Skip to main content
Loading Explore guides…
AI for Biology & CRISPR · Acedra
Home
Explore
Biology
AI for Biology & CRISPR
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
67 segments
Copy
Hide video
0:00
What if we can make biology almost more like computer science where you can >> run, experiment, and, the, next, day, you, know the results. >> Professor, Leong, pioneered, the, use, of crisper gene editing in human cells proving this revolutionary tool could work in living systems. But his next breakthrough might be even bigger. >> Instead, of, having, one, me, working, in, my lab, wouldn't it be nice? We can have 200 AI scientist in your team that only cost you hopefully just GPU compute. But
0:29
getting there meant abandoning most things he used to work on. >> I, really, moved, completely, from, you, know doing experiments in the lab with my hands to typing on the keyboard. Sometimes it's really I learned it from my own mistakes the hard way. >> The, lesson, that, changed, his, approach, to innovation forever. >> Failure, is, great, but, most, of, the, time it's great if you fail fast. If this thing takes you less than 3 minute and I think it's interesting
0:55
do it right now. Don't wait. Now he's building something that could accelerate every medical breakthrough by decades. >> What, if, I, can, model, the, whole, cell, or genome or disease purely in computational way? That makes the whole process almost eventually as quick as writing a code. We have already built a large community with over 20,000 people. Everybody come together, build a community and then to talk about how do you really improve AI for science. The
1:25
race to cure cancer, Alzheimer's, and aging itself just got exponentially faster. Today, my guest is Le Tong, an assistant professor at Stanford University School of Medicine. Professor Tong was one of the co-inventors who proved crisper gene editing technology could work in humans. After gaining industry experience as VP at Mona during the pandemic, he returned to Stanford where he's developing AI agents that can design experiments analyze results, and even replicate the
1:54
capabilities of world-class researchers. He's now building something even more ambitious, AI systems that can automate scientific discovery itself. He's also built a 20,000 plus member global community that's becoming the epicenter of AIdriven scientific breakthroughs. In part one, Professor Tong revealed the untold story behind Crisper's breakthrough. In part two, we covered his remarkable journey from hands-on crisper research to leading the AI for science revolution. We dive into his
2:20
three-level framework for AI in scientific discovery, from AI co-pilots to autonomous AI scientists to complete virtual research environments. Professor Tong shares his vision for multimodal biological models that can understand DNA, RNA, proteins, and clinical data simultaneously. We explore why he left pure research to gain industry experience at Monoa, his work building specialized AI agents like crisper GPT and stem cell GPT and how he's created one of the largest scientific AI
2:49
communities in the world. Let us get started. I would love to also dive into your life uh after the crisper. Uh so you did this great work of crisper in 2022. you started take a job of um VP of um Mona starting doing more industry experiences um tell us more >> yeah, like, I, think, maybe, I'll, share, two you know you know stage of my sort of career after uh the Christopher moment we can call it >> so you know in many ways right like we we're talking about Christopher is a
3:27
Caesar but what actually are the targets for them like yes I fix many things but what are the genes that can help me to cure cancer cure diabetes cure Alzheimer's right >> so, that's, why, I, moved, into, computation AI because I think we needed ability >> to, really >> scale, the >> or, understanding, of, the, data, and, biology >> in, order, for, us, to, identify, okay, this, is a way you can add the genome to cure disease and to prevent aging right so
3:59
>> so, you, first, invented, Caesar, and, now, you are looking for like you are now starting to solve the search problem. >> Yeah,, search, problem., So, I, think >> well, that's, why, I, worked, with, a, viv, gav >> who, you, know, basically, was, my, PhD advisor and she was a computer biologist >> and, I, I, really, moved, completely, from, you know doing experiments in the lab with my hands to typing on the keyboard to you know try to write write codes and to
4:25
do you know computational biology and machine learning. So you started connecting with your childhood dream of computer design. >> Yeah., But, which, is, you, know, not necessarily, the, exact, same thing, as programming a game but and then with a viv uh I really start to really appreciate the power of cotation. No the iteration just faster right you you write a program the same day or most the next day you will know the results of computation. of course uh you know so
4:50
the the the whole cycle of feedback iteration >> is, way, faster, in, computational, biology so that has been really fun and that's why you know I later started to focus more on AI AI for bio medicine AI for science and basically that's how I like you know after working at the viv when I moved to Stanford here in be in the silicon valley you know San Francisco I I reallyed start to build a group where we do experiments on one hand and then do commutation AI on the
5:22
other hand and then so that's one thing and then the two why I worked you know I spent some time with MADNA is because uh you know when during the pandemic there were a lot of you know restrictions >> and, of, course, I, really, respect, what, you know MRA technology was doing >> but, then, I, think, the, most, motivation, is not about particularly like MRA or something it's really about getting to understand the industrial scale of things >> because >> my, actually even my own advisor Aviv
5:55
>> before, I, started, a kind, of a, leave, of, sabatical, at, Madna >> Aviv, first, actually, moved, from, MIT >> to, become, the, head, of, pre-clinical research >> at, Ro, genetic >> uh, so, that, inspired, me, to, think, about >> if, you, wanted, to, understand, what actually it takes to make a And to be able to scale a technology from lab to business to really a life-saving medicine, I think I need to exposure. I need to get more exposure to industry.
6:29
>> You, have, been, really, passionate, about how to turn some idea or scientific ideas into something that is really practical and can be scaled that be used in the industry. And you also briefly uh mentioned about this AI for science initiative uh that you have been focusing in the past decades. Um can you tell us a little bit about uh with us about what what does it mean by AI for science? >> Yes., Uh, yeah, thank, you., Yeah,, basically as you you mentioned um one thing that
7:04
I've becoming more passionate about in the past couple years has been essentially trying to accelerate the whole process >> of, discovery. >> Uh, like, you, know, what, if, we, can, say, make biology almost more like computer science where you can >> run, experiment, and, the, next, day, you, know the results right or something like that. So I think AI present probably the most reasonable opportunity for us to accelerate science because like you said AI can write a
7:39
>> you, know, can, write, a, email, for, us >> can, write, a, paper, can, write, a, poem, but also can generate image >> right >> can, generate, a, movie, so, I, think >> AI, really, accelerate, a, lot, of, the process right now in text >> in, image >> so, what, we, think, are, going, most, helpful for for us right as a scientist >> is, if, AI, not, just, help, me, write, an, email but actually help me >> to, be, more, efficient more amazing at doing science by applying the AI technology and the
8:12
concept what you just mentioned scale or scaling law we can also be basically achieve three things in science number one >> we, can, have, AI >> essentially on first level >> be, a, co-pilot >> or, be, a, partner, for, all, the, scientists where if you need to find information to design experiment to run something or to run data analysis >> AI, will, help, you, do, that, and, the, second level >> uh, is, called, AI, scientist, where, like basically >> I, can, replicate, myself
8:48
>> like, I, can, build, an, AI, that, basically, is at least as good as myself or fun can replic scientist as good as fun or Jennifer D. So that would be fine. So basically like instead of having say >> one, you, know, me, working, in, my, lab wouldn't be nice or maybe not necessarily just me but maybe all my students like we can have >> a, lot, of, the, students >> uh, and, all, you, need, is, not, to, recruit 200 people >> to, your, team. >> Yeah., You, just, need, 200, AI, scientists, in
9:19
your team that only cost you hopefully just GPU compute and then you can have all of them work together >> to, not, just, so, it's, not, just, AI, helping one person but it's AI replicating the capacity of scientists so that you can have very quickly a team of people working on crisper working on Alzheimer's aging stem cells right this is the second and the last level I think is what we can call virtual >> research >> or, we, can, call, you, know, basically, like
9:48
like you said you know like you you know you can generate a video that's completely by AI you don't need to actually shot in a camera right >> so, what, if >> eventually, I, think, the, last, level, of, AI for science is like what if you can say I can model the whole >> you, know, a, cell, or, a, genome, or, a, disease purely in computational way so then >> when, I, do, research, it's, not, even, just, I have a team of AI scientists But also like I just essenti simulate you know a
10:19
lot of the disease in working in silicico right >> and, that, really, becomes, exciting, to, me because that makes the whole process almost eventually as quick as writing a code or testing a code like you can build >> there's, actually, a, lot, of, uh, similarity between the knowledge discovery process in biology but also a lot of the things that are happening other AI uh AI industry Right. >> Um, so, the, pattern, that, you, put, is, such, a beautiful summary of the three stages of
10:52
AI evolution. uh you put in the terminology with just biology but I also observe these patterns for all the industries and it's basically kind of evolution of AI for automating the launch discovery process and also applicable for many launch workers >> like, if, you, are, managers, if, coders, also works very similarly and these three phrases marks the different levels of agency >> yeah, it's, almost, like, you, know, the, L1, 2 3 4 of auto driving right
11:23
>> exactly, The, the, lowest, one, has, NAS agency so needs more human involvement. The highest level is completely the agency. So it's just discovery logic automatically by itself. >> Exactly. >> Yeah., So, that, is, a, very, cool, um, cool mission for for you um to have um AI to enable the science or the scientific knowledge discovery processes. >> Right. >> Yeah., Can, you, walk, walk, through, us, some of exciting projects you you did um in that direction?
11:52
>> Yes., So, specifically, we, have, uh, you, know in the past year we have released I think three different set of progress that I think can all help us to get to this maybe we can call it you know level one two three of AI for science >> so, the, three, thing, to, give, you, a, quick overview is one models right we've been building >> uh, large, RNA, models, uh, for, example, one of the model we are leading to build is >> uh, RNA, Genesis >> uh, which, is, basically, one, of, the, most
12:25
you know capable RA model that can model different type of RNA. >> For example, like, for example, if, you wanted to design a small RA drug MRA drug you can use our model to give you a quick AIdriven design that hopefully will work very well. >> Yeah. >> So, so, we, can, call, it's, almost, like, the same thing like you have chat GBT or uh you know cloud right for language generation. We also want need models for biological generation or generative AI in biology. So the first thing we've
12:54
been building is this models we kind of foundation models in biology and because there are a lot of people already working on proteins and DNA so we have more focus on RNA but I think eventually there will be a suite of models >> and, that's, number, one, models., Second, is called agents. >> Uh, that, is, the, next, sort, of, component right when you models can give you some simple output but like you you know if you want to do coding or web browsing you need a agent that is driven by you
13:28
know cloud or GBT to do the task. So I think in bi biology like you were saying >> all, the, agents, are, going, to, be, more capable to complete >> you, know, human, task, or, help, human, with completeness. So the first agent we built is called crisper GPT. >> Yeah. >> Which, is, a, specialized, agent, that, can automate design execution and analysis of genetic experiment crisper genetic experiment. >> Right., Right. >> And, so, that's, number, one., And, we, are, now
13:56
building also stem cell GBT for stem cells. >> Right., and, also, like, a, aging, agent, for aging study and then we'll have hopefully many more eventually maybe ecosystem for that. So that's second agent and the last thing I want to quickly want to mention we're also building benchmarks >> and, safety, you, know, sort, of, guard, wheels that are part of the requirement for AI for science right so when you think about how do I really know the AI scientist is actually working so we need
14:26
benchmarks data sets >> so, we, have, been, releasing, just, actually this week >> um, genome, bench, which, is, a >> large, bench, benchmark >> for, genomics, sort, of, valid, research >> and, basically, the, idea, is, that, we, can release a benchmark and we can see how different models different agents are performing on the bench and have everybody improve. So the genome bench is, one, of the, thing, we're, also, going, to release stem cell bench a lot of
14:54
arguments. So the final part of our work is establishing AI for science benchmarks and also also be able to also help people to understand the performance as of the safety of these you know different things. So, so that's >> those, thing, for, people, who, are, familiar with like AI research actually are the exact same terminology I described for uh AI research like foundation models you need agent and you need a benchmark >> yes >> and, I, also, realized
15:21
>> uh, that, a, lot, of, your, model, names, had GPT inside Christopher GPT storm cell GPT >> right >> it's, not, a, coincidence, and, for, for, our audiences like they may not have enough background to connect oh what is the relationship ship of the models to generate DNA or or the RNA with the model that generate the email. So what's the relationship between them? In fact that you put in early ways that they're all sequence and a lot of the AI what we call the large language models is
15:53
basically very powerful sequence processing machine and we just happen to apply this to our natural language sequence first. uh that is a pretty complicated sequence because the token size is big >> right >> but, our, DNA, on, RNA, is, also, sequence, in some sense is even simpler than the than the human language because there's like certain pattern there's not that many code book relative is smaller >> from, that, uh, so, a, lot, of, ideas, can, be can be shared among the two exactly
16:22
>> what, what, would, be, your, take, what, are areas that you're most excited about to push uh about like the the research in the in the biology side. Yeah. >> I, think, the, most, exciting, to, me, are, you know again you know basically focus on the three things we were just talking about on the model side I think is multimodality. So instead of just say oh there's a model for generating a protein or crisper RA >> can, we, have, a, sort, of, multimodal, it, was
16:54
like why you know there are what we call vision language models or you know you know these multimodality language models where you can take a image understand image and but also give you a a you know test description so they understand both language >> and, also, you, know, vision, imi, images, and videos right >> I, think in, biology, we, also, need, that, you know understanding for example the DNA RNA protein sequences and understanding all the for example the biology physiology or
17:23
clinical data right in patients I think this model modality is super exciting to us and we're doing some of the work and >> yeah, I, want, to, paraphrase, just, make, sure my understanding is correct and also help our audience to understand on this when you mention about a multimodality you essentially referring the data sources or sequence from diff that has different nature >> yeah, sometimes, not, even, sequence, also like uh you know anything that can tokenized like you said images you know
17:50
maybe electronic recordings or heart you know >> you, know, or, or, heart, you, know, you, know you know sort of you know we can call like uh cardio sort of measurements through like you know different type of you know device even like or personal health data like >> uh, in, the, form, of, like, your, you, know, you know brainwave data you know your like you know Apple health data. So a lot of these personal health or imaging data X-ray, you know, older stuff or even
18:21
sometimes the you know various forms of you know molecular data is all going to be >> capable, being, connected, through, this multiple >> you, you, actually, mentioned, such, an interesting point because previously there was many data in biology. There's DNA data which is has it's independent research area >> right? >> There's, protein, data, and, there's independent research. >> No,, it's, all, connected., Yeah, >> it's, all, connected., Then, there's cleaning trial datas or there's even
18:51
nowadays the medical note takingaking systems. >> There, are, a, lot, of, things, we, can, collect >> independent, before >> and, now, we, can, we, can, merge, them. >> You, want, to, merge, them., Yes., And, I, also realized you pub recently published a paper that even like you can infer the you basically doing a lot of analysis of this clinical data even like discussion from the medical forum. Yes. >> Is, that, align, with, this? >> Yeah., This, is, Yeah,, exactly., I, think, uh
19:16
but that's more on the agent side where like maybe very quickly going through the other two areas I'm super excited about wines like we're trying to build ecosystems for agents where you can have you know hundreds of tools and databases in a almost like a operating system for AI for science or AI for biome agents right essentially you build an operating system like iOS or Mac OS or Windows and then you can allow these agents to explore for different type of data and
19:46
this is I think going to be what we can call the agent you know moment for biio medicine where like you can now have the agent explore the biomedical database >> and, then, maybe, learn, from, by by themselves also like we can call autonomous evolving agent I think this is the, second, part, and, the, last, part like you said why we are getting some really interesting work done in the benchmarks and like extracting information from online discuss discussion is that we think once you
20:16
have a lot of the good models and agents, you need really really scientific benchmark >> to, help, everybody, contribute, and evaluate and this is where we think we need a community together. So that's why we're for the benchmark and you know and the safety side of things we're engaging not just a small amount of scientists where we have already built a large community with you know over like 20,000 people on a slack group where we wanted to have everybody come together build a
20:47
community and then to talk about what do you want to see in a benchmark how do you contribute to the data and how do you really improve AI for science that's kind of the three things that I think super exciting >> those, are, very, very, cool, direction, to push forward. So what I hear you say is that in this three area layout each side you actually have one area that most excited about. Yes. >> For, the, model, layer, you're, very, excited about this multimodel capability which
21:13
is merging the data from different sources. Yes. Uh connect those together. In the agent side you want to create this platform so that other people can build more useful agents and inject the knowledge into the platform. And also the last part in the benchmark side you realize that it actually more and more uh critical to ensure that we're moving into right direction which needs more sophisticated evaluation. >> Yes., requires, the, benchmark, and, you, also mentioned about the community because
21:44
that's also um a thing in like uh um that even in the nonbiology AI side people are talking that a lot by the way like everything talk about sound music to me because that's also exactly what AI research is pushing forward in the normal AI field. Uh and specifically I want to dive into the benchmark side. You mentioned about you build this AI for science community which already have 20,000 me members into that is incredible. Uh can you walk through us like why did you do that and how does
22:17
that relates to other efforts like the benchmark efforts and other efforts you have? >> Yeah,, I, think, this, really, connects, to like you said we need better ways in at least for AI for science. I think the data and the benchmark are limitation because like a lot of public information is more about like oh can you uh be smart at exams like GPT sorry GRE or uh you know all these exams or or you know but exam taking is not scientific discovery >> right
22:47
>> scientific, discovery, is, about, learning new stuff >> uh, you, know, discovering, new, ideas, never happened before >> so, it's, very, hard, for, one, lab, to, or, even a small group to be able to cover the exciting frontiers of science. >> Exactly. >> And, and, then, you, actually, need, something really challenging that's not just patent matching what's already on the internet. Sometimes these are needs to come from private data, private domain. So that's why we believe that we need a
23:16
bigger community and I just want to mention that this is not a personal effort. is really led by for example professor like Miwang you know even like young George Church fun a lot of people that I know are all coming together so this is a big effort >> by, many, leaders, and, I'm, really, just, one part of it the idea is that >> it, needs, to, come, from, scientists, doing the science because >> it's, not, one, group, has, all, the, expertise >> exactly
23:42
>> so, we, needed, to, have, a, global, network people so we are building this community that come together to do three things one we can pull our data expertise together to build really good benchmarks and two >> we're, organizing, a, conference, the, first one will be October this year >> uh, you, know, actually, at, Princeton tentatively where we can bring people together to talk about what's where we are in AI for science how can we make >> AI, more, intelligent, have, higher
24:11
reasoning capacity and what are the things we need to do like benchmarks evaluations that so I think the second is conference and the last thing is I With all these coming together, we can also essentially build some consensus uh on how we can move forward. it can be you know we hope that you know we can come from these community and and conferences some cool ideas that maybe not necessarily just you know in the form of research can be I don't know maybe a new company new new company
24:43
spin-off from the the discussions >> it, can, be, a, new >> collaborative, you, know, scientific reasoning model >> that, really, is, a, global, collaboration and make everyone do science better. I think there are many possibility like you said in the beginning we'll bring a lot of cool people together. >> You, can, have, magic, happen., So, that's what we hope. >> Yeah. >> Exactly. >> So, I, don't, know, what, will, happen, but, we hope >> uh, that, this, community, can, you, know
25:11
really be uh you know a centerpiece of what can come for AI for science. >> Exactly., I, think, that, many, people, have been talking about when we have the AI how to ensure that we human as uh species still is safe if we responsibly develop this new technology I think is as powerful as a nuclear weapon >> and, how, do, we, how, do, we, make, sure, that is you know developing responsible way I think the only way is cooperations and like you mentioned international
25:39
corporations with scientists from all over the places >> right >> and, all, contributing, on, this, idea, is, and in that way we will connectively know what is good and what is bad. Yes. >> Yeah., So, you, have, done, a, lot, of, amazing work in crisper and AI for science and building community. How do you balance all this work and still leave the healthy life? >> Well, I, I, think, yeah, two, thoughts, of course. one uh sometimes it is very hard to have work like that right in academia
26:14
which, is, part, of the, reason, why sometimes it's nice to take a break you know at a company for some time I do think you know um the number one thing I want to share is definitely like one this is not necessarily happening all the at the same, time, like, for example, I, did, crisper as a student and then did computational biology as a postto and now of course in my lab we did more the AI for science AF for biomedical stuff. I think it it does you know take time. So we think
26:42
this a lot of work are connecting and also for example the reason we started AI agents with gene editing was because my earlier work and the accumulated knowledge and knowhow right from gene editing and stem cell research that helped me to build the AI agents for crisper and stem cell. So I think the nice thing is even though it's seem to be different work but I I felt like they're all connected >> and, we're, also, building, on, our, private work to to to really enable the AI
27:11
agents models and benchmark >> and, the, second, dot, is, like, I, do, think, a lot of work are really collaboration and uh community like we work on AI agent ecosystems with Yuri Lasha here at Stanford computer science and also Aviva Gaff my previous advisor but who's now at genetic. So I think uh on the model side we work with you know Mong Dwang uh you know Russ Altman a lot of the people in in you know here in the community and then you know with gene editing we also
27:39
talked to got great benefits from Funjang and then um security community we got support from George Church you know fun even Jennifer down a lot of people in the field so I think >> the, second, part, is, really, like >> you, know, ideally, we, can work together >> and, pull, the, efforts, and, that, makes everyone's time amplifying, right? It's like 1 plus one more than two. >> That's, also, the, power, of, community. >> But, I, do, think, like, yeah,, like, you, said, maybe it's easier to do
28:10
>> in, science, because, everybody,, you, know, is is not always looking for uh a single person's credit. It's more like if we can make things happen together, it's benefiting everyone. So, so I think it does help to work on AI for science to >> Exactly., And, and, I, also, feel, that throughout our conversation gives a lot of credit to your collaborators like the fellow researchers who are researching the similar problem as you do. Uh it it is actually spirit of a science
28:42
scientific spirit. It's like share the credit. We are on this together to push the frontier for human not for individual of us and that make the collaboration easy and make each like feel more empowered. >> Yes., Um, the, very, last, question, I, want, to ask you is that um what would be your advices for the younger generations who want to follow your path will take you as a role model uh to do the research push the front and empower other people and have the real industry impact. What
29:15
would be your advice to them? >> Thank, you., Um, maybe, I, can, share, um, two quick thoughts that I think really helped me in some way. But sometimes it's really I learned it from my own mistakes the hard way. Number one is it's always okay and sometimes maybe even really great that you'll fail or but only if you'll fail fast. So I think this is really important like failure is great >> but, most, of, the, time >> it's, great, if, you, fail, fast., So, do, not
29:51
spend like say you know a year doing something that you don't know if it's you know something you believe in. You you kind of wanted to iterate faster and, this, is, part, of the, reason, I, think we needed to have young people really embrace AI in science because AI kind of accelerates things and I think so the first thing I I want really want to share is that I think failure is really good. You learn so much, but you wanted to be able to fail fast like either
30:20
through, you know, either through using AI, either through just do things harder, right? Like you said, working hard >> is, not, always, useful,, but, you, kind, of want, to, at least, work, hard, sometimes, so that you can gain more from your your your work experience. I think that's number one. And then se second thing I want to share is like you said I think there's always the idea that you know timing is important for doing everything. >> Mhm.
30:52
>> But, to, to, me, I, think, the, key, thing, here is timing is also about doing things early. So my advice for a lot of my own students is that if you believe in something there's kind of this >> I, have, this, sort, of, three, minutes, rule right in my lab where like if this thing takes you less than three minutes I think it's interesting do it right now don't wait >> the, power, of, now >> the, power, of, now, like, so, if, you, think, oh this is a cool person >> I, just, met
31:28
>> in, like, you, know, in, a, in, in, a, sort, of scientific conference >> maybe, I, should, email, this, person., It just it takes less than three minutes to email someone. >> Exactly. >> Just, do, it, right, now. >> Yeah. >> Don't, wait. >> And, and, I, think, you, if, you, get, to, know someone early, you know some more people in the early stage of a career, it actually pays more dividend in the long run. So in some way you want to act faster you when you're young and that
31:55
give you in some way more likelihood of getting the right timing. So >> yeah, it's, also, like, the, maximize, the curio Warren Buffett's po point about >> if, you, have, like, you, know, five% cumulative return over time you will get be a you know billionaire or something right anyway. So I think that's the two thing I think really want to share is like right feel fast and then really be willing to act now. >> Act, now., Yeah., Exactly. >> Wow., What, a, beautiful, uh, advices., I
32:27
think all the our audiences will be benefit from me. >> Thank, you. >> It's, such, a, great, pleasure,, such, an honor to have a conversation with you. So insightful. Thank you. Thank you so much for sharing the community. Is there anything that the uh fun community can help with you? >> A, ton., Right., like, I, love, to, meet, with more people from the community. I wanted to learn from everyone especially given that you know uh you know I'm more from academic background. We're interested in
32:55
you know learning from you know industry startup leaders people who have done startup how do we >> we, would, also, have, you, to, join, one, of our event children community but we'll leave you a comment in the video below so they can reach out but we would love to have you join our event >> and, learn, from, everyone, who, are, you, know doing cool things >> that'll, be, fantastic, thank, you, thank, you so much looking forward to chat more >> thank, you
33:19
>> thanks, for, watching, the, full, interview if you got value from this conversation please Like, drop a comment, and subscribe to Founder Coho. We've got more founder insights and deep dive interviews coming your way, so don't miss them.
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat