Set in early 2023, the case exposes students to the challenges of managing open source software at Google. The case focuses on the challenges for Alex Spinelli, Vice President of Product Management for Core Machine Learning. He must set priorities for Google's efforts in open source communities. An important priority is the TensorFlow framework. Released by Google in 2014, it supports many advances in modern artificial intelligence and continues supporting it. Spinelli must also manage open source activities in additional areas, such as the Android operating system for mobile handsets and cloud computing. The choices and strategies have consequences for Google's position as a technological leader among programmers who use and write code and for the safety and transparency of the code Google uses and supports. The case frames questions about what, if anything, Google should change with the explosion of interest in generative AI.

This case tells the story of Microsoft's 2018 acquisition of GitHub and the subsequent launch of GitHub Copilot, a tool that uses generative artificial intelligence to suggest snippets of code to software developers in real time. Set in late 2021, when Copilot was still in beta, the case asks how Microsoft and GitHub should roll out Copilot to the public.

It is fall 2022, and open-source AI model company Hugging Face is considering its three areas of priorities: platform development, supporting the open-source community, and pursuing cutting-edge scientific research. As it expands services for enterprise clients, which services should it prioritize? Will these projects be in line with Hugging Face's volunteer community? Further, Hugging Face decided to remove a model uploaded by a contributor, due to the potential harm the leadership felt it could propagate. Was it the right decision?

In 2021, the car manufacturer National Electric Vehicle Sweden (NEVS) faced the challenge of securing funding from its investor to launch an innovative mobility solution based on fleets of shared autonomous driving (AD) cars. The system was complex as it required the development of several interconnected components: from AD cars to a system to coordinate the vehicles with the city's digital and physical infrastructure. Furthermore, AD cars were still in a testing phase. The project was made even more complicated by NEVS' decision to target different customers: from companies which could use PONS as part of a leasing service or to transport goods, to municipalities willing to complement their public transportation services, or private citizens which could use it to commute or go shopping. NEVS felt the urge to launch the system. But was NEVS ready?

Threadless, an online apparel company and artist community which Jake Nickell founded in 2000, continued to maintain its status as a top company in the online apparel industry during its second decade. From 2010 to 2020, Threadless continued to operated its crowd-sourcing platform, while it transitioned away from traditional screen printing to a digital print-on-demand model. Concurrently, the company jettisoned its warehouse and built a worldwide network of manufacturers that could print and ship Threadless orders on demand. Threadless also launched a new platform called Artist Shops that allowed graphic artists to sell apparel in uniquely branded online stores, with the option of having Threadless manage their pricing and promotional events. The software Threadless developed to facilitate its manufacturing network and Artist Shops platform also led Threadless to increasingly view itself as a technology company performing intermediary services, rather than merely an online apparel company. The onset of the COVID-19 pandemic in 2020 accelerated the company's transition, triggering the sale of Threadless's office and a move to working from home. Nickell wondered what the next steps for the company should be.

This case discusses efforts made by biotechnology (biotech) company Amgen to introduce digital technologies into its manufacturing processes. Doing so is complicated by the fact that the process for manufacturing biologics-or therapeutics made from living cells-is subject to unforeseen variability and thus requires a highly controlled environment. Mistakes are costly, given that the manufacturing process takes several weeks from start to finish. Set in early 2020, the case asks students to evaluate two opportunities facing case protagonists Myra Coufal and Chris Garvin. The first involves working with a new team to build a standard multivariate model for a fairly new commercial product with limited production data. The second involves building a predictive machine learning model to automate one step of the manufacturing process for a top-selling product that generates sizable margins. The case includes a supplemental problem set that provides students the opportunity to analyze data and make an informed choice between the two opportunities.

After discovering that their cancer diagnostic tool, designed to leverage the cloud computing power of IBM Watson, needed greater integration into the clinical processes at the MD Anderson Cancer Center, the development team had difficult choices to make. The Oncology Expert Advisor tool used a combination of machine learning and the latest cancer care research to make recommendations to clinicians in the field. Was automated cancer diagnosis the future of cancer care? The development team, comprised of clinicians and data scientists, reviewed the results of their experiment to augment their implementation plan and better evaluate the efficacy of the analytics tool.

Korea Telecom has committed $4billion in investments and R&D to build a GiGAtopia, essentially ushering in the next generation of mobile (5G) and wired infrastructure. CEO Dr. Hwang and his team are considering which areas to prioritize in terms of new products and services in development. The top five sectors identified by KT's team include Internet of Things (including connected cars and smart city/homes), media, health, energy, and security and surveillance. Which might provide some quick wins both in terms of revenues and market lead. Should KT develop solutions that could be exported to other countries? Should KT go all in across all five sectors, or select one or two to prioritize?

Founded in 2016, DeepMap developed high definition (HD) mapping software and localization services for Level 4+ autonomous vehicles. Traditional navigational maps were accurate to a few meters, which was sufficient for drivers, but not for machine-driven vehicles which required centimeter level accuracy. Autonomous vehicles required a new form of map that was highly precise, produced a 3D representation of the surrounding area, enabled vehicles to locate themselves within the map, provided information for how to navigate safely using the correct rules of the road, and was updated continuously as road conditions changed. DeepMap was not selling a static mapping database, but rather licensing its software under a software-as-a-service model. As a startup with limited resources navigating a nascent market, DeepMap faced uncertainty across several dimensions: the timing of overall AV market adoption, which countries would adopt fastest, which AV segments would move most rapidly, which sensor technologies would become standard, and what impact regulation would have. They also faced the challenge of serving customers on a global scale. As DeepMap looked ahead, it had to decide how and where to focus and allocate its funding in order to achieve its short and long-term objectives.

In 2016, Affectiva-a Boston-based emotion AI software company with a long track record of building emotion-sensing software for market research-had attempted to expand into new verticals by releasing a mobile software development kit (SDK) that downloaders could adapt for their own use cases. The experience taught Affectiva that the company itself would have to bear most of the financial risk for adapting its technologies but also demonstrated that the automotive industry was very interested in using Affectiva's technology to monitor the emotions and cognitive states of drivers and passengers. In 2018, Affectiva executed a "90% pivot" to serve the automotive industry's increasing demand for "emotion AI". However, with automotive revenues not expected for several years and very expensive data collection requirements to enter the industry, Affectiva faced a challenging set of trade-offs between its market research business, its SDK, and its automotive aspirations.

Four years after being founded Amir Amir Konigsberg (CEO) and Adi Avidor (CTO), Twiggle had developed a search enhancement that plugged into the online merchants' existing framework. The company utilized advanced structuring and linguistic tools to build search technology that understood shopper intent and matched it with the right products. Twiggle was founded in 2014 by former Google executives Amir Konigsberg (CEO) and Adi Avidor (CTO), who believed e-tailers were losing enormous revenue due to poor search results. The team built a set of proprietary tools that created a human-like understanding of customer queries in the e-commerce search experience, using natural language processing that translated text into meaning. The team also constructed a semantic model of the e-commerce world for three product domains: fashion, home and electronics. The resulting "ontology" was a set of concepts and categories in a subject area that showed their properties and the relations between them. The Twiggle's tools could also unlock significant online sales revenue by increasing click-through and add-to-cart rates, ultimately improving sales conversions. So far, Twiggle had secured deals with more than half a dozen large e-commerce retailers and, so far, could improve search results in three product categories: fashion, home, and electronics. The company initially targeted large direct-to-consumer e-commerce players. Yet the cofounders encountered challenges pursuing this type of customer. The customer acquisition process included lengthy, intensive face-to-face sales cycles, expensive and technologically complex proof-of-concept testing, and requests for customization. Konigsberg and Avidor wondered if they should expand their customer focus to target a wider set of smaller e-tailers that still had significant volumes of online search queries, but their search quality tended to be less robust and might be easier to improve.

An Israeli startup founded in 2014, Zebra Medical Vision, develops algorithms that produced diagnoses from X-rays, mammograms, and CT-scans. The algorithms used deep learning and digitized radiology scans to create software that could assist doctors in making diagnoses. By July 2018, Zebra had developed seven algorithms to analyze scans for emphysema, liver density, compression fractures, bone density, brain bleeds, breast cancer, and a calcium score-used to detect calcified plaque in coronary arteries. For each scan analyzed, Zebra charged hospitals $1. By 2018 Zebra found itself in a race with its competitors to perfect these algorithms, create software tools, distribute the tools to physician partners, and create a market. Zebra already had several partners in the U.S. and Europe who gave feedback on its development. Management had to decide: What should they do next at Zebra? Should they work on the accuracy of the already-developed algorithms, or continue to develop many new tools? If they chose to develop new products, which applications should they address? And how should they go to market?

In two short years, Viacom's Data Science & Advanced Analytics team built a web platform called Science Central that allowed employees from Viacom's 20+ cable networks to access television audience insights through three data science apps. In the past, employees would have approached the team to carry out these requests. Vice President of Data and Audience Development Fabio Luzzi, who oversaw the 10-person team, believed that making data science instantly available in accessible formats would allow teams to make better-informed decisions. By June 2017, the platform had 600 regular users, but Luzzi wanted to expand its reach. He considered whether the team should focus on strengthening the platform or devote more resources to custom analytics requests that would allow the team to explore new problems and develop new insights.

In May 2016, Michael Rechtin, an expert in international data center law, advised global financial services firm Great Lakes Banking Group (GLBG) on its plans to upgrade its data centers. The bank's data processing and storage systems were in need of an update, and since GLBG last made heavy IT investments in the late 1990s, the technology had changed considerably. GLBG sought Rechtin's advice on whether or not it should build a new data center, pursue a wholesale colocation solution, rent space from a retail colocation provider, or store more data in the cloud.

Korea Telecom has committed $4billion in investments and R&D to build a GiGAtopia, essentially ushering in the next generation of mobile (5G) and wired infrastructure. CEO Dr. Hwang and his team are considering which areas to prioritize in terms of new products and services in development. The top five sectors identified by KT's team include Internet of Things (including connected cars and smart city/homes), media, health, energy, and security and surveillance. Which might provide some quick wins both in terms of revenues and market lead. Should KT develop solutions that could be exported to other countries? Should KT go all in across all five sectors, or select one or two to prioritize?

The net neutrality debate had implications for Internet service providers, content providers, and end users. This note aims to inform the reader of the various sides of the debate where open issues remain, and what aspects an entrepreneur, investor, or content provider-either an existing player, or a new entrant-should be aware of. From a management perspective, net neutrality rules could have major consequences on the business models of content providers, ISPs, and other businesses that depend on the transfer of data across the Internet.

In late 2013 and early 2014, Netflix service over the major U.S. Internet Service Providers (ISPs) suffered major slowdowns. What were the causes of these problems? What could Netflix do to solve them?