Pez.AI: The Grandmaster of AI-Powered Chatbots

Brian Lee Yung Rowe, Founder & CEO

In the fall of 2015 when Brian Lee Yung Rowe first noticed Slack business messenger app’s service integrations, it set in motion a chain of events that would one day transform the workflows for enterprises. The ability to consolidate data from third-party SaaS applications into Slack turned out to be a eureka moment for Rowe. With strong expertise in quantitative analysis and machine learning, Rowe had a flair for leveraging data to build state-of-the-art models for the financial sector. Slack triggered an avalanche of ideas in Rowe’s mind; he envisioned a data source powered by artificial intelligence (AI) with built-in analytics capable of providing all the information sought by analysts as well as customers through chatbots. Further, he intended to provide a more “democratic” interface that gave data insights without programming. The result was a chatbot platform built on top of Slack that won immense appreciation from companies that went on to become Rowe’s clients. Thus began the journey of Pez.AI—a trailblazer firm that has championed the use of chatbots as a vehicle for data-driven continuous improvement.

Chatbot-Driven Workflows

While Pez.AI has effected a paradigm shift in utilizing chatbots to create data-driven workflows and to optimize business processes, the company has a broader focus empowering organizations via a seamless conversational experience. “Everyone recognizes the importance of data, but transforming raw data into something useful is a challenge, irrespective of an organization’s size, as it requires different business workflows. Chatbots are more than just a new way of interacting with customers; they present an opportunity for organizations to be data-driven,” remarks Rowe, founder and CEO of Pez.AI.

The company offers a SaaS platform that supports chatbot interaction through different modes including its own web channel, Slack, voice, or even home speakers like Google Assistant and Amazon Alexa. The Pez. AI platform has three components, each of which functions as a part of a “continuous improvement cycle” of the clients’ workflow.

The first component is chatbot authoring, where customers select a turnkey bot role, like HR assistant or sales assistant. These roles come bundled with dialog packs and analytics that address specific industry pain points. Rowe cites the example of e-commerce, where companies can expect up to 30 percent of their orders to be returned. “Customers return products when their expectations of the product differ from the actual product, or the wrong size or color is shipped, or the product is broken during delivery,” comments Rowe.

He further states that chatbots provide a novel way of assisting customers with product returns. When users seek to return an item, the chatbot politely asks why they are returning an item. These reasons are collected and tracked in the analytics component.

Conversation design is difficult. Our dialog packs balance best-of-class conversation experience with personalization to create high-quality bots in just a few minutes

The second component of the continuous improvement cycle is managing customer responses. Customer service personnel are notified when the chatbot does not understand a question and can take over the conversation. Once resolved, the control can be transferred back to the chatbots. The platform groups escalated conversations and recommends which interactions should be added to the bot. Managers can use the responses of the customer service representative or use their own to train the bot. Over time, the bot coverage increases, providing further scale to customer service.

The coup de grace of the platform is the purpose-built analytics. At the heart of Pez.AI is a sophisticated natural language processing (NLP) engine that not only powers the bot but also the analytics. By analyzing the intent and sentiment of each user request, Pez.AI can show businesses what their customers want and highlight issues in their business processes. Returning to the e-commerce example, Pez.AI can rank the top reasons for product returns and also display the historical trend. Sentiment analysis helps to identify which reasons upset customers the most, helping decision-makers prioritize the issues. Managers can identify if the quality of control issues is acute or systemic and act accordingly. The manager can use the analytics to see whether the issue was materially improved.

Accelerated Deployments

Being a SaaS solution, Pez.AI’s chatbot platform requires no specific onsite deployment. Rowe mentions that for enterprise clients, in most cases, a fully functional bot can be published in just a few minutes. Pez.AI can customize the dialog packs by training the bot on existing customer transcripts. Doing so also provides immediate insight into some of the reasons why customers contact a business.

For the HR space, organizations can leverage the Pez.AI platform to create essentials like an employee handbook. Onboarding new employees requires hiring managers to give up a lot of time and attention. With Pez.AI, hiring managers can quickly configure a virtual HR assistant using chat. The beauty of the Pez.AI approach is that while the manager configures the bot via chat, the bot is also guiding the manager on the onboarding process. So the manager benefits from the chat interaction. Once the bot is configured, new employees can read and acknowledge company policies via the bot, saving hiring managers valuable time. This process is what Rowe calls seamless conversation, where all actors using the platform benefit from chat.

The Chronicle of Success

Within two years of its inception, Pez.AI’s solutions for chatbot-enabled workflows have won the attention of organizations in multiple domains ranging from finance, HR and benefits administration, travel agencies, and restaurants to e-commerce. In one of the engagements, a peer-to-peer lending company used Pez.AI to create a chatbot to answer borrower questions. Using the analytics component, Pez.AI identified that the majority of escalated interactions came from lenders, not borrowers. Training the bot with these additional interactions thus increases the bot coverage and lowers the load on customer service agents.

Exciting Future Ahead

Buoyed by the response from clients, Pez.AI is now engaged in creating turnkey chatbot assistants in different verticals, such as HR and e-commerce to streamline interactions and save time. Rowe sees these verticals as the best way to introduce organizations to the power of data-driven workflows. Over time, additional workflows will be added based on customer demand.

The company also transformed the employee onboarding process with some of its clients leveraging its chatbot solution to keep their employees apprised of the new policies, procedures, or locations of resources. The Pez.AI Chief of Staff dialog pack allows users to raise questions on policies and procedures to provide the most appropriate answers. For example, employees can bypass the process of logging in their sick leaves by merely telling the bot, which not only logs the leave, but also notifies the team manager. “If someone is falling sick more often than normal, it may indicate an acute health issue or that they are dissatisfied and interviewing for a new job. The manager can take appropriate action based on the insights provided by the bot,” says. Rowe.

Rowe mentions that the company is committed to the HR space and has partnered with a firm operating in the HR benefits administration space that owns a portal used by numerous Fortune 500 companies to provide benefit information to employees. Employees can ask the bot questions about the number of vacation days they have, whether they qualify for health insurance, and other benefits they have access to. Pez.AI provides the AI infrastructure that powers the chatbots.

Given the difficulties in identifying employees’ job satisfaction, Rowe cites a research study confirming people’s comfort in sharing personal opinions with bots than humans. Pez.AI surveys employees periodically on their state of being using the Employee Satisfaction dialog pack, wherein a bot asks easy-to-answer questions to employees as opposed to companies conducting long point-in-time surveys. The managers eventually get an understanding of their employees’ state of being, following which they can have a targeted discussion with each individual to resolve the issues affecting their job satisfaction and performance. Being at the forefront of innovation, Pez.AI is championing the chatbot-enabled workflows for niche domains.

- June Morgan
July 16, 2018