20VC YC's Daniel Gross on How YC Can Democratise AI & Reduce Incumbency Advantages, Why ML Enabled Software Will Eat The Software That Ate The World & Whether AI Will Produce Independent Companies or Be Technology within Incumbents

Abstract
Summary Notes

Abstract

In this episode of "20 minutes VC," Harry Stebbings interviews Daniel Gross, a partner at Y Combinator (YC), the world's leading startup accelerator. Gross shares his journey from an Israeli settlement to Silicon Valley through YC, which led to his company being acquired by Apple and his eventual return to YC as a partner. He discusses the significance of AI, noting its potential to revolutionize industries despite skepticism about its hype. Gross emphasizes the importance of democratizing AI to prevent large companies from dominating the field and outlines YC's efforts to support AI startups by providing expertise, computing resources, and data. He also touches on the importance of managing one's psychology for career success and shares productivity tips, including the benefits of decaf coffee and meditation.

Summary Notes

Introduction to Daniel Gross and Y Combinator

  • Daniel Gross is a partner at Y Combinator (YC), a highly successful accelerator.
  • YC alumni include Airbnb, Dropbox, Reddit, and Flexport.
  • Daniel has an impressive angel investment portfolio with companies like Opendoor, Cruise, Gusto, and GitHub.
  • Daniel was previously a director at Apple, focusing on machine learning after his company was acquired by Apple in 2013.

You are listening to the 20 minutes VC with me, Harry Stebings at H Debbings with two B's on Snapchat and it's day one of YC demo Day. And here at the 20 minutes vc we like to bring you the most current and timely of interviews. And so today's show, I'm thrilled to welcome Daniel Gross, partner at Y Combinator.

This quote introduces Daniel Gross and emphasizes the timeliness and relevance of the interview, given it coincides with Y Combinator's demo day.

Daniel heads up all things Ycai, having been a director at Apple, where he focused on machine learning as a result of his prior company. Q also a YC company being acquired by Apple in 2013.

This quote provides background on Daniel Gross's career and his connection to Y Combinator and Apple.

Daniel Gross's Journey to Silicon Valley and Y Combinator

  • Daniel Gross came to Silicon Valley through Y Combinator.
  • Originally from Israel, he applied to YC and was accepted after an interview process.
  • He moved to Silicon Valley to participate in YC, despite an initial setback during the interview.
  • Daniel's initial idea was rejected, but he was offered a chance to work on something different.

So I originally came out to Silicon Valley through Y Combinator. I was living in Israel, literally in almost a military bunker in an Israeli settlement, really in the middle of nowhere.

This quote describes Daniel Gross's origins and his initial connection to Y Combinator, highlighting the transformative opportunity YC presented.

And then Paul Graham, as I was walking away, Paul Graham called me back into the room and asked me if I'd be willing to work on something completely different than what I pitched.

This quote captures a pivotal moment where Daniel Gross was given an unexpected opportunity by Paul Graham, co-founder of Y Combinator, to pivot his idea.

The Challenges and Pivot During Y Combinator

  • Daniel experienced several idea changes and iterations during his time at YC.
  • He faced a significant challenge right before Demo Day when Amazon changed their terms of service, impacting his business.
  • Daniel had to quickly come up with a new product idea and build a prototype in 48 hours.

And then right towards the end of the program, I was working on something that was generating quite a bit of revenue using Amazon affiliate marketing.

This quote sets the stage for the challenge Daniel faced, as his business model was tied to Amazon's affiliate marketing program.

And literally 48 hours before Demo Day, which is the culmination of the program, where you, as a founder, get up on stage and present your work to the best investors in the world. Amazon changed their terms of service to deem what I was doing illegal according to their new tas.

This quote highlights the urgency and high stakes of the situation, with a major setback occurring just before the crucial Demo Day event.

Creation of Greplin and Rapid Development

  • Greplin was the new product Daniel Gross created, a search engine for personal content across various platforms.
  • The prototype was built under immense time pressure and was a technical challenge.
  • Despite the rushed development, Greplin attracted significant angel investment and led to a partnership with Robbie Walker, another YC alum.

So I decided to do that. And that is really when I started to build my company. Weirdly enough, in earnest. In those 48 hours, I built a product I always wanted for myself.

This quote illustrates the decisive moment when Daniel Gross chose to pursue the development of Greplin, marking the true start of his entrepreneurial journey.

And I was always a little bit afraid to build because I thought it be too technically challenging at the time it had a really funny name. It was called Greplin.

This quote explains the origin of the product's name, Greplin, and acknowledges the technical challenges Daniel faced in creating it.

Acquisition by Apple and Machine Learning Work

  • Daniel's company was acquired by Apple, where he continued to work on similar products internally.
  • He led various machine learning efforts at Apple, applying his expertise from his startup experience.

And that's what I did. The company was acquired by Apple, and I worked on kind of similar things internally at Apple, and I ran a bunch of different machine learning efforts there.

This quote summarizes the transition from Daniel's startup to working at Apple, where he applied his skills and knowledge to internal projects.## Transition to Y Combinator

  • Daniel Gross acknowledges his deep connection to Y Combinator, crediting it for the start of his career.
  • He expresses enthusiasm for working at Y Combinator, a place that offered him early opportunities.
  • In January, Daniel left Apple to join Y Combinator as a partner, aiming to contribute in a role similar to Y Combinator's founders.

"And so at the beginning of January, I left my oreg at Apple and joined Y Combinator as a partner."

This quote marks the transition of Daniel Gross from Apple to Y Combinator, highlighting his career move and his new role as a partner at the influential startup accelerator.

AI: Hype vs. Reality

  • Daniel Gross does not believe AI is a scam, but acknowledges Silicon Valley's tendency to be overly optimistic about the timing of technological advancements.
  • He suggests that while AI is not meeting the hype, it is not fraudulent and has made genuine progress.
  • Gross divides the technological landscape into three categories: commodities, cutting-edge research, and technologies that have recently become feasible.

"AI today is a little bit like that kind of big data trend everyone talked about in 2005 and six where it's changing the world in a bunch of different ways that I'll detail in a moment, but it's not really living up to, I think, all the hype, and that's probably fine."

Daniel Gross compares the current state of AI to the big data trend of the mid-2000s, acknowledging its significant impact despite not living up to its full hype.

Tangible AI Advancements

  • Gross highlights the ability of computers to "see" within images, a breakthrough in AI perception, as a tangible advancement.
  • He cites Google Photos, Amazon Go, and self-driving cars as examples of AI applications in various domains.
  • Another advancement is in automatic speech recognition (ASR), which has improved to the point of being practically usable.
  • Gross emphasizes the difficulty in predicting when far-end research will become commercially viable, drawing a parallel with the skepticism before the Wright brothers' flight.

"So there are a few different types of machine learning that are kind of culminating in this thing called AI. The first is the fact that computers, as of roughly 2013 or 14, can really see inside images."

This quote highlights a specific advancement in AI, where machine learning has enabled computers to interpret images, leading to practical applications like Google Photos and self-driving cars.

The Future of AI Companies

  • Gross explores whether AI will lead to the emergence of dedicated AI companies or if it will serve as a sustaining innovation for existing incumbents.
  • He notes that while deep learning has been applied in many areas, there haven't been breakthrough independent AI companies yet.
  • Gross presents two possibilities: either it's too early for such companies to have emerged, or AI will become a sustaining innovation for large companies.
  • He underscores the importance of democratizing AI to prevent large companies from monopolizing the benefits due to economies of scale in talent, compute infrastructure, and data.

"I think there are two possible takes on this. One is that we're just early and that's going to happen. And I think that's fairly true."

Daniel Gross suggests that the emergence of independent AI companies is still possible and may occur as the technology matures, indicating an optimistic view of AI's potential to foster new businesses.

Democratizing AI

  • Y Combinator aims to democratize AI by providing startups with the resources typically unique to large companies.
  • YcAI is an experiment and the first Y Combinator vertical dedicated to AI, offering domain-specific perks to participating companies.
  • The initiative seeks to replicate the impact of Amazon Web Services, which leveled the playing field for startups in web hosting.

"Part of my job at Y combinator is to kind of democratize AI again and to try to take those three properties that I mentioned that are today unique to large companies and kind of give them back to startups."

Daniel Gross discusses his role in making AI accessible to startups, countering the dominance of large companies in the field by sharing critical resources and expertise through Y Combinator's AI-focused initiative.## Combatting Large Company Advantages in AI

  • Y Combinator helps founders overcome the advantages large companies have in AI.
  • Founders receive support with talent density, capital costs for model training, and data set acquisition.
  • Y Combinator provides office hours with AI experts, free compute credits, and access to the latest GPUs.
  • They acknowledge the challenge of competing with large companies like Google, especially in data sets.

overcome the exact three proprietary modes I mentioned large companies have. So the founders are able to combat the talent density issue. Founders are able to book office hours, not just with partners like myself, but with true AI experts from folks at OpenAI and Amazon and other companies as well.

This quote explains how Y Combinator helps founders address the challenges of talent acquisition by providing access to experts from leading AI organizations.

Second, to combat the kind of upfront capital cost of training models, we give these kind of young companies in their embryonic phase, not just our standard Y combinator investment, but free compute credits on all the different kind of web services, as well as a company that literally racks gpus, the latest and greatest gpus in San Jose locally, which is needed for a few technical reasons that we can get into if interested, in order to combat that second mode I mentioned.

Y Combinator assists startups with the financial burden of training AI models by providing investments, free compute credits, and access to high-end GPUs.

And third, we are working. I don't yet have a great answer for this, but we are working on empowering these companies with data sets so that they compete with Google.

Y Combinator is actively seeking solutions to help startups gain access to competitive data sets, acknowledging it as a significant challenge in the AI industry.

Specialization as a Competitive Advantage

  • Specialization in data sets can help startups overcome the data incumbency advantage of larger companies.
  • True specialization involves deep diving into niche areas like calendar booking and scheduling.
  • The idea is that by focusing on specific domains, startups can create their own moats.

One kind of corrective action that one previous guest on the show, Dennis Mortensen at X AI, said is that you can build your own kind of incumbency advantage through true specialization of datasets.

Dennis Mortensen suggests that startups can build a competitive edge by specializing in particular types of data, which can help them stand out in the AI industry.

Reducing Friction in Starting a Company

  • Y Combinator has historically reduced the friction involved in starting a company.
  • The program provides structure, investment, and support that may be pivotal in the creation of successful startups.
  • Reducing the activation energy required to start a company can help founders overcome the data moat challenge.

I think that one of the great things Y Combinator did is it reduced the friction involved in starting a company to a level where it was counterfactual in the creation of a lot of these great companies.

Y Combinator's support has been crucial in reducing the difficulties of starting a company, which in turn has led to the creation of many successful businesses.

Transfer Learning and The Future of AI

  • Humans learn with fewer examples than software due to transfer learning across domains.
  • The future of AI may involve software mastering transfer learning, potentially democratizing the field.
  • Nature's efficiency, honed over millions of years, sets a high bar, but humans have succeeded through alternative methods, such as using jet fuel for flight.

That is to say, I'm able to understand an Atari game with far fewer examples than whatever DeepMind can cook up, because I have an understanding of gravity, of. Take Pong as an example of what happens to an object when it collides with another object.

Humans apply knowledge from one domain to another, which is a form of transfer learning that AI is still striving to achieve.

And I kind of wonder if that will be the case here as well.

The speaker is curious whether AI will eventually be able to learn as efficiently as nature or if it will require alternative methods to excel.

The Role of Ambiguity in AI

  • Ambiguity in AI can be indicative of the necessity for advanced machine learning techniques.
  • Simple machine learning or statistical techniques may suffice when there is little ambiguity.
  • Overapplying AI to areas without ambiguity can be counterproductive.

I think ambiguity could be a goal in the sense that when you don't have that much of it, you really wonder if simple machine learning techniques, or frankly even simple stats, statistics techniques could have solved the problem long ago.

Ambiguity in AI problems may signal the need for more sophisticated AI solutions rather than simpler statistical methods.

AI Business Models

  • AI can be applied to verticals creating business models similar to SaaS companies.
  • Platform companies in AI also have potential for business models, though less obvious compared to databases.
  • Incumbents like Google and Amazon are more forward-thinking, making it harder for new companies to establish a dominant platform.

Well, I would disagree. I think that there are going to be business models of taking AI and applying it to a vertical in the same way that you could literally take the advent of the database and apply it to a know software eats the world is a quote, I think originally attributed to Mark Andreessen.

The speaker believes that applying AI to specific industries will lead to viable business models, drawing a parallel to the impact of databases on software.

And so the companies are more forward thinking. And as a result, I think it's going to be a little bit harder for a company to build something that Tensorflow won't do over time.

The speaker acknowledges the challenge for new AI companies to compete with established players that are already innovating and expanding their capabilities.## Self Empowerment

  • The concept of children running the show was a source of fascination and empowerment for the speaker.
  • Re-reading "High Output Management" is a practice that helps maintain the right mental framework for managing people.

"The concept as a child that the kids are running the show was fascinating to me, and it really, I think, gave me a strong sense of self empowerment."

The speaker reflects on how the idea of children in control gave them a sense of empowerment from a young age, suggesting an early interest in leadership and autonomy.

Importance of Managing Psychology

  • Understanding the significance of managing one's own psychology for personal and career success is crucial.
  • Emphasizes the value of emotional intelligence (EQ) over a high IQ.
  • Practices such as meditation and good sleep are important for success.

"I wish I would have known how much managing your own psychology matters to your personal and career success."

The speaker wishes they had understood earlier the importance of psychological self-management for success, indicating a realization that came with experience in the industry.

Meditation and Self-awareness

  • Meditation has been a key practice for the speaker, using the Headspace app daily.
  • Meditation helps in becoming more self-aware and less reactive.
  • It enables acting according to one's goals rather than moods.

"I found it really helpful for me to become just more self aware and less immediately involved in my own line of thinking."

This quote highlights the benefit the speaker has found in meditation, which has helped them to detach from their immediate thoughts and become more self-aware.

The Challenge of Rejection

  • Rejection, especially of companies that have made significant efforts to engage, is a challenging aspect of the speaker's role.
  • There is an effort to provide intelligent and empathetic rejection feedback.

"Rejecting companies, in particular companies that fly from across the world to see us, it's really gut wrenching and the most difficult part of the job."

The speaker reveals the emotional difficulty involved in rejecting companies, showing empathy for the efforts those companies make to pursue their goals.

Framework for Saying No

  • The speaker emphasizes the importance of writing thoughtful rejections that show understanding of the business and clearly explain their position.

"We always try to write rejections, in particular for the interviews where people have flown in that are intelligent."

This quote indicates the speaker's commitment to providing detailed and intelligent feedback when rejecting companies, respecting the time and effort they have invested.

Productivity and Self-Criticism

  • The speaker is self-critical, which they believe drives them to work harder but may conflict with happiness.
  • Productivity hacks, such as drinking decaf coffee for consistent energy levels, are part of the speaker's routine.

"I am very self critical of myself, and in many ways that propels me to work harder."

The speaker acknowledges their self-critical nature as a motivating factor in their productivity, but also recognizes the potential downside in terms of personal happiness.

Reading for Diverse Perspectives

  • Slatestar Codex is a favored blog for its fantastic writing and diverse subject matter.
  • Reading widely is important for maintaining breadth in mental thinking.

"I love reading Slatestar Codex, which is a blog written by a gentleman who is, I think, a physician somewhere in the middle of America."

The speaker expresses admiration for Slatestar Codex, highlighting the importance of exposure to different topics and perspectives outside of their usual Silicon Valley environment.

Investment in Ripling

  • The speaker's recent angel investment in Ripling was driven by a personal desire for the product and its potential to streamline organizational processes.
  • Ripling automates onboarding and offboarding tasks, which the speaker sees as a valuable tool for small companies.

"Ripling just automates all of this. It's that Google Doc, but it's one click."

The speaker explains the functionality of Ripling, a company they invested in, which automates administrative tasks for hiring and terminating employees, emphasizing its simplicity and utility.

Acknowledgments and Resources

  • Thanks are given for the opportunity to be on the show and for the engaging discussion.
  • The speaker's and host's social media are shared for further engagement.
  • Pendo and Treehouse are mentioned as resources for product management and coding education.

"Thank you so much for having me on the show. Absolutely delighted."

The speaker expresses gratitude for being invited to the podcast, indicating a positive experience.

"Pendo helps companies create products that customers love."

This quote introduces Pendo, a platform for product teams, suggesting its usefulness for understanding customer engagement and driving product adoption.

"I went to Treehouse, the online school where you can learn how to build websites and apps."

The host shares a personal anecdote about using Treehouse to learn coding, positioning it as a valuable resource for beginners in technology.

What others are sharing

Go To Library

Want to Deciphr in private?
- It's completely free

Deciphr Now
Footer background
Crossed lines icon
Deciphr.Ai
Crossed lines icon
Deciphr.Ai
Crossed lines icon
Deciphr.Ai
Crossed lines icon
Deciphr.Ai
Crossed lines icon
Deciphr.Ai
Crossed lines icon
Deciphr.Ai
Crossed lines icon
Deciphr.Ai

© 2024 Deciphr

Terms and ConditionsPrivacy Policy