Drishti Technologies Inc. Managing Operations PDF 2024

case W19C80 January 5, 2024 M.S. Krishnan Drishti Technologies Inc.: Managing Operations through Computer Vis...

case W19C80 January 5, 2024 M.S. Krishnan Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics In the early 2020s, digital technologies were transforming businesses across industries. Business process automation and embedded intelligence through cognitive computing, as well as decisions based on insights from real-time data analytics, emerged as major advancements. For example, in retail, online shoppers saw personalized suggestions resulting from analysis of millions of customer purchases and real- time engagements. In the fnancial sector, billions of transactions were checked by artifcial intelligence (AI) algorithms to identify potentially fraudulent activity. In healthcare, the entire sequenced human genome was used to predict effective therapies and treatments. Lastly, in marketing, advertisements were personalized to drive consumers’ purchasing decisions. In general, big data and AI were leveraged to build predictive algorithms to improve the effciency and effectiveness of the business processes. Traditionally, a prevailing belief had been that business processes are most effcient when machines replace humans. This approach aimed to eliminate the variability that human involvement could introduce to the processes. Consequently, signifcant investments were made in process automation, including the deployment of robots to perform intricate tasks. However, both the traditional approach to shopfoor process improvement and the introduction of robots had their limitations. With innovations, Drishti Technologies Inc. inserted itself into the dilemma, positing that in some hard-to-automate areas the best solutions and performance could be achieved by marrying the skills of human and machine. Manufacturing Process Monitoring Limitations of Traditional Operating Line Process Evaluation In a typical factory or assembly line, hundreds of employees worked at different shop-foor stations to build a product. Often, production rates, quality, and safety-related data were depicted on hand-generated charts and dashboards. Key business metrics in many of these production settings included process effciency, product quality, worker safety, productivity, rework, and overall throughput. Manufacturing Published by WDI Publishing, a division of the William Davidson Institute (WDI) at the University of Michigan. ©2024 M.S. Krishnan. This case was developed by Don Borschel, Gilbert Pasquale, Yuning Ye, and Maria VanDieren, under the supervision of M.S. Krishnan, Accenture Professor of Computer Information Systems at the University of Michigan’s Ross School of Business. The case was prepared as the basis for class discussion rather than to illustrate either effective or ineffective handling of a situation. The case should not be considered criticism or endorsement and should not be used as a source of primary data. A representative from Drishti reviewed and approved the case before publication. This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 organizations used several methods and tools to improve these metrics, including time and motion studies, control charts, visual management systems, and continuous improvement programs. Businesses also spent considerable amounts of money on rigorous employee training to ensure meticulous adherence to the prescribed processes and to minimize production variance. These methods contributed signifcantly to the improvement of manufacturing processes. In fact, the U.S. Bureau of Labor Statistics reported that manufacturing productivity accounted for 20% of the nation’s growth between 1990 and 2017.1 Even so, valuable opportunities still existed to enhance the effciency of these processes. However, it was almost impossible to monitor all the processes—products, employees, and stations —in real-time using traditional quality improvement methods. While traditional sampling was informative when executed correctly, it could be easily misunderstood and performed incorrectly, especially in larger operations. Also, sampling studies were not always capable of specifcity to derive insights through analytics. For example, quality engineers might not be able to target a specifc product, such as a smartphone with serial number 35876212081768, in the event of problems or defects arising. Sampling techniques also could not identify in real time if and when issues were being introduced by, for example, a new process or employee. Therefore, the capability to observe and measure every movement, station, product and unit in the manufacturing process would be extremely valuable to companies seeking to maximize effciency. Limitations of Robots The capacity of robots’ production lagged the manufacturing demand. In 2018, more than 72% of factory tasks were performed by humans.2 Thus, the challenge of managing human errors in manufacturing was still present and signifcant. Traditionally, manufacturers addressed these challenges through continuous improvements in process design and training workers to follow standard processes (“standardized work”). Adoption of digital automation for shop foor operations was often questioned as over-emphasizing simplicity and process visibility. In the early 2000s, it was extremely expensive to equip a facility with internet of things (IoT) devices necessary to collect, store, and process the necessary data to build valuable analytics models and support these devices once installed. Consequently, manufacturing industries were relatively slow in adopting technology solutions. Computer Vision Comes into Play Computer or machine vision was a cutting-edge technology that had the potential to bring new levels of transparency to manufacturing. It applied computer processing capabilities to capture, interpret, and respond to visual stimuli with the same functionality as the human visual system. As IoT devices and sensors began to become dramatically cheaper, processing power, whether through on-premise servers or the cloud, was much more accessible. Since 2012 and the introduction of ImageNet,3 a large, hand-annotated visual database of images, the use of data-driven programming based on AI technologies surged.4 Traditional manufacturing organizations became more data literate, giving rise to new products and advancements such as automotive companies competing to build more autonomous and technologically connected innovations. Product development had been driven by data and automation, and now the manufacturing line could also be streamlined and altered with insights from big data. Recognizing that data could drive decision-making in the manufacturing industry, Drishti, a computer vision startup, was launched to revolutionize the collection and analysis of data from companies’ shop foors. In so doing, it could fundamentally rewrite how productivity and quality were quantifed and improved on the plant foor, a change perhaps as consequential as when Henry Ford, Frederick Taylor, and Frank and Lilian Gilbreth introduced their seminal time and motion studies in the early 1900s. 2 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Drishti Technologies: Putting the “Eye” in Innovation In 2016, Dr. Prasad Akella founded Drishti in Palo Alto, California, to provide insight into production foor processes where manual assembly was the dominant production method and where a sense of process information was largely absent. Drishti is the Sanskrit word for “vision”—aptly named because computer vision was central to the company’s goal of empowering the billions of people in the workplace, from warehouse factory foors to restaurants. (See Exhibit 1 for a timeline of company development.) Fundamentally, Drishti was uniquely creating a digital data set of manual manufacturing activities through multiple digital cameras placed in areas of operation. Drishti’s technology answered three valuable questions for its customers: 1. What just happened? Drishti captured and annotated videos of the manufacturing foor, allowing supervisors to search for specifc time stamps or products. 2. What is happening now? Drishti live-streamed current operations, including annotations of the movements being performed. 3. How can we improve what is happening next? Drishti applied artifcial intelligence and machine learning capabilities to help operators enhance effciency and productivity. Similar to how Google and Apple mined text and audio to radically transform business processes, Drishti’s “action recognition” technology could identify specifc actions occurring in a video stream, thus revolutionizing analysis of these processes. By combining the strengths of AI and human fexibility, Drishti’s technology could enhance human potential in an increasingly automated world. This translated to measurable ROI opportunities for its manufacturing customers, such as improvements in productivity, quality, and training. The Drishti method began by installing cameras on the factory foor to capture actions occurring at each station. The video was then uploaded, processed, and automatically annotated, enabling a manager to search the video using a timestamp, product number, station, or other identifers determined by the customer. This video could also be used to train employees and improve models that analyzed cycle time, machine performance, and product quality. Unlike traditional manufacturing process improvement methods like operational equipment effectiveness (OEE), which relied on small samples of visual or measured data, Drishti’s technology provided a comprehensive view of the factory foor in real time. Using AI capabilities in replacing traditional time and motion studies, Drishti provided previously unattainable insights to enhance overall performance on factory foors. Unique Founders’ Experiences Led to the Drishti Venture In addition to Akella, two other experts in the feld contributed to Drishti’s inception: Dr. Krishnendu Chaudhury and Dr. Ashish Gupta. Dr. Akella, the lead founder, CEO, and chairman of Drishti, had long been a pioneer in creating market categories that used technology to extend human capabilities. For example, in the 1990s he was a leader in a team drawn from General Motors, Northwestern University, and University of California Berkeley that created category-defning intelligent assist devices, later referred to as collaborative robots (“cobots”). The global cobot market was valued at 1.2 billion USD in 2023 and projected to rise to $6.8 billion by 2029.5,6 Akella also served in executive roles for various enterprises including Spoke, SAP, Thomson Reuters, and Intermedia; received numerous awards including the IEEE’s Anton Philips Award and the IEEE/IFR Entrepreneur Award; and was designated a Technology Pioneer by the World Economic Forum. He was awarded eight U.S. patents and published in many technical journals. Akella holds a PhD in Robotics from 3 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Stanford University, an MBA from the University of Michigan’s Stephen M. Ross School of Business, and a BS from the Indian Institute of Technology. He was elected a Fellow of the ASME and the SME. Chaudhury, Drishti co-founder and CTO, was a computer vision and deep learning expert. Over his 20+ year career, he was part of many technological innovations at Google, Flipkart, and Adobe. At Drishti, Chaudhury drove the creation of its action recognition technology that digitized human actions in real time—a challenge that pushed the boundaries of computer vision and deep learning in the spatio-temporal domain. He was the lead author of the book Math and Architectures of Deep Learning.7 Chaudhury earned a PhD in Computer Science from the University of Kentucky and a BS from Jadavpur University. Gupta, Drishti co-founder and board member, sought unexplored problem spaces in which to grow large companies. His career spanned inventions and innovations across the technical and business worlds. Working with colleagues, he helped invent virtual databases to aggregate structured and unstructured data at Junglee (acquired by Amazon), business process outsourcing at Daksh (acquired by IBM), online travel in India (MakeMyTrip/MMYT), and an Indian venture fund (Helion). Gupta earned a PhD in Computer Science from Stanford University and a BS from the Indian Institute of Technology, where he was awarded the President of India’s Gold Medal. Since its launch, the Drishti team grew signifcantly, operating in Silicon Valley, India, Europe, and Japan while hiring more than 225 employees. Drishti won recognition as a World Economic Forum Technology Pioneer, a Forbes AI 50 company, an NVIDIA Top 5 AI company, and a National Science Foundation SBIR awardee. The company raised a $25-million Series B round in June 2020, a $10-million Series A round in May 2018, and a $2.5-million seed round in June 2017. Its investors included Andreessen Horowitz, Alpha Intelligence Capital, Benhamou Global Ventures, Emergence Capital, HELLA Ventures, Kauffman Fellows, Micron Ventures, Presidio Ventures, Sozo Ventures, and Toyota AI Ventures. Drishti Goals and Products Since its 2016 founding, Drishti sought to augment humans in the manufacturing industry, assuming humans would continue to be critical in factories. To this point, in April 2018 Tesla CEO Elon Musk tweeted,8 “Excessive automation at Tesla was a mistake.... Humans are underrated.” (See Figure 1). In 2018, there were approximately 345 million factory workers with 1.7 million robots deployed alongside them. Humans still completed 72% of factory tasks and created 71% of value (yet introduced 68% of defects) in manufacturing.9 4 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Figure 1 Elon Musk’s Tweet on Manufacturing Source: @elonmusk. “Yes, excessive automation at Tesla was a mistake. To be precise, my mistake. Humans are underrated.” Twitter, 13 Apr. 2018, 12:54 p.m. https://twitter.com/ elonmusk/status/984882630947753984. Historically, to improve productivity, manufacturers used time and motion studies for operations management. These studies were signifcant to scientifc management in manufacturing. This feld was pioneered by Frederick Taylor and Frank and Lilian Gilbreth in the late 1800s and early 1900s. The traditional way to conduct time and motion studies involved industrial engineers physically being on the manufacturing foor, recording times spent on each step, and conducting subsequent analysis of the data to identify bottlenecks and opportunities for improvement. However, with study insights based on generalizations deduced from the data samples, many challenges arose. For example, employees could behave differently when they knew they were being observed. Also, it was impossible to observe all the employees’ actions directly and collect data in this physical setting. With these limitations in mind, the team at Drishti leveraged opportunities in emerging technologies such as computer vision, AI, and machine learning to provide an intelligent approach for management within the manufacturing industry. Drishti’s products centered around two core capabilities. The frst was action recognition technology,10 which automatically digitized the tasks that manufacturing line workers performed by analyzing the video streams from cameras directly above the workstations. The second was the delivery of a fully integrated and vertically focused product.11 In the same way that a lab technician can operate a sophisticated MRI machine without understanding how all the science of the machine works, manufacturing line associates, production staff, and engineers could use Drishti without needing to understand the underlying technology or requiring support from data scientists, who were often a scarce and expensive commodity. Drishti’s software products were verticalized for manufacturers which meant that the products’ functionality was built using the language of the manufacturing vertical, delivered KPIs tracked by the manufacturing vertical, and, generally, designed for professionals to use in a manufacturing environment. In the manufacturing vertical, products were designed for plant personnel (from the line associate to the plant manager) to beneft from using the detailed analytics that Drishti’s video-based AI analytics delivered on each single unit produced. Drishti worked with its customers through a three-step process: 5 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 1. Video capture: Cameras were installed throughout the factory to stream processes at every workstation. (Drishti said it did not capture personally identifable information; the captured video masked identifable features such as a human face.) 2. Data creation: The video was searchable and data was organized and stored. Process metadata was automatically generated, including time and motion, annotations, and descriptive tagging for collaboration. 3. Integration: This involved enabling quality and industrial engineers, supervisors, and line workers to use the Drishti-generated data and the software tools that they already used, like MES, QMS, and ERP systems, to drive improvements in productivity and quality and accelerate training across entire production lines. In sum, as Akella outlined in October 2022, in all its offerings Drishti relied on proprietary, AI-powered video analytics to deliver actionable insights for its customers in near real time. While monitoring the videos, the software identifed attributes, events, or patterns of specifc behavior, generated automatic alerts, and facilitated forensic analysis of historical data to identify trends, patterns, and incidents to enable smarter decisions. These capabilities were embodied in the following Drishti products developed over time: Trace allowed customers to use live and recorded video. It could be thought of as a combination of YouTube video cataloging plus Zoom, providing live video and remote access. It enabled addressing the questions of what just happened and what is happening now. Flow augmented human input through cycle identifcation and tracking. With Flow, customers used data and insights powered by AI to continually measure and improve their production. For example, a user could analyze cycle times to identify variation and understand outliers for optimizing a stable manufacturing process. In a manufacturing operation, a specifc station on the shop foor mandates a set of physical movements of people and parts. If there were deviations from these movements, Flow enabled managers to spot these variations and study the implications on product quality and process performance. Assist, an advanced AI product, augmented human input through step recognition. With Assist, customers could measure, understand, and improve in-cycle activities. For example, the team leader could understand whether Standardized Worki had been followed for every unit built on the production line. Just as a spellchecker helps writers focus on content while ensuring the document is grammatically correct, line associates could be coached while working on the unit, and defects could be pre-empted. Bolt offered semi-supervised AI techniques that required far less training data, time, and effort to deliver the high-quality detection that Drishti pioneered. Drishti ported its products so they could be deployed anywhere on the cloud-to-on premises continuum, to serve manufacturers with different information security requirements. Drishti was a SaaS vendor, selling the above capabilities as targeted solutions on an annual subscription basis. i Standardized work is the process of establishing precise procedures for each operator, taking into account the task time, sequence of operations, and standard inventory needed for each task. For more information visit: https://www.lean.org/lexicon-terms/ standardized-work/. 6 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Understanding Customer Value By October 2022, Drishti served about 25 global customers in three industries—automotive, medical devices, and electronics—including such huge corporations as DENSO, Ford, and HELLA. Customers experienced signifcant ROI realization in as little as 10 weeks. Examples reported by Drishti included a 20% throughput improvement at an airbag factory in Mexico, a 50% reduction in Kaizen duration at a Japanese Tier 1 auto supplier in Michigan, a 50% reduction in training time of new associates, and a 50% reduction in defect rates and 30% reduction in labor cost at an electronics contract manufacturer in Mexico and power tools manufacturer in the Midwest. DENSO, a $46-billion global automotive components manufacturer, used Drishti for root-cause analysis, process design and improvement, operator training, and in-station operator assistance. DENSO sought to achieve signifcant process gains while simultaneously helping its workforce add greater value. According to Drishti, key drivers of Drishti’s Trace product value for DENSO included:12,13 1. Visibility: Live streaming and instant replay, video sharing, collaboration, and time-based studies. 2. Traceability: Root cause analysis and birth certifcate/bill of process for a product. 3. Training: Standardized work and spot training. Trace helped reduce defects, accelerate training, improve processes, and enhance operator health and safety. Furthermore, its Flow product offered DENSO: 1. Knowledge: Creation of a large-scale human data set. 2. Insight: Identifcation of bottlenecks and opportunities for variability and line balance. 3. Action: Faster, more impactful Kaizen processes (manufacturing A/B testing) and use of outliers for learning/best practices. Flow enabled industrial engineers to deliver productivity improvements by analyzing the large and continuous data set. Finally, Drishti’s Assist product allowed line workers to execute their work consistently. Assist fagged potential deviations from Standardized Work, giving associates the opportunity to fx issues immediately while still at the station, avoiding costly rework, and to learn on the job and build skills. Raja Shembekar, DENSO’s vice president of process innovation, said there is a famous quote stating humans are incredibly smart but slow, and machines are incredibly stupid but fast. Drishti lets him marry the best of the two and simplifes the process of identifying issues while letting line associates and team leaders focus on what they do without changing their behavior.14 From the gathered feedback, DENSO engineers and leaders could make decisions on design and organizational management. Similarly, HELLA,15 a German-based global supplier specializing in high-performance lighting and automotive electronics, used Drishti’s line variability and trend charts from its Trace and Flow products to surface cycle time anomalies and examine videos. The results indicated a 7% decrease in cycle time, a 5% increase in productivity, and a 4% increase in overall equipment effectiveness. HELLA’s production teams used Drishti daily insights reports to recognize, confrm, and prioritize actions for the next 24 hours. HELLA implemented Drishti across many of its plants globally. 7 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Drishti developed the Trace, Flow, and Assist capabilities after beginning an ongoing dialogue with its frst customers, Flex and DENSO, in 2016. That dialogue focused on how customer needs can best be fulflled, with white space innovations that addressed needs that had not been recognized. When approaching Flex and DENSO, Drishti frst had to show that a problem existed to be solved. Drishti white space innovation followed this pattern: First, Drishti addressed a data gathering and collection white space. In traditional manufacturing improvement studies, a manager may physically observe an operating line with a stopwatch, taking a few hundred sample measurements—and possibly not seeing certain defects. By automating that process, Drishti captured every measurement for a line for each minute, hour, and shift to expand the customer’s understanding of its own line. This provided a record for individual line operators. With Drishti, the operators who were most effcient and produced the fewest defects were highlighted. Every cycle was recorded, tracked, and understood. When a defect did occur, a record existed. This eliminated the unproductive fnger- pointing that often occurs when others cannot objectively review what happened. Ultimately, Drishti data collection empowered operators. Second, Drishti was an artifcial intelligence solution within manufacturing. It used AI in its metrology platform to create datasets, and in its applications where the data can be more easily analyzed and presented to customers to help solve their operations problems. Through AI capabilities, on-the-job training could be provided even as line associates were working. Employee safety could also be improved by dynamically measuring the ergonomic workloads. Drishti addressed potential concerns related to the capture, storage, and security of personally identifable data, in addition to the negative perception of surveilling employee populations. In canvassing operators on the line, Drishti leadership found that providing its “record of truth” was welcomed and valued. Drishti’s video capture technology blurred the operators’ identities. A company could identify employees because it knew the schedule and stations of those employees, but by itself the video data did not store any personally identifable information. See Exhibits 2 and 3 for illustrations of Drishti capabilities. Business Model Choices: Product vs. Service Drishti continued to face challenges in the iterative development of its core products. As with many start-ups, the valuation of Drishti was based on growth as opposed to proftability in the early stages. It was vital for Drishti to understand customer and technical support requirements and capabilities before expanding its customer base and establishing its sales cycle.16 Because the sales cycle involved the iterative back-and-forth with a customer regarding both price and value, Drishti had to be able to answer the following questions: How long will that price and contract exist? As new features and supporting services are created, how are those incorporated into the pricing of the product? How should Drishti prioritize new features and supporting services? 8 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 In most software start-up companies, the strategies and frameworks to tackle such questions are created during the experiences with the frst few customers, setting the stage for rapid growth and scale as the start-up matures. This emerges as a collaborative exploration process with these initial customers. This process was more complex for Drishti, whose core technology required some level of personalization for each customer. Akella realized that Drishti must walk a tightrope in its development. It could not customize a solution to 100% of each customer’s specifcations because that would be too costly and slow-moving. Drishti must provide a consistent foundation, but its base must be confgurable so that its most valuable capabilities were adjustable for each customer. To grow strategically and walk this tightrope, the Drishti team expanded systematically (see Exhibit 4). In Drishti’s frst two years, Akella’s focus and investment were on addressing the hardest problem—the development of the core product technology. With the onboarding of the frst few customers, a customer experience team was built to both quickly address customer issues and to directly feed insights from the feld back to the product team. Application support and a customer success team followed. Finally, a sales team was established. This growth strategy allowed Drishti to better understand its customers’ needs before scaling its business model. In July 2021, to strengthen and broaden Drishti leadership, Akella and the company recruited an experienced chief executive, Gary Jackson, to be Drishti CEO as Akella transitioned to board chairman.17 Vision for the Future Drishti’s next steps were two-fold. First, the capabilities of the core technology would continue to expand and be able to mesh with a greater range of customers. Second, Drishti aimed to diversify its customer base by targeting more industries and usage applications. With its established core product and value proposition, Drishti could focus on rapid customer acquisition and revenue growth. Considerations in fulflling this vision included several strategic choices. What key characteristics and values should Drishti look for in future customers? Should it aim horizontally across industries or vertically, deeper into a few select industries? How should Drishti leverage its technology? To what extent would it customize per customer? And, how would it determine the pricing of products that had no predecessors in the market? 9 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Exhibits Exhibit 1 Drishti Technologies Company History Source: Created by the case authors. Exhibit 2 Drishti Architecture Overview Source: Created by the case authors. 10 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Exhibits (cont.) Exhibit 3 Drishti Solution Technology Stack Source: Created by the case authors. 11 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Exhibits (cont.) Exhibit 4 Drishti Leadership Team Evolution In Drishti’s early years, CEO Prasad Akella focused on investing in and building the core product technology and the product story. Then, as the sales cycle was iterated, the customer experience, fnance, operations, and other teams were added. Sales became a greater focus as Drishti entered 2020. In July 2022, Gary Jackson succeeded Akella as CEO. Source: Created by the case authors. 12 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Drishti Technologies Inc.: Managing Operations through Computer Vision, AI, and Video Analytics W19C80 Endnotes 1 “Trends in Manufacturing Productivity, 1990–2019.” Bureau of Labor Statistics, 7 Oct. 2021. https://www.bls.gov/opub/ ted/2021/trends-in-manufacturing-productivity-1990-2019.htm. Accessed 29 June 2023. 2 “The State of Human Factor Analytics.” A.T. Kearney, 2018. https://info.kearney.com/30/2769/uploads/the-state-of-human- factory-analytics.pdf?intIaContactId=BScnpdElTmEtvqn7k6tggQ%3D%3D&intExternalSystemId=1. Accessed 31 July 2023. 3 “About ImageNet.” ImageNet. https://www.image-net.org/about.php. Accessed 15 Aug. 2023. 4 Karpathy, Andrej. “Software 2.0.” Medium, 11 Nov. 2017. https://karpathy.medium.com/software-2-0-a64152b37c35. Accessed 29 June 2023. 5 “World Robotics 2021.” International Federation of Robotics, 28 Oct. 2021. https://ifr.org/downloads/press2018/2021_10_28_ WR_PK_Presentation_long_version.pdf. Accessed 29 June 2023. 6 “Collaborative Robot Market.” Markets and Markets. https://www.marketsandmarkets.com/Market-Reports/collaborative-robot- market-194541294.html. Accessed 4 Aug. 2023. 7 Chaudhury, Krishnendu. Math and Architectures of Deep Learning, Manning, 2023. 8 @elonmusk. “Yes, excessive automation at Tesla was a mistake. To be precise, my mistake. Humans are underrated.” Twitter, 13 Apr. 2018, 3:54 p.m. https://twitter.com/elonmusk/status/984882630947753984. Accessed 29 June 2023. 9 “The State of Human Factor Analytics.” A.T. Kearney, 2018. https://info.kearney.com/30/2769/uploads/the-state-of-human- factory-analytics.pdf?intIaContactId=BScnpdElTmEtvqn7k6tggQ%3D%3D&intExternalSystemId=1. Accessed 31 July 2023. 10 “Action Recognition Video.” Drishti. https://bit.ly/Drishti-ActionRecognition. Accessed 13 Nov. 2023. 11 “Drishti Essentials.” Drishtri, 8 May 2023. https://akellas.org/drishti-essentialswithsubtitles-221009/. Accessed 13 Nov. 2023. 12 “DENSO: $46B Japanese Auto Tier 1 Manufacturer.” Drishti. https://web.archive.org/web/20230605143729/https://drishti.com/ case-studies/denso. Accessed 29 June 2023; “Drishti Trace and Flow: AI Video Analytics and Traceability for DENSO’s Manual Assembly Lines.” Drishti Video. https://bit.ly/Drishti-Denso-AIPoweredProduction. Accessed 13 Nov. 2023. 13 “How AI Extends the Toyota Production System.” Drishti. https://bit.ly/Drishti-Liker-AIExtendsTPS. Accessed 29 June 2023. 14 “How AI Extends the Toyota Production System.” Drishti. https://bit.ly/Drishti-Liker-AIExtendsTPS. Accessed 29 June 2023. 15 “HELLA: Global Tier One Automotive Supplier.” Drishti. https://web.archive.org/web/20230830085413/https://drishti.com/ case-studies/hella. Accessed 29 June 2023; “Drishti Hella AI Powered Production.” Drishti. https://bit.ly/Drishti-Hella- AIPoweredProduction. Accessed 13 Nov. 2023. 16 Joisa, Srida. Personal interview. 10 Mar. 2021. 17 Akella, Prasad. “Thinking Long: Prasad Akella on Drishti’s New CEO, Gary Jackson.” Drishti blog, 20 July 2021. https://web. archive.org/web/20230629060820/https://drishti.com/resources/blog/thinking-long-prasad-akella-on-drishtis-new-ceo-gary- jackson. Accessed 29 June 2023. 13 This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.  Established at the University of Michigan in 1992, the William Davidson Institute (WDI) is an independent, non-profit research and educational organization focused on providing private-sector solutions in emerging markets. Through a unique structure that integrates research, field-based collaborations, education/training, publishing, and University of Michigan student opportunities, WDI creates long-term value for academic institutions, partner organizations, and donor agencies active in emerging markets. WDI also provides a forum for academics, policy makers, business leaders, and development experts to enhance their understanding of these economies. WDI is one of the few institutions of higher learning in the United States that is fully dedicated to understanding, testing, and implementing actionable, private-sector business models addressing the challenges and opportunities in emerging markets. This document is authorized for use only in Prof. M P Ram Mohan & Prof. Viswanath Pingali's Senior Management Programme (SMP-BL13) 2024 at Indian Institute of Management - Ahmedabad from Apr 2024 to Oct 2024.