Google Workspace Results
Immediate responses save hundreds of employee-hours per month
• Enables rapid deployment of an enterprise chat assistant with Google
Workspace and Google Cloud
• Responds in under one second to over 95% of all inquiries
• Accelerates Red Hatters easy access to critical information—saving time, reducing costs, and improving productivity
At Red Hat, open is more than just a word. It is the cornerstone of the organization’s technology, culture, and business operations. The company’s initial product, Red Hat Linux, was an open source operating system. Today, Red Hat contributes to more than 450 open source projects throughout the software stack—from the operating system and developer toolchain to apps in the cloud and automation.
Open source depends on the free exchange of opinions, ideas, and expertise. Red Hat has embedded these values into its culture. “Open is what distinguishes Red Hat,” says Tarique Aman Aziz, Software Engineering Manager in Red Hat’s Innovation Office. “Not only is our technology open source, but our people are open. They collaborate and share knowledge.”
Red Hatters (company employees) use Google Workspace to connect, create, and collaborate. But Red Hat’s commitment to cloud-based, open source business operations goes much further. It uses Google Workspace apps and Google Cloud products to build and scale Shadowbot—Red Hat’s innovative, enterprise chat assistant.
The genesis of Shadowbot
“Working collaboratively, we were able to quickly develop a Chatbot using Google Apps Script as a proof of concept for a more secure, scalable solution.” -- Jeffrey Daube, Innovation Architect and Program Manager, Innovation Office, Red Hat
When Jeffrey Daube joined Red Hat in 2015, the company was already using Gmail and Google Calendar. His role, as principal business systems analyst in the IT productivity group, was to expand utilization beyond Google Workspace and build custom internal cloud solutions for Red Hatters using tools and services available in Google Cloud.
“Google Chat is the single, IT-supported, chat platform for all Red Hatters,” says Daube. “We use it quite heavily.” He was using Chat and experimenting with bots when he was pulled into a project with Red Hat’s EMEA (Europe, Middle East, and Africa) sales group.
The group had developed an ad hoc messaging app to help salespeople find the phone numbers of technical specialists who could answer urgent questions. “The app was not scalable,” says Daube. “Working collaboratively, we were able to quickly develop a Chatbot using Google Apps Script as a proof of concept for a more secure, scalable solution.”
The project had a simple goal: get the right information to the right people in as few steps as possible. It started with solving 90 of the top sales queries, which could be answered by materials accessed through a smartphone or other mobile device for real-time use. It also formed the genesis of Shadowbot.
Shadowbot and the Innovation Office
The name Shadowbot is an homage to Shadowman, in the old Red Hat logo. Initially conceived and developed as a virtual assistant for sales, Shadowbot has progressed to become an enterprise-wide assistant. When Aziz joined the team, he led the Innovation Office development team to rebuild the Apps Script prototype using the Red Hat build of Quarkus (a full-stack, Kubernetes-native Java framework) running on Google Cloud Run for scalability and modularity.
With Aziz’s contribution, Shadowbot is now focused on user personas that extend its value beyond sales teams. Modules have been added for the global delivery and products and technologies teams as well. One new module, SKU Lookup, is a way for sales to access detailed product information directly from Shadowbot. Previously, a ticket would be opened and handled by a Red Hat associate. Since introduction, over 1,300 SKU queries have been handled directly with the application—saving time and reducing cost.
A marketing module is planned for introduction next, with a long-term goal of enterprise-wide utilization by over 95% of Red Hatters. According to Daube, “The vision for the future is that any Red Hatter can ask a question and Shadowbot will answer it in a meaningful way or learn from the question to inform future answers.”
An open source powered Google Cloud Platform solution
“One of the most unique aspects of Shadowbot is the seamless interplay between Google Workspace and Google Cloud Platform.” -- Jeffrey Daube, Innovation Architect and Program Manager, Innovation Office, Red Hat
According to Daube, “One of the most unique aspects of Shadowbot is the seamless interplay between Google Workspace and Google Cloud Platform.” Shadowbot is accessed through a modern message UI in Chat on Google Workspace. When a user enters a query, Google Chat API forwards it to the Red Hat build of Quarkus for serverless containers running on Cloud Run.
Quarkus on Cloud Run enables a response time of under one second for over 95% of all Shadowbot interactions. In addition, Cloud Run allows for rapid horizontal scaling from two containers, which are always live, to as many as nine during major Shadowbot announcements. This ensures optimum performance regardless of the interaction frequency.
Quarkus reaches out to Postgres, an open source database where Red Hat’s persona attributes are stored, via Google Cloud SQL. If needed, the message can be sent to the Google Translate API, which can translate over 100 languages in a few seconds. The message is then sent to Google Dialogflow, for natural language understanding of the intent and parameters of the user query. Once the intent of the query is determined, a reply is provided in the appropriate language.
If someone in sales asks, “What is ZVO?” Shadowbot will reply that “ZVO” stands for a zero-value opportunity and provide links to information and videos explaining the term in more detail. Similarly, someone could enter, “I feel lonely.” Daube says, “This may sound weird, but someone working remotely from home during a COVID office shutdown might feel lonely.” Shadowbot will use its semantic capabilities and reply with links to Red Hat’s well-being initiatives.
In addition to the question-answer workflow, details of every query and reply are logged into Google BigQuery. This is important because the information guides learning and curation.
Curating open access
A dedicated team of Red Hatters curate Shadowbot content using data captured in BigQuery. Emphasis is placed on identifying when and why Shadowbot did not provide a meaningful answer and closing the gaps.
The initial sales-focused genesis of Shadowbot included direct replies to about 90 of the top sales queries. The number of curated replies in the latest version of Shadowbot is increasing at an accelerated rate over a wider range of topics. In just 8 months, the number of replies nearly doubled to well over 1,000, exponentially improving efficiency and performance.
Another form of curation involves linking access to existing information within Red Hat. Daube says, “We have a lot of excellent information on Google Drive, on our intranet, and in our digital asset management, including marketing materials and presentations.”
Curating open access to existing information rather than creating and storing new information increases content efficiency. “We're not changing the source, we're just helping a Red Hatter find the right information using natural language understanding,” says Daube. “You don't have to know where the information is hosted, you just ask Shadowbot, and it points you to the right content.”
For Red Hatters by Red Hatters
“Connecting BigQuery to Google Sheets enabled us to quickly and easily build a dashboard where we can track the increasing usage of Shadowbot in real time.” -- Tarique Aman Aziz, Software Engineering Manager, Innovation Office, Red Hat
“Connecting BigQuery to Google Sheets enabled us to quickly and easily build a dashboard where we can track the increasing usage of Shadowbot in real time,” says Aziz. It shows that daily interactions more than doubled between January and July of 2022. During this period, a total of 2,900 unique Red Hatters had over 33,000 interactions with Shadowbot. Initial inquiries indicate Shadowbot has decreased the time required for Red Hatters to find the right information by between 2 and 12 minutes per interaction.
Beyond the numbers, Aziz sees the enthusiasm of Red Hatters from different departments working together and building Shadowbot for the benefit of everyone. He is confident it will achieve its goals because, “It’s for Red Hatters by Red Hatters.”
His confidence is echoed by users who praise Shadowbot’s quick and easy usability, extensibility, and potential to significantly improve productivity. “I believe Shadowbot could be a game-changer,” states a senior director of global product marketing. “The ability to easily search and access customer-facing content, sales tools, and specific enablement anytime from anywhere... I would have killed for that when I was in the field.”
*Google Workspace was formerly known as G Suite prior to Oct. 6, 2020.