9 Ways to Use Machine Learning to Build Startups YC Wants to See

Part One

by Lily Adelstein

November 5, 2021 5 min read

For years, YCombinator (YC) has published their Requests for Startups (RFS). This is a broad list of ideas that YC is interested in seeing startups work on. The list includes AI, Biotech, and Brick and Mortar 2.0., among 18 others. While AI is given its own bullet point on the list, we wanted to further examine the intersection of Machine Learning (ML) and each of the 21 ideas listed. As Andrew Ng describes it, “Machine Learning is one of the most exciting recent technologies.” It is currently being used in ways that affect people every single day and already we are seeing startups use ML to advance ideas in many of the areas identified on YC’s RFS. In this series, we will go through each idea on YC’s RFS and identify the ways ML is already being used or can be used. Here are the first three categories. More to come... 


AI is changing the way the world does business. YC recognizes this and is interested in seeing startups tackle domain specific problems using AI. While AI very generally encompasses all forms of computer capacity to behave like a human, machine learning, which is a subset of AI, refers more specifically to a computer's ability to learn about something from data, without the need for a human to program that learning explicitly into the computer. For more information on the difference between AI and ML, check out this post. ML allows companies to create tools that are more intelligent, personal and expedite certain tasks. 

Here are a few ways companies using ML to create standout offerings: 

  1. VisualOne is a company using computer vision and custom image detection to create personalized security cameras. Customers identify and train the model on events that they are interested in monitoring — for example, a stove left on, the backdoor open, flooding in the basement — and the system can then alert them if these events are detected.
  2. Datasaur.ai uses ML to create an assisted data labeling tool. Data labeling is a time consuming task; often repetitive and mundane. Datasaur.ai uses ML to learn from data that is already labeled to label other batches of data.
  3. Turing is a company that uses ML to simulate different combinations of ingredients to create the ideal consumer packaged good product, like detergent or packaged food. ML models can be trained on previous combinations of ingredients and past outcomes to better predict future ingredient combinations. For example, a company may be interested in creating a new, better performing laundry detergent. Turing would use past data on the ingredients that went into different types of laundry detergents and how those different types of detergents performed, then predict new, high performing ingredient combinations. 

Image source: Turingsaas.com; Past ingredient combinations and outcomes are used to predict new combinations.


Biotechnology is changing the makeup of the physical world. From the food we eat, how medicine is made and even how we understand and interact with human DNA. Advances in biotechnology open up the possibility of a disease free future but also a more complex relationship between tech and humans. One area that has gotten a lot of attention recently is gene editing. CRISPR is a revolutionary gene editing technology that allows scientists to remove, replace and repair certain genes. 

Here are a few ways companies are using ML now and could use ML in the future to advance innovation in gene editing:

  1. Synthego is a startup using ML to predict the possible effects of CRISPR editing, which could be used to treat or prevent human diseases, like cancer. In a blog post, they wrote, “While efforts for making CRISPR more predictable have been ongoing for years, researchers have only recently been able to model CRISPR editing and repair mechanisms using large well-described data sets and advanced ML models.”
  2. Metagenomics is a startup, founded by researchers at UC Berkeley, using ML to analyze 4 billion years worth of microbial evolution to discover novel enzymes that would improve the efficacy of CRISPR editing. 
  3. University of Toronto researchers have discovered several use cases for ML in genomic medicine including predicting how changes in genetic materials impact phenotypes (think eye color, hair color, freckles). This would allow scientists to better understand the relationship between certain genes and how they are expressed in humans. 

Brick and Mortar 2.0

Our relationship to the physical world has changed and continues to change. Offices, restaurants and public buildings were left vacant during early months of the Covid-19 pandemic while many big box stores shut down even before that due to the popularity of online businesses like Amazon. To start a business in 2021 means, very often, building a website, not a store front. But the future is not black and white, online or offline, instead it is a combination of innovative uses of physical and online spaces. We turn online often for efficiency, but we look to the physical world for experiences. For example, we might want to be able to easily make a reservation for a hot air balloon ride online, check out reviews, and get a receipt for the purchase, but we want to experience the hot air balloon ride in the real world. Startups are taking advantage of this hybrid future and using ML to make data-driven decisions.  

The following are a few ways retailers are using ML to optimize their online and offline presence: 

  1. Intelligent Search Assistant— ML can make employees more knowledgeable and help answer difficult questions from customers. For example, if a customer comes into the store, asks an employee a question, the employee can type it into a search of the company's database to find a relevant answer without needing a supervisor or HQ. ML can classify the question being asked, classify the types of answers that exist, and detect if an answer is spam or low quality and potentially return a high value answer to the employee.
  2. Product Recommendations — In-store recommendation systems can equip employees with valuable recommendations for customers based on personalized store data and customer history of purchases. 
  3. Just Walk Out technology — Amazon is pioneering technology that allows customers to walk into an Amazon Go store, pick up what they want and just walk out. Computer vision allows computers to track and monitor customers, and allows the company to be able to identify who is taking what out of the store. 

The Request for Startups published by YC highlights some major trends in the type of challenges startups are tackling. As machine learning becomes more accessible, the technology will likely play a growing role in the ability for small teams to develop innovative solutions.

The next post in this series will highlight use cases of machine learning in carbon removal technologies, cellular agriculture and clean meat, and responses to Covid-19.

Want products news and updates?

Sign up for our newsletter to stay up to date.