Digital Transformation of Business

Digital Technology Transforming Business

• In the early 2020s, digital technologies significantly reshaped businesses across industries
• Business process automation & embedded intelligence through cognitive computing led to major advancements
• Real-time data analytics decisions became prominent

Use Cases of Digital Transformation

• Retail: Personalized suggestions for online shoppers through analysis of purchase data
• Finance: AI algorithms checked billions of transactions for fraud
• Healthcare: Sequenced human genome predicted effective therapies
• Marketing: Ads personalized to drive consumer purchasing decisions

Big Data and AI

• Big data and AI were used to build predictive algorithms.
• These algorithms improved the efficiency and effectiveness of business processes.

Traditional Business Process Compared with New Business Processes

• Traditionally, machines replacing humans was believed most efficient for business processes
• The traditional approach aimed to remove variability introduced by human involvement
• The traditional approach led to significant investments in process automation; Robots are deployed for intricate tasks
• Both traditional shop floor process improvement and robots pose limitations

Drishti Technologies Solutions

• Drishti Technologies Inc. suggests marrying the skills of humans and machines for optimal results
• This is especially true in hard-to-automate areas

Traditional Process Monitoring

• Traditional manufacturing involves hundreds of workers at shop-floor stations to build products
• Production rates, quality & safety data were on hand-generated charts/dashboards
• Key metrics included process efficiency, product quality, worker safety, productivity, rework & throughput
• Organizations leverage time/motion studies, control charts, visual management, etc. to improve metrics
• Significant investment into employee training ensures adherence to processes and minimize variance
• Manufacturing productivity accounted for 20% of the nation’s growth between 1990-2017

Limitations of Traditional Methods

• It was almost impossible to monitor all processes, products & employees in real-time using traditional methods
• Traditional sampling, while informative if done correctly, could be easily misunderstood and performed incorrectly, especially in larger operations
• Sampling studies lacked specificity to derive thorough analytics
• Sampling isn't product specific when problems arise
• Sampling fails to identify in real time when issues are introduced by new processes/employees
• Observing and measuring every movement, station, and unit in manufacturing is valuable for maximizing efficiency

The Limitations of Robots

• Robot production lagged in 2018, only 28% factory tasks are automated
• Managing human errors remained significant
• Continuous process improvements and worker training addressed these concerns traditionally
• Digital automation was often questioned for over-emphasizing simplicity and process visibility
• Equipping facilities with IoT devices in the early 2000s was expensive; It was needed to collect, store, and process data for valuable analytics models
• Manufacturing industries were slow in adopting technology solutions

Computer Vision Introduction

• Computer/machine vision allows new levels of transparency to manufacturing through use of computer processing.
• It captures, interprets, and responds to visual stimuli like a human visual system

Decreasing Costs

• IoT devices and sensors become cheaper, processing power (on-premise servers or cloud) is more accessible
• Since 2012, data-driven programming based on AI tech surged with the introduction of ImageNet
• Traditional manufacturing organizations became more data literate
• Automotive companies compete to build technologically connected innovations
• Using big data insights, product developments were driven by data and automation

Drishti Launch

• Recognizing data importance for decision-making, Drishti launched
• Revolutionizing data collected on shop floors on a fundamental rewrite productivity and quality quantification
• Drishti's AI driven insights comparable to Henry Ford & Frederick Taylor's time and motion studies

Drishti Technologies: "Eye" in Innovation

• In 2016, Dr. Prasad Akella founded Drishti in Palo Alto, CA
• The company provides insights into production floor's manual assembly processes
• "Drishti" sanskrit for "vision" to empower workers in factories, warehouses, and restaurants
• Creates digital data sets of manual manufacturing via digital cameras
• Technology answers 3 questions for customers in manufacturing: "What just happened?", "What is happening now?", and "How can we improve what is happening next?"

Drishti Technologies Core Questions

• What just happened?: Drishti's tech captures video and annotates, supervisors search by timestamp or product
• What is happening now?: Live streams of current operations, annotations of movements
• How can we improve?: AI and machine learning enhance efficiency and productivity

Action Recognition Technology

• Similar to Google and Apple mining text/audio, Drishti identifies specific actions occurring in a video stream via "action recognition" tech
• Revolutionizes process analysis; integrates the strengths of both AI and human flexibility
• Technology translated to ROI opportunities via improvements in productivity & training

Drishti Camera Method

• Method starts by installing cameras on a factory's floors
• Video uploaded and process, enables a manager to search using timestamps/identifiers
• Video used to train employees, improves models on cycle time, machine performance, and product quality
• Traditional methods use small samples of visual/measured data
• Drishti gives a comprehensive/real-time factory floor views
• AI replaces traditional time and motion studies, offers unattainable insights to increase performance in the factory

Venture Origins

• In addition to Akella, Dr. Krishnendu Chaudhury and Dr. Ashish Gupta played key roles in Drishti’s beginning
• Dr. Akella pioneered using robotic tech to extend human capabilities

Prasad Akella Background

• Dr. Akella founded categories using tech to extend abilities
• 1990s, lead team from GM, Northwestern, and UC Berkeley
• Team created intelligent assist devices, "cobots."
• The global cobot market stands at $1.2B in 2023, projected $6.8B in 2029

Akella's Executive Roles

• Akella served as an executive for Spoke, SAP, Thomson Reuters & Intermedia
• Received IEEE's Anton Philips Award, the IEEE/IFR Entrepreneur Award, and was a Technology Pioneer at the World Economic Forum
• He was awarded 8 US patents and published scientific journals
• Akella gained PhD in Robotics from Stanford University

Chaudhury's Contributions

• Drishti co-founder/CTO was a computer vision/deep learning expert
• 20+ year career, was part of tech innovations at Google, Flipkart, and Adobe
• Chaudhury drove creation for action recognition tech, challenge pushed boundaries of computer vision and deep learning
• Lead author of "Math and Architectures of Deep Learning"
• Gained PhD in Computer Science from Kentucky and BS from Jadavpur University

Gupta's Background

• Drishti co-founder/board member looked for unexplored problem spaces
• His career spanned inventions/innovations in the technical/business worlds
• Helped invent virtual DBs at Junglee, business process outsourcing at Daksh, travel at (MakeMyTrip/MMYT), and venture fund
• Gupta earned PhD in Computer Science from Stanford University
• Awarded the President of India’s Gold Medal

Drishti Growth

• Operating in Silicon Valley, India, Europe & Japan, employing 225+ employees
• Won the World Economic Forum Technology Pioneer, Forbes AI 50, an NVIDIA Top 5 AI company, a NSF SBIR awardee
• Company raised $25M Series B (June 2020), $10M Series A (May 2018), $2.5M seed (June 2017)
• Investors included Andreessen Horowitz, Alpha Intelligence, Benhamou Global Ventures, HELLA Ventures, Toyotas AI Ventures

Drishti Goals and Products

• They want to augment humans in the manufacturing industry, humans are important in factories
• In April 2018, Tesla CEO Elon Musk tweeted that excessive automation at Tesla was a mistake and that humans are underrated
• Humans still complete 72% of factory tasks, creating 71% of value, introduce 68% of defects

Time and Motion Studies

• Early methods improved productivity, manufacturers used time & motion studies for management, time and motion studies were important in scientific management
• Field pioneered by Frederick Taylor, Frank and Lilian Gilbreth (late 1800s/early 1900s)

Studies Implementation

• In the field, industrial engineers went to the factory floor
• Recorded times, analyze data to identify bottlenecks
• Gathering insights created generalization challenges
• When employees knew of being observed, they could behave differently
• It was impossible to observe all employees, so Drishti leveraged emerging technologies (computer vision, AI, and machine learning)

Product Capabilities

• Drishti products centered around 2 core capabilities
• Product 1: Action recognition tech, tasks automated, video streams analyzed using cameras
• Product 2: Delivery through integrates product, associates in the manufacturing line can use Drishti without understanding the science

Software Products Functionality

• Products tailored for use in the manufacturing vertical, delivered KPIs which deliver the needs of professionals for usage in manufacturing

Three Step Process

• Step 1. Video Capture: Installing cameras in the factory to track and record processes
• Step 2. Data Creation: Video searchable where data is organized/stored and includes time/motion, annotations/tagging
• Step 3: Integration quality and industrial engineers to apply Drishti-generated data to other software for continuous improvements

AI-Powered Video Analytics

• Drishti relies on AI-powered video analytics to deliver insights in time
• The software IDs behavior and patterns, generates alerts, and facilitates analysis allowing for smart decisions

Primary Products Capabilities

• Trace: Use live and recorded video, provides address so what just happened and what is happening now
• Flow: Measure production with cycle identification and tracking, used data/insights powered by AI, optimizes a stable manufacturing process
• Assist: Advanced AI product that recognizes human input through step recognition, understand/improves in-cycle processes
• Bolt: High quality with less training
• Deployed on-premises and on the cloud

Understanding Customer Value

• By Oct 2022, Drishti had about 25 global customers in the automotive, medical devices, and electronics businesses
• Customers saw a major ROI realization in 10 weeks
• Increased the throughput at a factory in Mexico by 20%
• Reduced Kaizen duration at a Japanese auto supplier 50%
• Reduced new associate training by 50%
• Reduced defect rates by 50%

DENSO

• DENSO sought to achieve significant process gains while helping its workforce add value
• Key drivers of Drishti's Trace product value: visibility, traceability, and training to standardized processes

Improved Health and Safety

• Trace reduces defects, improves training, enhances the process and enhance health by improving human operation
• Creates a large-scale human data set
• Helps identified bottlenecks
• Action makes Kaizen processes and best practices more efficient
• Allows line worker to perform task accurately and consistency
• Drishti allows human to work together with machines

Raja Shembekar Quote

• Raja Shembekar quotes Humans smart but slow, machines are stupid but fast
• Drishti simplifies identification of issues while letting associates without change
• DENSO leaders and engineers can do organizational management
• It's trace and flow products can produce a 7% decrease in cycle time, a 5% increase productivity and %4 increase in quality
• HELLA uses daily reports to recognize and prioritize daily task

Drishti: Flex and DENSO

• Drishti focused its dialogue on customer needs best filled
• Drishti first had to show a problem existed to be solved
• When approaching Flex and DENSO, Drishti innovation followed with this pattern
• Addressed a data gathering as well as collection
• Captured every measurement for a line expanding customers overall analysis
• With Drishti, efficient operators and defects were highlighted
• Drishti data collection aided operators
• Solution was developed within manufacturing, uses AI to improve jobs, provides safe measurements and ergonomic support

Concerns for Security and Personally Identifiable Information

• Drishti leadership discovered by providing the 'record of such' was valued
• A company knows employee info from schedule and stations, video data doesn't store identifiable employee data

Focus Challenges

• Was based on growth as opposed to early profitability
• Had to understand customer and their technical support
• Price and contract exist
• How can those features and service support exist in order will that particular pricing
• What features had to be prioritized?

Product Solutions

• Strategies and frameworks tackle questions and are created when the stage matures and company scale up.
• Technology requires personalization
• Akella could not customize solutions that are costly so, she must walk a tightrope in order that development wouldn't be. Also it must be customizable so what is applicable must be adjusted to each customer. So they can grow what team can be made better and systematically fix it
• Akella focused in investing and addressing in the hardest problems
• Customer experience was built to fix customer issues
• Next application support, customer support and team and Finally sales. The growth allowed and understand the custom before scaling

Two Fold Steps

• Core tech would continue to expand customer needs
• Diversity can be built customer core product value
• Strategic Choices that look at characteristics and what values should Drishti look at future customers

Questions moving Forwards

• Should It look at horizontal across industries
• How should It leverage tech?
• How would it determine pricing that wasn't in the market