Download as PDF, PPTX
Uploaded byDr. Tathagat Varma
Building an AI Startup
The document outlines key considerations for building an AI startup, emphasizing the importance of addressing real-world problems that potential customers are willing to pay to solve. It discusses factors contributing to startup failures, such as the lack of market demand and the distinction between genuinely AI-enabled businesses and those masquerading as such for investment appeal. Additionally, it highlights the challenges faced in adopting AI technology, including low ROI and scalability issues, underscoring the need for a clear understanding of the specific problems AI can effectively address.
More Related Content
Similar to Building an AI Startup
![[DSC DACH 24] Harnessing AI for Sustainable Results: Strategies for Long-Term...](https://cdn.slidesharecdn.com/ss_thumbnails/johannesschauer-240921154533-335db038-thumbnail.jpg?width=640&height=640&fit=bounds)
Building an AI Startup
- 1. BUILDING AN “AI STARTUP*” (* Y E A H … W H AT E V E R T H AT M E A N S ! ) TAT H A G AT VA R M A S T R A T E G Y & O P E R A T I O N S , W A L M A R T D O C T O R A L S C H O L A R , I N D I A N S C H O O L O F B U S I N E S S
- 2. DISCLAIMER T H ES E A R E M Y P E R S O N A L V I E W S !
- 3. TOP 3 REASONSTO “BUILD” A STARTUP (ANY STARTUP !) 1. Solve a real-world problem! 2. Solve a real-world problem!! 3. Solve a real-world problem!!! PS: there are other equally meaningful and “rewarding” ways to pursue interesting ideas, complex problems, hobbies, and passions, etc. Let’s just don’t call them a “startup”!
- 4. GENERAL PRE-CONDITIONS TOSOLVING ANY PROBLEM BY BUILDING A STARTUP THERE IS A PROBLEM WORTH SOLVING! SOMEONE WANTS IT SOLVED! YOU ARE INTERESTED IN SOLVING IT!
- 5.
- 6. WHY IS THEMARKET IMPORTANT? • Markets that don’t exist don’t care how smart you are. • In a great market - a market with lots of real potential customers - the market pulls product out of the startup. • The #1 company-killer is lack of market. - Marc Andreesen https://web.stanford.edu/class/ee204/ProductMarketFit.html
- 7. SO WHAT KINDOF PROBLEM TO SOLVE? Potentially affects many people, or a hard or a long- term problem Affects only a few people, or an easy or a short-term problem People can or want to solve it themselves (or even willing to live with the status quo) People potentially ready to pay others or buy a solution or service to solve their problem Touch Me Not: Even if the problem is interesting enough, you probably have no market because people are willing to solve in “Do ItYourself” mode, or even live with the pain but not buy a solution or pay someone else to do it! Self-Help Groups: Perhaps a limited market, but might offer interesting ideas to study, understand and maybe even develop it further! Ideal Opening: Despite looking like a small market, maybe offers interesting opportunity ahead, and a “natural” MVP testbed to try out some ideas Perfect Problem: Ideal end-state scenario, but maybe already too crowded (it is it too good, expect cutthroat competition already!), and hence not a great choice for entry strategy
- 8. WHAT TO BUILD? “Don’tsell what you can make; make what you can sell.And that means figuring out what people want to buy.” - Alistair Croll and BenjaminYoskovitz, Lean Analytics: Use Data to Build a Better Startup Faster (Lean Series)
- 9. SO, HOW TOKNOW WHAT PEOPLE WANT? Primary Research Secondary Research User reviews Interviews Observations Polls Focus Groups Surveys Literature survey Internet Industry reports Demographic data Customer Feedback Trials Experiments Ethnographic Research
- 10. HOW DO YOUKNOW YOU HAVE FOUND THE RIGHT PROBLEM? The key to qualitative data is patterns and pattern recognition. Here are a few positive patterns to look out for when interviewing people: • They want to pay you right away. • They’re actively trying to (or have tried to) solve the problem in question. • They talk a lot and ask a lot of questions demonstrating a passion for the problem. • They lean forward and are animated (positive body language). - Alistair Croll, and BenjaminYoskovitz. Lean Analytics: Use Data to Build a Better Startup Faster
- 11. SO, WHAT’S AN“AI COMPANY”? • Clearly, no one single definition! • AI company,AI-enabled company,True AI company,AI-led company,AI factory,AI-first company…. • Bigger question to ask is why even bother about an “AI company”…after all, do you ever say it is an “electricity company” (rather than a lighting company) or a “timber company” (rather than a furniture chain) or a “milk company” (rather than a sweets shop), a “currency company” (rather than a financial services), etc.
- 12. STILL, WHAT’S ANAI COMPANY? “A firm that assists, augments and potentially replaces human judgment in decision-making by delivering socially-acceptable and appropriately explainable prediction solutions by data-driven continuous learning.” - My view (WIP!)
- 13. HOWEVER…NOT EVERYTHING THAT GLITTERSIS GOLD!!! • A new report from London-based venture capital firm MMCVentures found no evidence that artificial intelligence was an important part of the products offered by 40 percent of Europe’s 2,830 AI start-ups. • The findings raise questions about how the term AI has become a blanket phrase for start-ups looking to attract investments. https://www.cnbc.com/2019/03/06/40-percent-of-ai-start-ups-in-europe-not-related-to-ai-mmc-report.html
- 15. WHAT KIND OF PROBLEMS SHOULDAI BE USED FOR? “If a typical person can do a mental task with less than one second of thought, we can probably automate it using AI either now or in the near future.” - Andrew Ng, https://hbr.org/2016/11/what-artificial- intelligence-can-and-cant-do-right-now
- 16. CHOOSING PROBLEMS FOR“AI STARTUPS” “Investors are interested in startups that are building tailored AI solutions for previously unsolvable problems,” said Jay Srinivasan, who is the CEO and co-founder of atSpoke.“So focus on areas where there are many inefficiencies and repetitive human processes, such as call centers and back-office paperwork processing. Investors want successful AI solutions that address specific workflows and problems, such as a legal document review.” - https://www.forbes.com/sites/tomtaulli/2020/08/08/how-to-create-an-ai- artificial-intelligence-startup/?sh=44cd988719a4
- 17. WHAT ARE YOUR“PAIN METRICS” • Hollinger, Ph. D., who directs the National Retail Security Survey, retailers lost 1.7 percent of their total annual sales to inventory shrinkage last year.The surveyed portion of the retail economy transacts over $1.845 trillion dollars annually, making the loss worth over $31.3 billion. • Based on a number of different ecommerce studies, the average shopping cart abandonment rate is 68.81% with the most recent study showing 74.52%. • Indian public sector banks collectively owed approximately 6.17 trillion Indian rupees in non-performing assets in fiscal year 2021.
- 18. SOLVING THE DATAISSUE! • Problem domain, customers, etc. • Government (data.gov, data.gov.in, data.worldbank.org, etc.) • Companies (PhonePe Pulse, https://cloud.google.com/solutions/datasets, etc.) • Data brokers • Synthetic data • …
- 19. ESTABLISH THE “FINISHLINE” “Without productivity objectives, a business does not have direction.Without productivity measurements, it does not have control.” - Peter Drucker, Management:Tasks, Responsibilities, Practices
- 20. LEAN ANALYTICS APPROACH Much ofLean Analytics is about finding a meaningful metric, then running experiments to improve it until that metric is good enough for you to move to the next problem or the next stage of your business. - Alistair Croll and BenjaminYoskovitz. Lean Analytics: Use Data to Build a Better Startup Faster (Lean Series) (p. 27). O'Reilly Media. Kindle Edition.
- 21. ADOPTION ISSUES GALORE! •87% pilots never make it to production - https://venturebeat.com/2019/07/19/why-do-87-of-data- science-projects-never-make-it-into-production/ • 70% of companies report minimal or no impact from AI. • It was estimated that 85% of AI projects will fail and deliver erroneous outcomes through 2022.
- 22. SOME FAILURES OFAI… • IBM’s “Watson for Oncology” Cancelled After $62 million and Unsafe Treatment Recommendations • Microsoft’s AI Chatbot Corrupted by Twitter Trolls • Apple’s Face ID Defeated by a 3D Mask • Amazon Axes their AI for Recruitment Because Their Engineers Trained It to be Misogynistic • Amazon’s Facial Recognition Software Matches 28 U.S. Congresspeople with Criminal Mugshots https://www.lexalytics.com/lexablog/stories-ai-failure-avoid-ai-fails-2020
- 23. ROI IS ELUSIVE! TheROI for AI projects varies greatly, based on how much experience an organization has. Leaders showed an average of a 4.3% ROI for their projects, compared to only 0.2% for beginning companies. Payback periods also varied, with leaders reporting a typical payback period of 1.2 years and beginners at 1.6 years. https://www2.deloitte.com/us/en/insights/industry/technolog y/artificial-intelligence-roi.html
- 24.
- 25. SCALING-UP ISSUES • Firm-wideend-to-end use cases • Experimental learning,A/B testing, feedback • Model performance, Concept drift • Data governance, bias • Explainability, transparency, traceability • Social, ethical and safe • Industrialization of AI, MLOps • Internal organization design commensurate to delivering the goals of an AI-led ROI • …
- 26. RECAP • Technology ≠Startup! Proving a technology in the lab is rather “easy”, building a startup around the idea is an altogether different ball game! • Think ROI! While today’s AI is all about data, think of what business problem you want to solve and why? • Data delivers value! While data is the biggest source of competitive advantage for AI companies, it is also the bane. However, there are options available. • Outside-inThinking! Think of building your AI startup “outside-in” rather than “inside-out”