitative AI company to offer natural language querying of customer feedback, announced the completion of a $5 million fundraise primarily for growth, research, and new hiring.

From product-market reach to customer feedback via NLP, you''ll get product-market experience.

Erickson, an engineer by trade, cofounded Viable with his identical twin brother Jeff, who is a designer. Both have been in the IT industry for about 15 years, having skipped college to start growing early on.

As Erickson said, the two have held senior roles at several startups, and their career paths have intertwined over the years.

The Erickson brothers decided to start their own business, focusing on solving Dan''s product-market fit challenge over the years. That was the start of what was initially called Viable Fit.

The company developed a product to assist individuals navigate the otherwise known superhuman product-market fit process. The process is centered around a survey, followed by a discussion to help owners and their customers navigate the markets roadmap.

The Viable team developed some of their natural language processing (NLP) techniques to make it work at an all-time high. It quickly discovered that this turned out to be the most significant component of their entire approach.

Gaining traction: Push and pull

The Erickson brothers decided to focus on their NLP models and find something new among small businesses.

Viable began focusing on aggregating customer feedback across platforms. Viables'' platform also offers a full analysis service that enables feedback to be obtained from written analysis. This technique may be applied to products such as customer experience, or marketing.

The analysis Viable offers can be accessed in two ways: push and pull. For the push mode, a report is sent on a weekly basis that covers what happened in your customer feedback during the last week. The report includes things such as the top complaints, compliments, questions, and requests from customers. The scope of the reports ranges from a dozen to a few hundred paragraphs.

When people read those reports, they have questions they need answered to be successful. Viable assist users in making a difference by providing a natural language question and answer system. Users may type in a question about the data and Viable provides a reply, all in plain English.

A separate subscription level for the service is also possible for Zendesk, Intercom, Delighted, iOS App Store, Play Store, and Front. As well as the ability to ingest data via.csv files, the company also offers an out of the box integration.

Under the hood

It may sound simple and obvious, so it''s crucial to wonder how no one else would have done it before. In theory, anyone including the Zendesks of the world itself might have done it. The answer is twofold.

Viable has a head start since they started in 2020 just before GPT-3 came out. As Erickson shared, they were among the first to integrate some of GPT-3''s capabilities in a commercial setting. Second, the fact that it integrates data from many different sources makes it a lot easier.

Viable is a little more than a thin wrapper around GPT-3. It employs the OpenAI API, including embeddings, as well as the actual GPT-3 completion engine. However, Viable also has its own models that work with GPT-3, which have been trained and fine-tuned throughout the last two years.

It also has its own data repository as well as its own ingestion pipeline. Whenever a new piece of content is created, it''s pulled in, along with any metadata that might be available. From there, it goes into a pipeline consisting of different versions that Viable has developed, along with some GPT-3 capabilities that will classify the text.

The classification process identifies whether the text is a complaint, a compliment, a request, or a question. It also identifies different topics within the text and performs some sentiment analysis, emotion analysis, urgency analysis, and noise detection.

At this point, the platform is geared towards text analysis but it cant directly connect to sources such as databases or spreadsheets. However, it can use what Erickson described as customer traits to slice and dice the data.

Those may include job titles, location, or even numerical answers to multiple choice questions, such as how many times a week do you use the product. Alternatively, users may have the system perform tasks like write a report for my product manager enterprise customers in the Bay Area who use the product one to two times per week.

Viable has developed a completely unsupervised approach for thematic analysis based on GPT-3 embeddings along with a proprietary thematic analysis engine on the top, according to Erickson. That means the system does not have to be given any context for what kinds of things its looking for other than requests, questions, compliments, and complaints, so it can be effective in any domain.

Limits for avoiding bias and abusive language

GPT-3 may be one of the most impressive feats in engineering and AI, but its not without its shortcomings. One of the most powerful aspects, such as toxic language generation and hallucination, is generating authoritative-looking statements that aren''t just based on facts. Viable has managed to circumvent those by custom training.

Is there any need to define a theme? Weve built out a fully fine-tuned version of GPT-3 that keeps it on the rails? So, it''s going to do all of those curse words or anything like that. So, on the hallucination side, we have done a meticulous job of putting out the training data set so that every single example that we pipe in is only directly using facts from the feedback that is inserted into it. I do not want you to be creative here.

Beyond GPT-3 and customer feedback

The above should provide valuable free advice to anyone looking to build a business around something like GPT-3. Not only in terms of how to overcome its shortcomings, but also in terms of how to add value on the top. The cost of using GPT-3 is covered in Viables pricing points, as well as other processing costs, and a healthy margin.

Viables investors deserve it. Streamlined Ventures led the $5 million round owing to its interest in applied AI, with participation from previous investors Craft Ventures and Javelin Venture Partners. The round included investment from Merus Capital, GTMFund, Stratminds, Tempo Ventures, Micheal Liou, Bill Butler, and Samvit Ramadurgam. The total amount of funds to date in Viables has been $9 million.

At this point, the company has a total of nine employees and a dozen paying customers. According to Erickson, the company has a few high-profile clients who are pleased with the product and Viable has made the move to expand beyond customer feedback.

According to Erickson, we strive for any type of experience, whether it be employee experience, partner experience, customer experience, or for assisting people analyze their qualitative nature.

You may also like: