Richard Socher On Selling A Company To Salesforce And Raising $99 Million To Build An AI Assistant Helping Users Be More Productive, Creative, And Informed

Richard Socher, an AI pioneer and accomplished entrepreneur, has a story that exemplifies innovation, resilience, and the ability to transform cutting-edge technology into impactful businesses.

Richard has had an exceptional journey founding and scaling AI-driven companies like MetaMind and You.com. His latest venture You.com has attracted funding from top-tier investors like Marc Benioff, Salesforce Ventures, NVIDIA, Gen Digital, and SBVA (formerly Softbank Ventures Asia).

In this episode, you will learn:

• Richard Richard emphasized the power of first principles thinking in developing neural networks that learn directly from raw data, revolutionizing natural language processing.
• With MetaMind, Richard sought to democratize AI by making neural networks easily accessible to businesses of all sizes, eventually leading to a successful acquisition by Salesforce.
• At Salesforce, Richard’s team pioneered advancements like prompt engineering and AI-generated proteins, proving that big companies can drive groundbreaking research and innovation.
• Richard founded You.com to challenge the ad-heavy search model, aiming to create a productivity engine that combines accurate AI answers with verifiable citations.
• Richard learned that great technology alone isn’t enough—effective marketing and distribution are crucial for driving adoption and revenue growth.
• Guided by mentors like Marc Benioff, Richard cultivated skills in sales, stakeholder alignment, and organizational management to scale his ventures effectively.
• You.com envisions AI as a tool to empower knowledge workers, boost productivity, and enable organizations to unlock their full potential across industries.

 

SUBSCRIBE ON:

For a winning deck, see the commentary on a pitch deck from an Uber competitor that has raised over $400M (see it here). 

Remember to unlock for free the pitch deck template that founders worldwide are using to raise millions below.

About Richard Socher:

Richard Socher is the CEO/Founder of you.com, the first chat-search assistant. He also invests in and mentors other startups at AIX Ventures, where he is the founder and managing partner.

Richard previously served as the Chief Scientist and EVP at Salesforce, where he led teams working on fundamental research, applied research, product incubation, search, customer service automation, and a cross-product AI platform for unstructured and structured data.

Before that, Richard was the CEO/CTO of the AI startup MetaMind, which Salesforce acquired in 2016. Richard received his Ph.D. in computer science at Stanford University in 2014 and later served as an adjunct professor in the CS department at Stanford.

He is widely recognized for bringing neural networks into natural language processing, inventing the most widely used word vectors, contextual vectors, and prompt engineering.

Richard has over 199,000 citations and won awards such as the best computer science PhD thesis awards at Stanford in 2014, honorary doctorate (Dr.-Ing. h. c.) from the TU Dresden, the ICML 2011 best paper award, 2016 Young Global Leader at WEF, the Microsoft PhD Fellowship, the test of time award at ACL 2023 for a paper from 2013, the PAMI Longuet-Higgins Prize for ImageNet, 2024 WEF Technology pioneers with you.com, the 2023 Time Magazine’s Time100 AI, first place at the local LaserTag one random Friday evening and more.

Outside of work, Richard enjoys paramotor adventures and photography.

Connect with Richard Socher:

• LinkedIn
• Tracxn
• TheOrg
• Crunchbase

Read the Full Transcription of the Interview:

Alejandro Cremades: alright Hello, everyone, and welcome to The Deal Maker Show. so Today, we have ah a really awesome founder, you know a founder that has exited you know companies, a founder that has raised money, ah and they and also a founder that knows a thing or two about this whole world of AI and LLMs and and you name it. you know Today, we’re going to be talking about the good stuff that we like to hear, that building, the scaling. ah and and Again, you know there’s going to be a lot that we can learn you know from the way that the they thought, you know from being you know just technologists to really making money. you know Now, they’re riding this rocket ship you know that we’re going to be talking you know with Richard about. But again, without further ado, let’s welcome our guest today, Richard Shocker. Welcome to the show.

Richard Socher: Hi there, great to be here. How are you doing?

Alejandro Cremades: Very well. So i’m very excited to have this conversation with you and they obviously, you know, quite the upbringing that you had born in Germany and also raised, you know, between Germany and Ethiopia. So give us a walk through memory e lane. How was life growing up for you?

Richard Socher: Life is pretty good. you know i ah like my My parents, especially my dad, wanted to get out of East Germany, and Ethiopia was one of those places that made that possible as part of the East Bloc. And so lived there for a couple of years, and then and did my undergrad and leipzak in Montpellier in southern France as an Erasmus exchange student back then, and then the Max Lang Institute for my master’s, and then looked at where the best researchers are in AI and who have the most citations and most impact on the field of AI and most of those were in the US. So I moved eventually to Stanford and did my PhD there and

Richard Socher: I was very fortunate to be at the right time to hear and see people talk about neural networks and keep learning for computer vision and speech recognition. And I thought, wouldn’t it be amazing if we can use those same ideas of just giving the AI raw data and then having it learn the features and apply that and use those kinds of ideas and adjust them, modify them, and make them work for natural language processing.

Richard Socher: And at the time,

Alejandro Cremades: you say quiet quite early because i mean when you thought about AI and getting into it, it was more like around 2009 before you know people were really talking about it. so That’s the time where you went to Stanford. so What really got you to start you know getting excited about the world of AI and what was possible with it?

Richard Socher: Yeah, I mean, I started even before that in 2003, I studied Linguistic Computer Science, which is essentially a different word for natural language processing, which is one of the most exciting areas of AI. ah Certainly the most interesting manifestation of human intelligence. And if we understand intelligence better, it helps us understand ourselves better. If we can recreate a capability, ah then we understand that capability better. And so it’s just been a fascination of mine since ah long for a long time. um Even in high school I enjoyed math and languages and they kind of marry also when you try to get computers to use math to understand languages. ah And then in 2010 there’s kind of this idea trying to go from first principles. You know a lot of the people at the time were designing specific features to try to solve one task really really well.

Richard Socher: And they became, you know, they talked to a lot of linguists and they tried to get really, really good at understanding sentiment or translation and different kinds of things. And I thought, well, wouldn’t it be more amazing if you just give it raw data and have the eye figure out all of those steps? So it was a little bit like first principles thinking.

Richard Socher: And then a lot of initial pushback from the community, but a handful of people did enjoy that line of of research and were supportive. Of course, my two advisors, Chris Manning and Andrew Ng, but also, you know, Yoshua Bengio, who’s the first to invite me to come give a talk somewhere else. And we won a Turing Award a few years ago for that deep learning work as well. So a little bit of a mix of being stubborn and first principles thinking.

Alejandro Cremades: So eventually you um graduated, ah you become a professor, and then also you were doing research um you know there at Stanford all the way up until 2018. But that was the time where you thought maybe it made sense to ah do something of your own. So what was that thought process like?

Richard Socher: Yeah, you know, in 2014, when I graduated, I saw most of my smart friends go to two or three companies, mostly they went to Google, some of them went to Facebook, some of them went to Amazon, but the majority of them went to Google. And I thought, well, it’s kind of unfortunate that this amazing technology is stuck in the hands of a few very large companies when really every company could benefit from this. And so I started MetaMind with the idea of making it very, very easily accessible to create your own neural networks

Richard Socher: and and do whatever you want to do with them. You can drag and drop images into the browser and create you know three lines of Python code to create your own an image classifier. We did natural language processing and we did some research even in the startup. And so it went really well, ah but we also realized, wow, this technology in the hands of a company that had amazing distribution, and no pun intended, an incredible sales force, ah would be even more impactful. And so after less than two years, we got acquired by Salesforce, where we became chief scientists, and then tried started to merge a lot of the different machine learning and AI efforts within the company, and eventually we became EDP.

Richard Socher: and did a lot of research. We started a whole research group there, but also ah actually shipped product, which was very motivating fun.

Alejandro Cremades: I mean, first company, first exit is is quite amazing, quite remarkable, especially when you’re doing an exit to a company like Salesforce. I mean, that’s saying as spectacular. I mean, how was that journey too as well? To be able to see the full cycle of a company, you know, when you’re going from creating it, building it, scaling it to all of a sudden, you know, you’re reaching the finish line and especially doing it with with a company like Salesforce, that’s incredible.

Richard Socher: It was it was really, really amazing and very meaningful. um It was also quite helpful that in 2016 I think Salesforce stock was at around 75 and now it’s at, you know, 300 and of course still.

Richard Socher: a holder. um And so i I have to say, you know, it’s interesting, a lot of but of other startup founders often came to me in after and asked actually, how bad is it? I’m like, what are you talking about? It’s amazing. They’re like, oh, you know, it’s big company politics. And I’m like, I felt like and In Salesforce, everyone was just trying to do the right thing. And yeah, sometimes people disagree on what the right thing is, right? So you have to argue for it and you have to talk about why you think one way of works better than the other. But everyone had their heart in the right place and and tried to do and build amazing products for our customers. So I actually really enjoyed my time. um And of course, you do have to spend some time understanding the organization, understanding how to make things happen.

Richard Socher: um But it was a really interesting and exciting challenge. I felt it was very meaningful because we really pushed the research world forward. you know We invented prompt engineering while at Salesforce. um We built the first large language model and actually synthesized proteins from those AI-generated biologic proteins. um and And they work, they fold it and everything. That line of research still hasn’t really gotten the limelight, I think it deserves. Those several startups have already started from that work, including the first author of the paper and and various others. And so it was a very meaningful time, shipping real products, seeing impacts on thousands of different companies, as well as really pushing the technology and the research forward.

Alejandro Cremades: What was it like to to experience a big organization like Salesforce? What kind of perspective do you think that gave you? Especially, you know, having been, you know, everything that you knew, you know, really being a founder of something that started from nothing where you were building the pipes, all of a sudden you were at a big company where the pipes were already very well, you know, designed and optimized and and and being able to see it more from like a larger ah enterprise type of type of approach.

Richard Socher: Yeah, i learned I learned so much. um Where do I start? I guess a big one is just the incredible ah sales motion also that that Salesforce has, but also the values that Marc Benioff instilled in the organization, ah making sure that you knows stakeholder capitalism, and at the same time, you know improving the shareholders’ values also and and showing that alignment that’s possible.

Richard Socher: um i I learned a ton from Mark just in terms of you know a good mix of listening and decision making and collecting data um and and from time to time saying like this is this is how we’re going to do it. um I have to be honest also that you know after 10 years of academia ah sales didn’t come naturally to me right and so that was like something that I’ve learned a ton about and that we’re now using at u.com also and you know trying to hire for your weaknesses, becoming a better manager. Salesforce was very good to me also in terms of just training and having mentors and coaches and things like that ah to keep improving. And I think if you have that growth mindset, ah being some amount of time in a large company can really help you as a founder also.

Alejandro Cremades: So once an entrepreneur, entrepreneur always an entrepreneur. and So after you know about four years and a half in Salesforce, you know you decided it was time to put the notice in. So what triggered that?

Richard Socher: You know, we invented prompt engineering in 2018. And what that meant is we could train a single neural network for all of the different and hardest tasks of natural language processing. You just trigger that one neural network in natural language to solve any kind of task. What’s the sentiment? What’s the translation? What is the summary of this paragraph? ah Who is the president? You know, all of these different questions. um And I thought,

Richard Socher: Well, clearly, if we have this one powerful neural net, the world of search and finding answers online, you should certainly change. And then fast forward two years to 2020, and like the world hadn’t changed at all. Google was still getting worse and worse, just more and more ads on top of fewer and fewer relevant organic links. They figured out that you know when you’re a monopoly and you just need to increase ah shares and revenue every quarter, it’s very easy to just go from three ads on top of the search results to four ads on top of the search results. And people would just stick with it because they didn’t know that there are any alternatives. And so it felt like now was the time the technology was right. We can use neural networks to massively improve how people find information online. The problem wasn’t anymore to make them accessible. ah The problem was to deal with that influx of massive

Richard Socher: of information. And so the the idea was to basically summarize search results instead of giving you a list of the links. And then we’ve actually transitioned more and more from a search engine via an answer engine to what we now realize is really the killer app, which is to be a productivity engine, help people be more productive when it comes to knowledge work.

Alejandro Cremades: So now let’s talk about what you guys are up to with you. And then just for the people that are that are listening to really get it, what ended up being the business model you know and how you guys make money? Because I know that that was a pivotal moment too of figuring out how to monetize. So walk us through that.

Richard Socher: Yeah, so high level, u dot.com basically combines the best of what chat should be key and Google offer to give you the most accurate answers with references in order to make AI agents and knowledge workers more productive.

Richard Socher: What does that mean? If you ask a question, you get an answer. Now that answer has real citations. A lot of our competitors fake ah that citation problem and half of their links or citations or references or just random numbered links sprinkled into an LM text, which you can’t really trust. So there’s a lot of hallucination.

Richard Socher: ah And it turns out some users are okay with hallucinations right if your life like it doesn’t really matter that much what answers you get you can use a lot of different tools and if they’re 70% accurate and you have a lot of time on your hands you can kind of deal with it and double click into it and search it again on Google and verify it and whatnot now we figured out like that the organizations and the people that really care about accuracy are knowledge workers. So we have one of the largest hedge funds in the United States as a customer of v.com. We have one of the largest European insurance companies as a customer, largest press agency of Germany that has thousands of different TV and newspaper stations as a customer. um As a nerd, I’m really excited ah to have the Institute of Advanced Study in Princeton

Richard Socher: be a customer, and what they pay for are two things, either APIs, where they get answers and they confuse them into their own products, or just a subscription and enterprise site license, ah where we combine their company internal or organization on internal data, as well as company the external data on the web to give them extremely accurate answers. With actually correct citation, you can click on and then scroll directly down to where it found its facts. No one else does that.

Richard Socher: and you can even build agents where you can have it automate an entire workflow for you. And that’s sort of what helps people unlock their productivity with AI.

Alejandro Cremades: So how much capital have you guys raised to date and what has been the journey of raising the money?

Richard Socher: We have raised ah a three rounds um and for a total of just shy of $99 million dollars so far.

Alejandro Cremades: And what has been that journey of going from seed stage to series A to, you know, the emotions too?

Richard Socher: The motions, yeah. So it’s been a really exciting time. In many ways, my biggest lesson is that when you’re ahead in terms of technology, that’s just one piece of the puzzle. It’s not really what drives um revenue as much and so a big lesson for me is that ah best technology is only useful if you also have really strong marketing. We’ve had companies that are massively inferior in terms of technology burn through millions of dollars in marketing and it’s kind of working right and they’re like semi-lying and I’m like ah I’m as an academic and a German I don’t really want to overhype things so much and claim we got 60 million new users when it’s really a few thousand and some

Richard Socher: partnership deal ah but I have to acknowledge that you have to kind of override things a little bit sometimes. I’m trying to adjust but I don’t really want to lose my my factual sort of correctness and so we’re we’re hiring better and better marketing ah folks into the team and yeah that was that was a big lesson. Technology is amazing.

Richard Socher: but you gotta distribute and you gotta actually sell it. And so we’re now, this this year we’ve transitioned kind of from just like, oh, look, we have the most amazing technology to like, how do we help a customer really become more productive with this? And that’s how we’ve closed all these deals I mentioned and how we are now an exponential curve when it comes to revenue. We doubled revenue every quarter this year. This quarter looks like we’re more than triple it again.

Richard Socher: And so with that like literal hockey state curve, i’m really excited to look into the next year.

Alejandro Cremades: So you were talking about a growth. You were talking about marketing before. I mean, I got to ask you, how the hell did you guys get the domain u.com?

Richard Socher: That was one of the many amazing inputs from Mark Benioff. He had actually bought that domain over 20 years ago. And um and you know I explained to him what I wanted to do with the new company and he’s very gracious. He became our largest and lead investor and board member for the Series C. And as part of that also gave the domain.

Alejandro Cremades: That’s incredible. So obviously when we’re thinking about investors, you know, they’re betting on a vision, right? Because I’m sure that Marc Benioff and the others were like excited about the future that you were living into. So with that being said, if you were to go to sleep tonight and you wake up in a world where the vision of you is fully realized, what does that world look like?

Richard Socher: That world will have a lot more knowledge and a lot more productive people ah in in on the planet. um People will have access to unbiased information that they can verify quickly and that makes them in insanely more productive. um And so to help overall organizations massively excel.

Richard Socher: And you know one one big epiphany we’ve had, just as an anecdote, was we had a company ah called Minecast. And they bought um like about 200 seed licenses. And we thought, amazing. This is an incredible customer. So we decided to do a workshop with them. ah They’re a cybersecurity company. And we said in the workshop, hey, you like feel free to bring other people, not just the engineering team and the initial teams, but bring everyone.

Richard Socher: sales, service, HR, recruiting, analysts, engineers, service folks, like everyone. And then they bought many more ah licenses after that workshop because they realized that, wow, this is really foundational technology. And at this point, I’m almost the best way I can describe it is if your organization could benefit from moving from paper to a computer,

Richard Socher: your organization can likely benefit from using a computer with u.com that’s how broad it is and that’s hard to explain to people without going through very specific examples so let me give you a concrete example let’s say you’re in marketing and one of your workflows is that every couple of weeks you get a big pdf file with a bunch of interesting new features that your product and then design and sales and so on concrete built, and you say, all right, please market these new features to your industry. So then you take that PDF, you look at the new features, you go on the web, you compare them to your competitors, you write two email marketing campaigns for two different industries, and you set out to send like five to 10 tweets and LinkedIn messages. You basically can just describe that entire workflow that you do.

Richard Socher: to an agent on u.com and then next week when a PDF comes in you just drag and drop it into your browser and it’ll just go through all of these steps and do all of those workflows for you fully automatically and you’re done and so we’ve had companies and employees tell us like holy shit this saves me like 10 hours, 20 hours a week of what I used to have to do manually. And so as we extrapolate that, I think the more you care about complex knowledge work, the more you get that 10X experience from switching from Google to U.com.

Richard Socher: ah And so in that world, in the future, I think we can become a multi-trillion dollar company, right? When you think about all the productivity, all the answers and inflamm information that you can get on u.com, it’s it’s just an incredibly exciting future for knowledge, for learning, and for productivity.

Alejandro Cremades: So in parallel, you’ve also been investing, you know, AIX Ventures, first fund 50 million, second fund 200 million. So ah do you how do you how are you able to really, you know, do both in parallel? And I’m sure that there’s been a lot that you’ve learned from the from wearing the investor hat, and seeing patterns, seeing what works, what doesn’t, you know what some founders that are succeeding are doing versus the ones that are failing. ah How do I identify you know also ah potential good hits and and good returners? So what when it comes to patterns, what have you learned?

Richard Socher: Yeah, you’re absolutely right. There’s a ton of synergy between these two. um you know It startedi started because after the acquisition, I mostly kept my grad student lifestyle, stayed in a pretty small apartment, didn’t buy you know a bunch of land bows and whatnot.

Richard Socher: um and instead I just invested in all my smartest friends and interns and employees and students and I love AI and so i you know AI is omnius technology and everyone is exploring different ideas and it turns out I was very fortunate to have an incredibly smart group of people around me and so as I just wanted to hang out with them and we talk over dinners and so on, I was like, hey, can I just put some money into your like company? Or they asked, like hey, do you want to invest in this startup that I’m thinking about? And so that worked out really well. My own angel portfolio has over 17x since 2016. And so I thought, well, how do I scale that? And my time doesn’t scale. I really want to build still.

Richard Socher: But you just partner with other people, and there’s you know a lot of synergies. And so we brought in other investing partners, what we call them, who are full-time faculty at Stanford and Berkeley, full-time founders like the Anthony Goldblum, the founder and CEO of Kaggle, who sold it to Google, now started a new company called Samble. And we we all have just naturally deal flow by being part of the AI community, in many cases for over decade or two decades. And so ah that means also that we’re just getting that deal flow, but at some point it becomes unwieldy and invest in a bunch of different companies. And I didn’t want to spend a lot of time on paperwork and some basic diligence things. And um and so we basically now in the fund combine a headquarters scheme that does all the traditional great work with a team that actually stays at that cutting edge of the eye, the investing partners like myself.

Richard Socher: And we think that’s really important and the feel like AI that moves so quickly. Imagine you’re an expert in AI five years ago. It wouldn’t matter that much anymore what you did if you stopped being an expert and started just investing. I think it’d be very hard for you to understand what the cutting edge ah possibilities are in AI. And so that’s that’s basically the foundation of AI expenditures. Combining investing partners with practitioners of AI with a strong headquarters team ah to get a new kind of VC model going and you know honestly it’s sometimes

Richard Socher: feels like a lot of the VC um work is barely, really feels like work. So what I mean by that, you know I sometimes have events here at my ranch, I bring a bunch of friends in, we discuss the future and I love these little unconference styles where we just have people discuss the future and think about you know where things are going and and you debate it and and you have little you know sessions.

Richard Socher: And then it turns out and that’s my friends and some of their friends come, a lot of them wanted start companies, and then you just had this event and like, that doesn’t feel like work to me and if I go to dinner and I talked to a really smart ambitious founder who wants to change the world and build incredible new technology or just invest in this amazing company called parallel bio they built.

Richard Socher: humanoid lymph nodes uh human lymph node organoids uh that are essentially tiny little lymph nodes that in a petri dish and that allows you to test uh drugs much much more quickly uh that company at scale will has a potentially saved hundreds of millions of animal lives let that sink in one company could save hundreds of millions of animal lives that are just bred with the worst kinds of genes to get the kinds of cancers that you then want to treat and learn how to treat. It’s a terrible life for those animals. And so you can do so much good in this talking to founders that have those kinds of visions and want to make people more productive, learn things, help improve health care, improve biotechnology. And all of this is just fun. And I enjoy spending my evenings and have dinners with folks like that. And on top of that, you just have to work a whole lot.

Alejandro Cremades: So imagine I put you into a time machine and I bring you back in time to maybe that moment where you were thinking about starting MetaMind. And let’s say you know you had the opportunity of having a chat with your younger self and being able to give your younger self one piece of advice for launching a business. What would that be and why given what you know now?

Richard Socher: I think the biggest thing probably is just to be even more constructively optimistic. I kind of coined this term constructive optimism for myself. And, you know, even though I love the eye and I pushed the eye forward, um and I had a lot of these foundational pieces and foundational ideas that led to chat LGBT, I didn’t scale the models that I’ve had like invented and co-invented with my co-authors and also my co-founder Brian McCann, our CTO here at u dot.com. We could have scaled them even further and just tried to push harder and harder to get funding for 30, 50, $100 million dollars to train a single model. And so I think that would be the biggest advice is go bigger with the ideas that you’ve had.

Richard Socher: train even bigger word vectors, contextual vectors, and prompt engineering single neural networks that can solve every task in NLP, including, for instance, language modeling, which was literally on our to-do list for this prompt engineering paper as another task to add.

Richard Socher: And, you know, ultimately, you know, our papers as an academic it’s great because they got cited they inspired other folks like those that open the eye to train a single model and just try to prompt that model with different question, but we could have gone even bigger, even earlier, I think that would be my main advice to manage.

Alejandro Cremades: I love it. So, Richard, for the people that are listening that would love to reach out and say hi, what is the best way for them to do so?

Richard Socher: Send me an email at Richard.U.com or follow me on Twitter or LinkedIn. Try to know answer all reasonable requests.

Alejandro Cremades: Amazing. Well, hey, well Richard, thank you so much for being on the Deal Maker Show today. It has been an absolute honor to have you with us.

Richard Socher: Thanks for having me and thanks for listening, everyone.

*****

If you like the show, make sure that you hit that subscribe button. If you can leave a review as well, that would be fantastic. And if you got any value either from this episode or from the show itself, share it with a friend. Perhaps they will also appreciate it. Also, remember, if you need any help, whether it is with your fundraising efforts or with selling your business, you can reach me at al*******@pa**************.com“>al*******@pa**************.com

 

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Google Podcasts | Spotify | Stitcher | TuneIn | RSS | More