Moveworks Intelligence Engine™

A machine learning model is only as good as the data used to train it. By analyzing employee requests from over 250M support tickets and pioneering a technique to collectively learn how employees ask for help at work, Language Core™ deeply and accurately understands all kinds of support language across different companies, industries, and domains.

When an employee submits a request, this NLU-powered system breaks down the employee’s question into structured pieces by tagging each word, identifying parts of speech, spotting synonyms, predicting the relative importance of each word in context, and more.

Once the language is standardized and analyzed, Language Core can correctly determine the type of issue and the employee’s intent with high accuracy. And since this system's models learn from user feedback, they continuously improve with every employee interaction.

Most chatbots follow pre-programmed paths, but conversations are dynamic.

Dynamic Flow is the framework for deciding the next steps needed to solve an employee's problem. Able to switch topics and generate dialog on the fly, Dynamic Flow ensures a more natural chatbot experience. 

At every step of the conversation, Dynamic Flow generates clarifying questions to further understand the request, handle ambiguity by presenting multiple solutions to move forward, add comments to a ticket, pull information from your knowledge base, and more.

Most chatbots can’t topic-switch. If an existing conversation is interrupted halfway through, the flow will break and your employee will have to restart their conversation from the very beginning, wasting their time with a poor experience.

With Dynamic Flow, if an employee changes their mind mid-conversation, they do not get stuck in a loop. They can jump out of the recommended flow at any time, and the bot will probabilistically move the conversation in a different direction — all while remembering and coming back to the previous flow, as needed.

Given that conversations are unpredictable, at each “turn” or change of topic, there needs to be a way to make decisions on the fly. The best way to make a great decision is to incorporate all of the possible ways to move forward in context. 

Unfortunately, with businesses constantly changing, traditional conversational platforms don’t have the flexibility to automatically update decisions if the conversations are pre-scripted. In addition to this rigidity, pre-programmed bots also typically require a whole lot of bandwidth to maintain for support teams.

Action Bid, an intelligent bidding system calculates the probability of all of the possible paths to resolving an employee’s issue: a relevant form, a troubleshooting article, an employee handbook. The path with the highest likelihood to solve the issue — the highest bid — is surfaced to the user, resolving their request.

While some bidding systems fall short, failing to incorporate additional context from past conversations or updated knowledge bases without manual effort from support teams, Action Bid dynamically generates the most relevant bids based on the conversational context.

Most conversational systems will point users to a pre-programmed, hardcoded solution, sending a knowledge article or a link for the user to navigate away to resolve the request. Because these systems map solutions directly to requests, employees have to search through an article or find the right form to get what they need.

A truly conversational chatbot does both — first, understanding both the request and the possible solutions, and second, serving up only the most relevant solution directly in chat.

Semantic Match goes a step further from traditional enterprise search systems, taking every piece of disparate internal information into account, from the words and phrases employees use to the resources available to the person asking to the location of the request.

With Semantic Match, a Moveworks chatbot can surface a personalized and highly relevant solution to the employee asking — without them ever having to leave the chat platform.

When a subject-matter expert writes an article, they are trying to document everything. But this kind of detail can be difficult for employees to find, read, and understand.

Knowledge articles can be thirty pages of potential WiFi issues, including everything from VPN complications to troubleshooting routers. And when it comes to HR — healthcare benefits articles can be hundreds of pages of text, featuring information from a dozen different countries. 

The key is figuring a way to get the right information in the right hands. And we’ve done this with Enterprise Cache — a system that indexes available resources, so every question is answered with a single, precise snippet of information. Even when there are updates to various knowledge systems, Enterprise Cache automatically incorporates new information.

And if the answer doesn’t live in your knowledge base? Enterprise Cache also ingests knowledge from trusted, public websites to make your answer store the best it can be.