Latest / Practice Ground / Tony Kipkemboi: "Stop Agentifying Everything" - What AI Builders Keep Getting Wrong | Ep 7
Transcript
- Nifemi Aluko: Tony Kipkenboy, thank you for joining Practice Ground podcast today. I'm very, very grateful for your time. And just to give a little intro by yourself, you're an AI engineering contents creator. On YouTube, you're known as the how to guy, where you teach technical content on building AI automation and agent systems. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah, thank you. Thank you for having me. Nifemi Aluko: You've worked at a variety of places, crew AI, I kind of worked with a lot when I was at deep learning AI. Snowflake, I think I saw Bloomberg somewhere. And you're now at Guild, which is very interesting what they're doing in terms of trying to upscale talent and close that talent gap. In general, you're passionate about open source AI agents and developer. education. I think what I really value in this space is when I see people that are, you know, taking technical, taking technical, will I say knowledge or technical things and try to translate it in a very simple, accessible way so that a lot of people can get in, but also holding that technical clarity because, you know, the hardcore engineers or hardcore technical people are are very ready to call you out if you're not, know, keeping it, keeping it 100. So I know it's a very challenging thing that you do daily. you know, in terms of like AI and agents, you've worked at one of the places that I felt was the place where, you know, building AI agents comes top of the mind to me, at least crew AI. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. Yeah. Nifemi Aluko: Right. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah. Nifemi Aluko: Building a crew. You know, I spent some time with Zhao himself, very passionate guy. And, you know, AI and agents, they just keep improving. Right. And we're seeing them improve a lot of stuff that can be done based on all like you've seen in the industry, what you've built. You've been around this for a long time, educating people. What is the most important thing to keep in mind when you're designing and building like an agentic? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah. Nifemi Aluko: system. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah, great, great, great, great, great first question to begin with. you meant you mentioned Joao there. Yeah, he's solid guy. Loved working for him. so really when I think about how to build agentic systems and how to think about that, mostly even with within the company right now at Guild and some of the companies that I spoke to previously at crew that were trying to build agents, I think the first thing that I always kind of emphasize is understand what you're trying to solve. Nifemi Aluko: Yep. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: like deeply. Cause right now there's a lot of tendency of we agentify that. We can agentify this. But I think that's just jumping to the to the tool without having figured out the problem set very well. Cause sometimes most of this problems don't even need agentic solutions at all. So I always go back and say if you can map out your entire process, like how a human would do it manually, and then figure out okay, how much constraint should I put on each of those nodes of what somebody would usually do. Then that would help you map the agentic framework on top of or the agent system on top of that workflow. So workflow for me is the first citizen, and then the agent comes as the second layer. if you go even deeper, then there's data that it's sitting on that that the agent's gonna sit on top to to basically drive the logic. So and even to go a bit deeper on that. Currently, when I joined Guild, the first thing that I you know during the interview and even talking to folks. Nifemi Aluko: Mm-hmm. Yeah. Thank you, nothing. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: The thing that comes to mind, and it's not only a guild thing or a this company thing, I think it's a universal thing for most companies. most of these automations and AI automations and agentic automations that companies are going to build, for me, I think it's gonna act as a magnifying glass to their data infrastructure. Most of the data was not built to be consumed by agents. They were built for humans. So they're in silos, some of those and be We have data in Snowflake, we have data in Databricks, all these other different systems, and then Nifemi Aluko: Okay. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Exactly. So h how do you have all of them in one semantic layer where the agent can then consume them without having to go like learn how this system is structured, this other system is structured. and of of course you've seen companies probably that are trying to solve that problem, like Glean, they connect all the data, they have that layer, and then the agent can start on top. So the way I think about it, I think from those fundamentals first and then we kind of build out from there. And then of course iterations and improving it. Yeah. Nifemi Aluko: Yeah. Okay. Yeah For sure, for sure. think, you know, one thing I also learned, a big thing I learned was a lot of people want to build these things, want to get right into it, but there's so much in terms of that data layer, in terms of data engineering that needs to be done to make sure that the data is served properly before you even get into all these, any fancy thing on the top, you know, in the application layer or the inter-layer. How do you know when... How do you distinguish between a question or a prompt or a project that is worth pursuing, right? Or one that is just merely solvable. Like, let me give you an example. I remember talking to, while I was at Deep Learning AI, I remember talking to one of the curriculum developers and I was like, man, I wanna use an agent to, I don't know what I was trying to do. I was trying to build something that can help with sorting the courses. And I talked about this whole elaborate way of an agent to read this and this and that. And the guy, an older engineer was like, I think you could just do some clustering. And I was like, clustering? Then I had to go back and start learning even more the foundational, some of these deep learning algorithms, supervised, unsupervised learning, all that kind of stuff. and what I was doing was we, what I was trying to do was a lot, I was trying to make it way more complex than what could easily just be solved. Right. So when do you know, when do you distinguish between, you know, a problem that is worth pursuing where you can automate it. And one day you're just like, let's just solve it with, you know, let's just use a calculator to solve this one. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. That is I think I face that every day because currently I'm I'm at a good position at company where everybody comes to me, either I discover the problem or something that needs to be automated, but usually it's the opposite. Everybody because my customers are my internal are internal employees. How can I make you much faster, more product productive within your role? So everybody comes to me, Hey, I have this process, I recorded a Nifemi Aluko: Hmm? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: a a video of how I do it manually. Can you help me automate this? and usually comes in different flavors, right? It's whatever they think could be automated. I have to be the filter of looking at the problem, understanding and then deciding this is gonna be a genetic solution. This is just gonna be a Python script. And I'm the the goal is to get to a point where I let people also self-serve. It for for things that maybe just require script, you don't need to come to me anymore. Fire up Replit, because we have a Replit license for example Nifemi Aluko: Yep. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Build it yourself using the agent there and then just deploy it and then you can use it for your own personal needs, right? You don't even need to know how to code. So the way I think about it is going back to the same problem set. I tell you to really break down how you do it manually. And that's why I love video recordings of somebody doing it manually, because it really helps like they will miss some stuff when they're explaining. Cause for you, it feels it feels innate. You're used to doing it. You're like, I just do this like this, this. But when you're building these systems, it's like, Nifemi Aluko: Mm-hmm. Exactly. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: You really, really have to dial into the point, okay. I click here, I go here, I do this because of this. Cause otherwise you're gonna meet a s a small edge case and people are so not forgiving when AI makes a problem versus a human. If somebody makes a problem, they'll be like, All right, dude, you get it right next time. But a agent does everything nine things right and fails one, it's like, this stuff is is horrible, we're not gonna touch it anymore. Yes. So so those things make it very Nifemi Aluko: Yup. Yeah, they're gonna make a total of that video. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: For me, it makes it very important to really understand who I'm building for and why I'm building it. And then letting them test it over time. And let me iterate. So I to some extent, the way I pitch to them once I've built the solution for them is this is the worst it's going to be. Just test it. It's gonna fail, I guarantee you. Just tell me. Just tell me, give me the feedback. I'm gonna record some, but give me the feedback of what you think how things should go. Basically give me the evals of how this thing should work. And then I make it better for you and tune it. Nifemi Aluko: Yep. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: And I think going with that mindset helps them a little bit more, not be so defensive and say, okay, this thing just doesn't work on the first fail. but going back to the problem set, the way I distinguish between even a script and agentic kind of workflow that I want to pursue is if this system is going to touch a lot of data sources and it needs to have extra logic. For example, there's a function maybe calling queries from from Snowflake, another one using an MCP from this other server to to to to get data from it. And then there's Within each of those spots, there needs to be some form of thinking or transformation that needs to happen to synthesize a final answer. That's when I think, okay, we kinda need an agent here. Because each of those nodes is gonna have a thinking, which basically in a little minute, that's gonna decide what happens there. Of course, and I add guardrils within each of those nodes. But if a a a problem is like, for example, I just wanna be able to get data from Snowflake, remove this column, and then Nifemi Aluko: Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Put it in a spreadsheet and then send it to somebody else. That doesn't need an agent. You just put a sc put a script, clean it, put it out. but but it but if the problem is a bit deterministic, depending on what the inputs are, things could go either way and there's so many edge cases, then I would put an agent, but then put guardrails on top of that on observability. Yeah. Yeah. Nifemi Aluko: Yeah. Yep. Yep. Man, Tony, you know, you're talking on so many things that even for me, you know, that I learned over the past year or two is even having like this mental model of how the thing should be built. Right. You know, people just want to jump in and then you don't think about guardrails in terms of do you want your personal PII or, you know, your personal identifiable information to be. you shared, right? So you're sharing it with this large language model or do you want, you don't want it to say explicit stuff, right? So when you're using it or the, you know, the other part, know, so, okay, yes, you put it in a guardrail, you can't put the constraints, but the other part is evaluation, right? Where, as you said, because it's, it's probability, every time you use it, gives you a diff, it sort of gives you a different response. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes, yes. Nifemi Aluko: How do you build like a systematic way of evaluating, that response good enough? Right? You know, without thinking about those things, it's very easy to just want to jump in and just say, I saw like someone did something on YouTube. I think I can build it myself without thinking about all the steps you have to do to actually get it to a actually very good state, right? For you to be able to. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. Nifemi Aluko: keep improving the product or the solution. I'm sure you see this, but now that you're working in an organization where you have to, as you said, your customers are your internal stakeholders, do you always have to educate them about all these steps or do some of them already know about this step? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. Yeah. Yeah. I I think it's a continuous process. For me, it really helps for me. I think it at my current role where I'm being a builder. But then on top of that, I'm also an educator in a sense as well. And I think it's just a perfect fit for me because I was in DevRel and I was that was my job. And then now I'm not doing the current role is not purely DevRel. It's really the building aspect of it. Nifemi Aluko: Yeah. Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: But the DevRel skills are directly impacting what I'm doing on a day-to-day. It makes me even better at my job. So for most folks, they come to me first with a problem solution. Hey, I want this thing to be solved. So right off the bat, I'm more of a PM because I have to look at the I have to look at their problem. Ask them, hey, can you I give them templates of a PRD, for example, they generate, they use that, they just put it on Cloud as a skill and tell it, hey, this is my problem. They're gonna they can describe it, transcribe. The cloud skills takes that and Nifemi Aluko: Yeah. Mm. Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Populates a PRD template that I gave them. And then when the PRD gets to me, I review it and tell them, hey, could you add more examples of like this is a question that you would typically ask? This is the process that of things how would happen. And this is the final answer of what it looks like. And give me a real example, so long as I'm privy to the data. And then from that, I already have my first eval data set right there. So when I build it, I directly test it against those evals directly and see, okay, I think this system is heading to the right direction. So by the time I give it to them. Nifemi Aluko: Hmm Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: It's giving them at least what they expect off the bat and then improve beyond that. So I have to tr in just throughout that process, that's already training them and telling them this is what I expect. This is kind of what things look like. And through that process, they are learning also how to think like I would think. So it's like I'm sharing my mental models with them. So I like to think of that as just like yeah, I'm just scaling the the the the knowledge that's out there already propagating to them because it will come back to me in a better format. Nifemi Aluko: Mm-hmm. Yeah. Yeah, yeah. Yeah Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Eventually, because I I'm gonna expect them to come up with more workflows. And over time, it's like I'm doing reinforcement learning on them as well as humans. So the PRD should be getting better eventually, right? And my job gets much more easier in terms of just building. So I I think those are very and sometimes I just even once I build a thing, I record a loom and share it with them explaining all the technical details. I start first with like over eye overview. This is a UI, this is what it does. Nifemi Aluko: Yep. Yeah, yeah, yeah, yeah. Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: And if somebody just benefits from the first three minutes, four minutes, and they're like, I'm good, I don't want to know the technicals, great. If they want to dive into the technicals, depending on their domain, there's breath they can jump into directly within that whatever twenty minute video. So I found that very useful for me and also for everybody in the company as well. Yeah. Nifemi Aluko: Oh, nice, nice. So, you just to do a quick reverse, right? Where you're at now, you're a builder, you're an educator, but similar to, I know myself, I'll talk about, you know, where I'm at now, I never envisioned that, you know, when I was growing up, when I was in Lagos, right, growing up, I never envisioned this like... 20, 22 years afterwards, right? So that's a quick reverse. Like, can you tell me how you got into this all the way back from Kenya? How, if you could, like just a little bit of that journey on how you got to where you are right now. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah. Oof. sometimes I've had a lot of people tell me, you need to write a book. just about the journey. It's yes, I I need to I need to learn some tips from you, you know, on how to do this. So my journey begins in Eldered, Kenya. And this is in two thousand and ten, December. I had just graduated high school. and like most African schools, you go to high school Nifemi Aluko: Yeah. Hey, come join me. Come join me. Thank Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: you study, there's no multiple choice, of course. Like you you really have to really put in the effort. And after graduation, there there's a certain percentage of people that get selected to go to uni. the government kinda pays for that. And then for the rest of you and they pick your major for you, of course, right? So and then the rest of you either you have a rich family or you just go back to farming. so I grew up in the rural in the rural Kenya, you know, north western part of Kenya. So I was a I was I was a farm boy, you know, just taking care of Nifemi Aluko: Mm-hmm. What's the name of the area or the name of the area you're talking about? Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Of this of the s small city. It's called Captaget. For for the people in the running world, you you hear Captaget, it turns like bells go off. This is where you if you see any long distance runners, they're from there. So I grew up. Most of those people that you see out there are people I either trained with on a daily basis or they live close to me, their neighbors. so names like Kipchoggy. Nifemi Aluko: I see. Mm-hmm. Were you also running? Were you also training? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: No. No. So the so the funny part is not not initially, not initially. So the story is very long. So it initi initially, in Kenya, I think here in the US is different. Athletes are like revered. People loved athletes here. back at home, at least the sentiment has changed a little bit. But but growing up, if you went to athletics and started running, and because Kenya are known, Kenyans are mostly known for that. If you went to running, they would you usually the sentiment was you're not good at school. Nifemi Aluko: Okay. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: So you decided to go for this thing because you gotta make money somehow, right? And so most of them graduated from high school and never went beyond that and just went to running straight away. Some of them didn't even go to high school, they just started running. because it gets you out there, it gets you making money faster. It's hard, but that's kind of the thing is. But the tribe I come from is called Kalanjin. And so if you look at my last name is Kip, if you also go and look up most marathon runners, most Kenyan athletes, they have a kip in their name, Kip Choggy. So Nifemi Aluko: Mm-hmm. Okay. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: It's a tribe. Most it's Nifemi Aluko: Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: mostly one tribe that produces a lot of runners. Other tribes also have runners, but most of the long distance runners are from the Calling Jin tribe. So that my journey started there. I graduated high school and like every other person, you know, limited opportunities in Kenya. I figure I gotta do something with my life. I think there's the chance to go to the US and the way is you start running because coaches come from the US and they come recruit high schoolers. Nifemi Aluko: Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Just finished high school to come run for NCAA here in the in the US and they give you full scholarship. I was that sounds like a good plan. There's a lot of camps that are within my area because just runners everywhere. So I was like, I'm gonna I'm gonna join one of them and I'm gonna study for the SATEs and take that and apply to colleges. And then when they when the scouts come, I go run at the track, compete with like 100 people on a dirt track. They pick maybe top ten, you get go to the embassy, try your luck, because the embassy is also a fluke, you know, depending on how you get it or not. So Nifemi Aluko: I mean, you be a little missed. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. Yeah, you know, you know, you know how it is. So I got lucky in twenty twenty eleven, December I went on actually was the twenty fourth, which is almost a day from from today, it's kind of like time reference. Yeah, and I went, I got a visa and January eighth I was in Georgia in college. and since then, you know, decided to come do biology, I majored in biology while in that small college called Young Harry's College. And then a year in I transferred to Grambling State in Louisiana University. And I got recruited there for NC developer division one, ran cross country there as well. And my my life has been more of like getting recruited a lot. So at Grambling, I spent a year. So my junior year, I got recruited by the Army because they were taking people with the languages, the Mavini program. I don't know if you're familiar with it. Yeah, it's military associations. Yes. Nifemi Aluko: Mm-hmm. Yeah, yeah, yeah, uh-huh. Yeah, yeah, yeah. I know some, like a few Nigerian friends that have done that, yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: That went through that. Yes, yes. Yeah, I have some friends too that we joined together that are from all the different countries in Africa mostly. So so we joined that in twenty fourteen. went to basic training in Fort Seal, Oklahoma. in the summer. It was too hot for life. one times. Graduated there, went to San Antonio, did a medical program. on top of being a soldier, you have to have this other extra job, MOS, military occupation specialty. So mine was medical laboratory technology. Nifemi Aluko: Yeah Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: which was kind of closest thing to those I was majoring in, which is still in kind of medicine space. And after that, got stationed in one of the army hospitals and worked as a medical lab tech. So if you go to the doctor, I would be the guy taking your blood, taking it back, make measuring the samples, sending the results to the doctor. So rotated through a bunch of chemistry, urinalysis, microbiology, hematology, all those cool stuff. And Nifemi Aluko: next. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: The last four year three years of that. So I served seven years. The last three of that I relocated to Maryland and the army has a research lab here in Maryland for infectious diseases. And at that point I was thinking, I want to go to medical school, but at the same time I was I don't know. After sharing some doctors in in Virginia where I was stationed, I was like, Maybe this is not the life of I d I don't think being a doctor is my thing now. Yeah. Like I I loved it. But then after watching the doctors all the time, I was like, oof, I don't even want to do that anymore. So Nifemi Aluko: Mm-hmm. Yeah, yeah. Yeah, yeah, doctors are special. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: My Yeah, they're special. I gotta I gotta give him the props for that. I think the lifestyle didn't I just didn't feel like I wanted to do to to do that. So I was like, okay, maybe I can go to research then, go to do a PhD program in biology, microbiology or something. So I came to the research facility here, it's called USAMRIT, US Army Research Institute of Infectious Diseases. they're mostly known for some if you look up the hot zone movie or something, they it's based off that stuff with anthrax back in the day. Nifemi Aluko: Yeah. Okay. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: th this have a some history that could be good or bad depending on how you look at it. So came there and this was the height of COVID. So 2019, I got stationed there. And my time there, it was very coincidental when when COVID happened. So COVID broke out. One of the first few samples of the virus was sent to University of Washington in in Washington, Washington State. And we got one of those early samples in USAMRIT. And our job was to allocate each of those samples and distributed within all the labs within the country so that they can start working on the vaccines. And I was like in the middle of that. So just every day when people are quarantined, I was not on quarantine. It was just like go reconstitute this. Let's get them out. Let's ship them out. So it's fun times as well. And I worked in different departments. I worked in aerobiology, which was pretty unique by itself. We would measure the breathing capacity of whatever animal model we were testing on. And we would then expose it to the agent that we were exposing it to and let it breathe it. And then we would Nifemi Aluko: Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Another team would monitor it and see how that affected the animal over long term. And then that would model into the human clinical trials at the end of the day too. And after that I also worked in genomics, which is where I started picking up programming. coincidentally. And this was in 2020, actually. I wrote my first Hello World in April. I still have that on my GitHub. but I got interested. I was like, there's this I do most of the stuff on the wet lab side of genomics. I go run the assays and they would come take the data. And I was wondering. Nifemi Aluko: It's your heart. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: They're buying from a bioinformaticians. And I was What do you guys gonna do back there with all this data? They like, You wanna come see? I was like, Cool, yeah. So I went and saw it and it was just a bunch of Python on the screen. I was like, Okay, this does looks complicated. And they were like, you're interested in it? Here's a book, you know, Python for dummies. And I took it read in the home and was like, Code Academy it is. I spent in Code Academy like a couple of weeks and I got hooked. I was like, This is what I want to do when I get out. So I got out in twenty twenty one. So twenty twenty to twenty twenty one was my window to get something happen and switch my career entirely. Nifemi Aluko: Thank Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: So I spent most of my nights really and my wife, I had one kid at that point, and I would come home, we would just hang out. As soon as they go to bed, I was up until maybe about three in the morning. And then I would wake up an hour later, go to morning practice with the army and then go to work. And that became a cycle for a couple of weeks. It was like pure, just pure dedication. Because I was like, I'm getting out of the army. I don't have much time. I gotta make something happen. Cause I don't wanna go work at the lab when I get out. And Nifemi Aluko: Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: The good thing with the army, they have that last six months of your service where you can go intern at whatever company to help you transition out of the army because it's a it's a hard transition. And I started applying to jobs and most of them were like, Well, you're great at what you do, but you don't have any technical experience working in tech. You just have personal projects, you don't have schooling, so it's great, but we can't even offer you an internship. So at that point I was okay, maybe I need to enroll in a college for for folks to take me a bit seriously. So I ended up finding a program from University of Pennsylvania. They just started their Masters in computer information technology and they were taking non-traditional students. I applied, got in, and from then it's just became history, you know, got some internship at Merck, the biopharma, did data engineering there. And then I got out in twenty one, went to Bloomberg in New York. that was that was wild just coming off the army and straight into that. it was also a very good learning curve. So I did some data analysis there with financial data and just working with them. Yeah. Nifemi Aluko: Yeah, you have a lot of data. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah, you know a lot of data. A lot of data. A lot of six hundred billion ticks in in a day. so that was fun. And then this was also during the height of COVID. So family didn't really love the vibe of New York, New Jersey. And they were like, Man, we don't wanna, you know, or concrete blocks, kids are used to being outside and then like we just it's just stressful. So we ended up moving back to Virginia and I moved to Virginia, took a job with Booz Allen as a contractor there, doing data engineering as well. And then after that I stu I was doing stuff on on YouTube. So since I'm mostly self taught. I would learn something and then I would record a video and post it. And that just became a thing. And eventually I was like, I'm just gonna keep doing this because it's fun. So I was doing something with Strimlet and a role popped up with Striomlet for Developer Advocate. And they had just been acquired by Snoff like about like I think a month in. And I got in and started working with awesome folks like Chinin, and and some of my awesome managers there. The team was really awesome. And from there, Nifemi Aluko: Mm-hmm. yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Was there for almost two years. It was an awesome time working there. And this is also when AI started becoming at least generative AI started becoming really prevalent. And I was doing a lot of stuff with generative AI with at Strimlet already. So I ended up interviewing Joao. He had just built a package. I think it was about a thousand stars. It was so early. Like nobody was talking about agents then. So I brought him on, I interviewed him, and then we just kept in touch. And when it was time, I just told him, dude, I'm looking to Nifemi Aluko: Thank Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: To switch to go fully into AI space and this in 2023, 2024. And he was like, Yo, dude, I'm I'm hiring. You wanna come on? I was sure, you know, and the rest is history. We worked on the product there from seed stage to post series A and a lot of customers. It was really awesome. And now I'm at guild, and I think the transition for my process here has been I've learned how to build those systems. I know how things work under the hood, but a lot of companies run that right now are struggling with just how do we then implement this? It's like boards are pushing all the companies. Hey, you need to have AI strategy, you need to have an AI strategy. But then it's a different thing from just pushing everybody to use ChatGPT within the company, but how do we systemize it and actually can you see them ROI coming from it? So I think that's where I felt I I'm at a good intersection to do both. So and that's kind of like my journey in a nutshell now. Nifemi Aluko: Yeah Thank Yeah, dude. I know, and that's a very interesting, like you've taken on the twist and turns of getting to where you are. And the interesting thing I've learned and even hearing your stories, it's not like careers are so linear, right? Some people just think like, it's just linear. Some people do have that. They could tell you this step-by-step story of like, knew exactly what I was going to do in my life. But you know, there's some, as you said, you went through the stages of like being recruited and then, you you had this moment where they're like, what are you guys working on? wow. You know, that's what you guys are doing with all this data. Let me go learn this thing. And as you said, even the roadblocks of now trying to interview and people are like, but you don't have the technical, you know, expertise and then going, okay, Penn State now has this thing and they're taking. you know, non-traditional. There's always like, sometimes there's persistence, but there's also, you know, right place, right time, or just finding a way, right? How, for you, how did you, the more you were pivoting and finding this thing, when did you start getting a sense that, you know, what you're working on now is what maybe you've wanted to work on, you know, in the past, like? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. Yes. Nifemi Aluko: Now things can change obviously in the future, but now in this state here, how did you start getting in the sense like, oh, this is what I want to do. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah. I think for me one thing that stuck with me in the beginning, early days, I think it's something that it's innate. Maybe it started like when I decided I was gonna come to the US, and relentlessly pursuing that objective. And one good story for that is when I started running in Kenya, you know, nobody in my family, at least most families within my area within Captaget, they have somebody that's a runner and they they they're doing pretty well. but my family tree line, nobody had ran, nobody had started running. Not because they didn't want to, but it just nobody just never ever tried it. so I was kinda like the first one to just kind of like, I wanna go to the US and the path for me is running. So when I started running, most of my aunties, most of my uncles and even my dad was like, are you doing this for physical fitness or what? Like it they do they were like, You're not serious. They were like, dude, nobody in our family ever ran. Like we get it, but like, are you serious or are you just, you know? Yeah, you know. Nifemi Aluko: Yay! Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: And for me at that time I was like, dude, I'm actually serious. In my in my in my I was like, if y'all don't believe in me, I know I'm believing myself and this is the path I wanna I wanna pursue. So I remember running and some days I would wake up in the morning, like it would be really hard to wake up because you're like, oof, I'd wake up at four, go for this two hour long run in the in the bushes with all these marathon runners. They would just leave me after like a mile in and I'll just be by myself the entire time. And it's dark and you gotta come back. Two hours later, and then you sleep for a few hours at 10, because we're training three times a day, four in the morning, ten, and five in the evening. So morning would be like a long, hard run, midday would be something like mile splits on the track, and then evening would be like a slow five K jog, which the jog is relative. It's a bit speedy. but for me, that was kind of like the opening. I would say that when the my bulb went off in my mind, it was like you've really gotta go after things. Like there's no Nifemi Aluko: Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: It's not gonna be easy. It's just gonna go after things. And then since since then it's just been like that for me. And the way I like to kind of think of it, even with all those pivots that I've that I've made, I think one thing that stood out with me as well is I think I was listening to one of the episodes and I think Jeff Bezos were in there and talking about the regret minimization strategy. You don't wanna down the road, you wouldn't wanna have a lot of, I should have tried that. I should have maybe if I did that, this then my life would be like this. So I took that in my mind. I'll say, well Nifemi Aluko: Hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: I don't know what I want to do right now, but I think I'm interested in this thing. And the moment I said I'm interested in this thing and I feel like I'm interested in it, my goal has been can you investigate? Can you figure out if you're gonna like it? Just try it and see if you like it. If you don't, you can always back out and try something else. I I think it's the two-door strategy. If there's two doors, just pick one. Don't spend a lot of time thinking about it. If there's one door, then relax and really think through it because it's one way. If it's if it's a two-door, you can go in. This one, if it doesn't work, come out, go to the other door. So Nifemi Aluko: Mm-hmm. Yeah. Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: That's what I've it's helped me a lot. So even the deciding, okay First, maybe let's think about medical school. And then, maybe not, after shouting to some doctors. Maybe research. worked on research a little bit at the infectious disease lab, talked to some of the PIs, principal investigators there. And I saw how their life was. And also they explained to me. They were like, dude, our life, our salary, everything is dependent on funding. So if it goes down, we're we're out of our work as well. And they were like, you really gotta do this if you really, really, really love it. And for me, that's when I was like, ooh, do I really, really love it? That if I don't get paid, I'm still gonna do it. And that's when I was like, I I probably do, but at the same time, I feel like there's a lot at stake for me. At least at this point in my life, I'm not ready to have that type of commitment. So I was like, okay, I need something that I'm gonna enjoy. The way I figure out what I enjoy right now and I and I feel like this is what I wanna pursue is the fact that even if I'm not working, I'm still living my phone. If you go to like the past ten searches for the last day, ten days is something that is involved with my work. Like it's it's Nifemi Aluko: Hmm? Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: I am obsessed with it. Like my social media feed, who I follow, is basically that. Every conversation I have with even my friends when I go out, it's basically that. So it's like I'm living in that obsession to the point where you sometimes my wife is like, Do you ever have an off day? Like in your mind, I'm like, I my off day is when I'm asleep. Like my brain is shut down. Otherwise, it's like I'm thinking about these things, even when I'm just sitting up still still thinking. It's just obsession. I think that's what I what I would. Nifemi Aluko: Mhm. No man. Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: when I found that I'm super obsessed with it, like even on my time that I call free, I'm just doing it. Cause I think at some point my alliance blurred from cause when I was in the MA for example, I would know, okay, I gotta wake up, I gotta go to practice, I gotta go work, and then I gotta come home. And I when I come home, I'm not thinking about work, I'm not thinking about anything. I'm just on the TV, I'm just like nothing. Like it's just like, thank God it's Friday, thank God it's Saturday. But with the current job that I'm in, there's none of that. I finish work, it's Nifemi Aluko: Yep. Yep. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: I try not to do this, but it's just it just happens. like your mind is already thinking about that. You go on the phone is that and it's not even work. It's like it's really hard to distinguish the lines of kind of blurred between work and actually you just enjoy these topics. And that's when I feel like this is what I enjoy. Yeah. Nifemi Aluko: Yeah Yeah. I know, I know. You Tony, you're saying something very interesting because, you know, I feel the same way, whether it's like me building my business or, you know, any like creative projects that I'm interested in. very, a lot of my projects are very like self-directed, right? It's not like someone told me to write a book. It's not like someone tells me to make music. I want to do it, What I'm learning now is trying to find that balance because like you, sometimes it's hard to switch off, right? So even when the laptop is closed, my mind is still thinking like, okay, tomorrow this is I'm going to do or, there's this other idea that I need to work on or whatever it is. like, look, it's not, yeah, it's like, to be honest, it's always like, it's like a gift and a curse, right? Because Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. I can relate. Nifemi Aluko: The gift is it doesn't sometimes feel like work because you're just going, right? The curse is sometimes you can't stop or you can't, you almost have to tell yourself, okay, like for instance, now, you know, it's holiday season, right? I'm telling myself, I'm going to take vacation, but it's like, really? Are you really going? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. Yes. Yes. Nifemi Aluko: So, you know, it's now trying to find like this balance, right? And you know, one question, you know, as you were talking about practice and everything, you know, there are two things I want to know. You know, those people you were running with, right, waking up in the morning going, were they, did you get a sense that they were in their own form of obsessed with what they were doing or were they just doing it as a way out to be like this is a way for me economically to you know get out and help my family for instance or did you get that sense of man this person there's that one guy that just loves this running thing like you can't stop talking about it he's always thinking about it what was your experience with that Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah, that's very interesting. there is definitely there was definitely a mix. I would say a good chunk of them was there's a good chunk of I I would segment it as there's folks that just like I love this, my family's been doing this, it's a family thing. My dad ran, my uncles ran, my brother runs, and this is the family thing, it's like a family business to some extent. Like it's just expectation, there's that burden of it. And they just that's all they know. They grew up knowing that they grew up. Taking their dad to practice, watching their dad run in the morning. Some some of them even run with their dad during the morning run. So that's them and they love it. And then there's folks that they love it, but at the same time, they're doing it for the economic value. Then I think all of them is still tied to economic value, but they they they it started to maybe off as trying to look for economic, you know, uplift, but at the same time, it just turned into like, I actually love this thing. And then there's other folks that were just like, I'm just going this for the money. And if you see the folks that went in for the money. Nifemi Aluko: Yeah Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: the the period of lasting is not long and the also the level of achievement is not but folks that are either in the two camps of I just love this thing versus it's money and I love it I just happen to love it as well. Those are then that kind of work out great for for them. I do have some folks that really enjoyed it. Like even when we came to college here, we ran together and 'cause we would run out the NCAA work, we would run, college would pay our tuition fees, room and and move Nifemi Aluko: is not wrong. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: food and books and everything. Our job was to run and maintain a three point GPA in college. That was the requirement. Some of the folks that I was with, for example, they ran we graduated, like like we did. Some of them ended up joining the army as well. But when they joined the army, they went to the army world class athlete program, which the army has, and basically your job is to run and compete for the army. And they competed all the way to the Olympics. And there's one good friend of mine called Paul Chelimo and he went on Came in second in the Olympics in 5,000 meters for in for the US. And that's somebody that loves it. Right now he's transitioning to marathon. He has a YouTube channel now. And he's all the day's content is just running content. He just loves it. he also comes from my we went to the same high school in Kenya, and we we all come up from the same region. Yeah. So so it's he for for me, I'm like, I know he loves it. You know, that's that's a person that loves what he's doing. that's an example of it. And Nifemi Aluko: It just. Yeah Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: For me, I think it was more of a means to an end. I it's not and you could you can see what I from what I told you. Nobody in my family ever ran. So I didn't have that like somebody from the family that did this, that I'm okay, no. But I n I have neighbors, I have a lot of friends around that I saw them running. my my goal was I'm gonna go there for to help me with education. Running will help me get educated. And that was kind of my means there. Yeah. Yeah. Nifemi Aluko: Yeah. Mm hmm. Yeah, yeah, yeah. Nice, man. And you know, you did say something like, you know, waking up, going for those runs, being left behind. And you're like, it was this hard. Right. And almost like the light bulb of anything you want to do, there's going to be some hardship. And, know, to be honest, personally, I'm trying to even shift my mindset of like, oh, things have to be hard or whatever it is. But the reality is, you know, I have friends that like want to run their own business or whatever it is. And they're like, man, you know, I want to come run my business. I'm like, hey, feel free to come run a business like it's hard, right? Having a corporate job to his heart. Being a parent is hard. Being single is hard. I'm like, let's choose whatever hard you want, like, just stick with it, right? And, you know, through some of my conversations, Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. No. Yes. Yeah. Nifemi Aluko: on this podcast. One thing that has been coming around was someone saying this like, helping people learn how to do hard things. Right. And I feel like we're in this era where, you know, there's this constant need for like a dopamine hit of whether and I'm, you know, guilty of it of like, whether it's like scrolling on social media, whether you're posting stuff and you want immediate response or whether you're building something like I'm building something here and I'm like, yo, this thing needs to work. Come on. Right. but I'm also learning, you know, the hard way or okay, there's a process to post into online to eventually build trust. There's a process to building agent systems that actually work. It's hard, but I'm learning how to do these hard things where, you know, as an educator yourself, right. when this era where students can easily get access from Gen AI, what does your job become in terms of trying to teach them stuff? what do you see as your job? that helping them get through hard things, helping them get that mindset? Is that part of it or is there something more? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah, I think there's something more to that as well. things have become easy, like you've noted. Yeah, these days you just you don't have to think. You we've kind of offloaded our thinking to the models. it's like, yeah, I don't have to think. You think for me, and then I synthesize, and then I let you think again and become like a the thought partner, right? And it it's great, it's helping, but at the same time, I feel like the fundamentals are getting eroded to some extent. People are not trying to dig deeper into specific things. But that could be argued the other way too, with These agents and LLMs now, they allow people to dig deeper into things that they would have previously had to pay for course or go find a video about certain things or read a book, which it's in itself, there's a lot of friction to like getting to that point. Whether whereas this, you just ask you the question, it just goes does digging for you and surfaces like the most the easiest way for you to learn. You're like, I understand this now, I don't understand this. Explain a little bit, let's go back. There's you can go on tangents and then come back in, back and out, in and out. So I think. Nifemi Aluko: Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: My biggest role as an educator, from the way I would think about it, is how do I then teach these learners how to do that? Even like as as a thought partner, how do you use an an LLM, an agent as a thought partner, instead of just kind of depending on it purely on like, I'm just gonna depend on you to tell me everything that that you know about this, and I'm just gonna sit there and kind of like digest it. I think the goal is know something, understand a little bit of it, and then be able to ask questions to expand how you're thinking about it. And make you more holistic. It's more of like it fills in the gaps that you would initially in the past would have had to grab a book, read it for a couple of days, and probably you're reading stuff that is not even contributing because books are dense and they they will take you on tangents that you're like, I this is exactly what I wanna learn. Yeah. Nifemi Aluko: Yeah. Yeah. Yeah. Yeah. Stories and all that might not be necessary. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Exactly, yes. And the point is like on page two hundred and five, you know. So so with with LLMs, I think it's like teaching people how to how do I use it in the most effective way to help me be better at what I'm what at what I'm trying to do. And I think for educators, and that's where things are going, be kind of like the the curator of helping them understand. Okay, this is how you need to be thinking about solving this problem. And I think the other thing here as well is just Nifemi Aluko: Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: There's the tools. So I like to think of this AI and LMs. These are just tools. But at the end of the day, it's like, what are you trying to solve or help other others make their work easier, make your work easier, make others' lives better, in the grand scheme of things. And so thinking about how this becomes like a your thought partner and how you help it kind of help you get to where you're going. So those is that's where currently I'm thinking in terms of just helping folks rather than rather than yeah. Nifemi Aluko: Yeah, yeah, yeah, yeah. Cool. So, you know, do you have a particular way that you create conditions for someone to discover something they didn't know they needed to learn? Right. It's one thing I learned from working with a lot of dev rels, a lot of AI educators is, you know, you want to give people that aha moment like, now I know. Okay. Now I know why this thing is important. How do you, have you thought, how have you thought about creating those moments for people? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah, so I think there's has to be a few things. First of all, there's the environment. Do they feel safe and do they feel there's a sweet spot where they can just they have a playground where they can experiment safely without without really causing any issues for them or for for the system or for anybody else that you're using it? So once they have that room space and safety, I can actually play with this thing here and I'm not gonna, you know, not gonna have any problems. So Nifemi Aluko: Hmm? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: You you as an educator have to set those constraints. Like, okay, this is the constraints of the system. And then, you know, you play within this space. And then the other thing is once that is there, it's actually helping them by giving them more of like hard problems to solve within that environment as well. It's like, okay, you now understand there's this place where you can play, but here's a problem that I want you to start thinking about. Start resolving this problem, using this tool within this environment to solve this issue. And that in itself, I think the core part of this is to go back to critical thinking. How do you and I think this is more important now, even with AI? The more critically you think, the more you're able to like because when when I even give talks, one thing that I've been mentioning a lot is we have tools now that are so powerful that the main limiting factor right now is just our imagination. If you have an idea and think of it today, you can build it literally almost within a day. Cloud code is here. It can build a website for that idea. Linking to Stripe. Building the product, hooking it up and putting it out there for people to use is so easy now than ever before. So the biggest thing is the idea. Maybe start looking at your experiences. What am I doing currently that it annoys me that I could build and like make it use you useful to help me? And then there's probably somebody else somewhere that is running through the same problem and finding those people and marketing to them and getting that product out there as well. So I think my lenses have shifted to let's go back to problem solving. So I think topics like philosophy is becoming very important right now. than how it's been in the past. Nifemi Aluko: Yeah, yeah, yeah, yeah. And you know, it's funny you say that because that's actually even why for me, I'm even asking these questions and doing these, you know, interviews because obviously I know I've been around a lot of problem solvers, a lot of people do building stuff and you just like the limits, there are no limits to what we can solve now or the tools we can use and there are a lot of big problems. So it's now how do we solve it? What's the approach we're using and why and what is having conversations about what is actually important to solve their big problems, inequality, climate, whatever it is. So how do you then choose? So just three more questions if you have the time. In this, feel like the AI space in the last two years has become very noisy. You're like, man, finally. one finally came out. Great. Then, Cloud Code. my God. I need Cloud Code. my God. What's the next one? Well, before Cloud Code was, know, better start using Cursor or Windsurf. Then Cloud Code came around. And then are you using warp? And then, you know, the next one, know, and then just keep going. Just keep going. Just keep going. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yes. Codings. Nifemi Aluko: And if you're not careful, you can find yourself like feeling like you're always behind, right? How do you, how do you stay aligned with your taste or your values in what you want to do in such a noisy environment? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah, that's a good question that I still also am looking for best answers there. But I think the way I like to think of it is going back to what you said, values, right? And I think that's the most important topic. I think figuring out what are your values, what do you enjoy doing and how does this new technologies that come out then shape how that changes and how you approach that. I think everybody within every role currently right now in the world, any job really, there's one way or another that AI is affecting that in a good way. I think as well as in a risky way, for example, support. If you're working in support, customer support, then there's currently a lot of issues there in terms of it's getting better. The voice models are getting much better and they get a point where it needs to be steered. So I think the biggest thing that I think about is for example, if your support, you need to learn how this system is working a sense, not even the technical details, but how does it work? How is it doing what I'm doing as a human? Where is it falling short? How can I help it as a you become like the thought partner for it? 'Cause it still needs evals and you are the evaluator. Nifemi Aluko: Hmm? Exactly. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: And so that's still the job. You still if you do good at that, then you're really good at doing this manually. So you're the expert in this case. And so when these technologies are changing so frequently, I also try to calibrate myself. Well, this is changing. How is that changing how I do stuff now? For the for for DevRel, LLMs and AI can generate content now. They can write blogs, but they don't write great blogs. They like okay blogs. So maybe I just take all the blogs I've written. Nifemi Aluko: Yeah. Okay, yep. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: feed it to so it knows how I write. And then we explore a topic together with a L L I just we just par just tell it, okay, here's a thought I have. And then it tells me goes start some research, gives me the thoughts of what and then I give them my counter arguments of why I think the current thinking system is wrong and this is how I'm thinking about it. And then it synthesizes that with me and I tell it, can you critique me if you were a subject expert of this topic? And it just gives me the critique and then I enrich my thought process. And at the end of it I come up with something solid at the same time. So it's like Nifemi Aluko: Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: It's just another partner that you didn't have that's there with you all the time. And for me, the way that's shaped my thinking is more of like I see this as an abundance thing more than it's reading me off some stuff. It's definitely making me better in some sense 'cause people have argued it's making people dumber. I get it. If you fully rely on it in a different I think there's a there's a skill issue and usage issue here as well. But to begin with, if you're just relying on it purely and just like telling it, write me a blog about this X and it just does it and you just get it and you Posted that is the AI slob we don't want. I think it's like the third partner. And given your own taste. And I think the taste part is what do you agree with? What do you disagree with? I think you have to be very thoughtful on that aspect. Be be open with what your internal thinking is. Like, I don't disagree, I disagree with this because of this. And make those beliefs known and let it let it let it dispute your beliefs. It also helps you actually understand yourself better because you're like, I disagree with this, but then if Nifemi Aluko: yeah. Yup. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: few counterpoints that maybe makes you realize, actually actually I'm wrong or I'm right or yeah. So yeah, 'cause if you just let it be your yes man or or yes device just says yes to you, it will do that very, very well. Nifemi Aluko: Yeah, exactly. You can definitely be a huge yes man in life. was watching, you know, I was watching this documentary about this designer and he was talking about one of his mentors and he said, you're going like long walks through the woods, through the city. And that mentor always would point at like houses or buildings and they'll say, tell me one thing you like and one thing you don't like about this building. Right. And the more he did that, he was creating his own taste or even being able to like articulate his taste, right? And that made me think like, that's very interesting. Like it was like a practice I would like to do because we kind of do it internally. For instance, you read a book and you're like, I like the book because of this, but I don't like it because of this. And everybody has their taste. Do you have a way of like, you know, bringing Do you have a ritual or things you do to kind of bring meaning to your life when things start to kind of feel blurry? Do you have a practice or way you think about things? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: r rituals, yes. so for me I have a I don't know if I have a few. I probably have some. Most of these are, like you say, innate. You don't even spend time thinking about it. But I think some of these are some of these are formed from just childhood experiences as well. like you know, things have changed, maybe regulations have changed about a certain topic that you disagree with, but then how do you adapt to it? How do you how do you just coexist within that environment? So for me it's become more of Nifemi Aluko: Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: It also challenges some of the notions I had in the past, especially I I would say most Africans, the way things are shaped in our continent, the way our cultures, because we come from a culture that's more rigid, and then come here, it's things are kind of loosey in a sense. there's no specific really it's mostly guided by regulations, how people think about specific things, versus there's really cultural, because you could boil down even to the tribe level. Yeah, and then the country level, there's still big differences within all those cultures. So I've had to tune off some of the things that Nifemi Aluko: Cool. Yep. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Over there was like it's huge, but here it's it's a little different. So I think for me is more of looking at when something new changes. Like for example, AI came around and the first thing is like, wow, it's gonna do this to my life, it's gonna change this, I'm not gonna have to do this. Well, does that then make me relevant at my job? Then how do I adapt to it? So I think for me, the way I like to to look at it usually is from a sense of just myself thinking, okay, this has changed. How do I stay Nifemi Aluko: Thank Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Okay with things that have changed. So I think always learning is always my thing for for for to kind of keep myself sane. It's like this thing has come out. I just want to learn about it. I want to learn about it, understand it, and understand how it's changing me. And that's one thing to just get keep me afloat and sane. And then just also leveraging everybody around me. Why what are you with what do you how do you think about this topic? I'm thinking about it this way. How do you think about it? And I think that helps me get it get grand a little bit because. Nifemi Aluko: Mm-hmm. Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Thinking for yourself is great, but at the same time, there's like this silo thing that you just sit in a silo. Like I think a good example is even right now with AI, I think there's a something with with just being within that group of people, for example, Bay Area folks. Yeah. You know, you're you all think this is how the entire world is operating. But then you move out of there, it's like chat GPT, some folks is still on foreign concept. And you're like, what? Nifemi Aluko: Nope. Exactly. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Exactly. So I think there's there's that aspect. I try to break on the dog bubble and like get some grounding a little bit on what's going on as well. Nifemi Aluko: Yeah, yeah, yeah. That's a perfect example for sure. Building in isolation. For you, how does building technology shape your sense of purpose or contribution to others or to the world? Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: There's a direct correlation, especially right now with within AI. things that we're building right now are helping people. I think a good examples of stuff that are out there right now is even with Neuralink, for example, things have changed a lot. You've seen people that were not able to perform tasks, just being able to do stuff and enjoy life. And I think the main purpose of us on on this planet really is how do we make everybody else lies better in in a sense and better is relative as well. How do we just make people enjoy the little things they need to enjoy rather than suffer? Or at least reduce suffering in a sense as well. And so for me, when I think about that is how can I, even at the smaller scale, somebody that works at a company, how do I help them do their job more efficiently? They're tired of clicking these different buttons on these different screens. How do I then make their life easy just like maybe this thing can run on a schedule and then they can start thinking about other things that could be more beneficial. Maybe it even relieves them time to actually go learn about this AI thing, right? Because they've they've gained some extra hour or day from what they were doing to actually go learn about this thing. Or even just not do anything as as well. That's still a form of benefit. So my point of view is that how can I make this better for folks? And then not only that, how can I help them understand it by just the teaching aspect of it? Because it's one thing to just kind of do it and then they just okay, it's great, it's working. But then what? So I think there's that aspect of it works and then teaching as well. So for me I find Really great satisfaction from doing both of those as well. Nifemi Aluko: Mm-hmm. Yeah. You know, even for me, what you just said really resonates because, know, sometimes when you're searching for purpose, you're always like looking around like, like do this or that. I remember reading a book by this, I forgot his name, Yuval Harari. He was talking about something about 21st century, 21 lessons. And one thing he said was, just reduce suffering. Like that's a clear thing that you can actually focus on because everybody's suffering in their own way. but if you can reduce collective suffering, that is actually like a straight up goal, right? You don't have to think about like the multiple ways, just find your ways to contribute to that. my last question for you, and this is kind of like a playful question that I've been asking everybody, right? So if the world went dystopian, and dystopian is we're all here working for the machines, we don't have agency. We're reporting to them, they're the higher lords at this point, controlled maybe by a few people. Or we go to a utopian world where we have friendly AI, we have trustworthy AI. It's doing all the work for us. Then we just have to sit back and try to figure out how we enjoy life. Everybody's taken care of. As you said, there's abundance of resources being used. Maybe there's some kind of like universal basic income. Maybe everybody, nobody's hustling to do that. So you have these two situations, dystopian or utopian, right? You have to choose one. You have to choose one, right? One of those situations. And then you tell me what would your daily practice be in either of those, in this situation that you choose between dystopian or utopian. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Wow. that's a really interesting questions question i in itself. So picking one, oof. I think my answer will cut through both. so it's not an option, but I think I'm gonna add a third option for all in this case. And I think one thing that I would focus on specifically is skills. And I'm gonna use skills as leverage in this case. Because in a dystopian society or or a utopian society, everything is there already. It's everything is great, every everything everything works, works as as it should. And in a dystopian society, I have to Nifemi Aluko: Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: make things work because just necessity is there. So the the basic thing that I would think about is continuously learning useful skills that are portable that I can use it in different ways. So I think things like communication, mediation, because disturbing there's gonna be conflicts everywhere and and just kind of chaos everywhere. How can I learn that? It's maybe even me br being a bridge between the the new AI overloads and the regular humans, which now sees us as just like dummies. Nifemi Aluko: Yeah Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Maybe I can be the mediator between the two, right? there's like the machines are like, We're so advanced, like you guys are but I'm like, look I Nifemi Aluko: Yeah. Huh? We need someone to do the translation. Tony's like, I got you. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Exactly. Exactly. Or I can learn and be really good at fixing a specific component of them that they're like, okay, we rely on you because you're cheaper, but you know how to fix this thing, like you go do it, right? So I'm still useful in a sense. or even for food for food like growing food, right? Like maybe they they the machines are like, We don't want do that. It's it's mundane work, it's just somebody can do it, you're human and you can do it. We don't pay you, just go do Nifemi Aluko: Yeah, Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: So learning those, I think that can contribute in good times and also, you know, keep me alive in the bad times as well. So in good times, even running grow food in in a utopian society, that's still great. It keeps me alive, it would just help everybody else. But at the same time, in a bad condition, it will keep me alive and others alive as well. So it kind of lets me contribute in in in both. And I think of the skills more of a as a stored option value, like a options in the company. That's the stored option that I'm putting it in. And I can Nifemi Aluko: Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Accumulate a lot of those skills. Since it's easy, AI is there for you know, in the you've this the learning it is not the problem. It's really do you how much of that do you want to amass and learn about? So I think I would focus pure in just kind of acquiring a lot of these skills and then storing them. And then it can help me throughout that entire aspect of it. The other thing is really community mindset here. And this is just like the regular checking in on people. I think that's Nifemi Aluko: Yeah. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Very important, of course, like checking in on people, resolving conflicts, you go back to communication aspect of it and kind of like knowledge sharing as well. And these will cut across the machines and the humans as well. So it's kind of really being there and understanding how each one of those work. Because it the difference really is more of like the isolation and a function of like mutual aid b within the network of humans and the machines when that works out. So if I was to say two things is amassing a whole whole bunch of skills. Nifemi Aluko: Yeah. Yeah. Mm-hmm. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: And then putting community first. I think those are the two things that I would do to survive in both environments. Nifemi Aluko: Yay. For sure. Hey, Tony, man. Thank you very much for your time. I really like how you ended it there. There's something big there about community. I'm hearing from a lot of people and definitely picking up the skills of just keep learning, right? Just keep learning, keep going. So I really enjoyed the conversation and thank you very much for your time and for joining me today. Tony Kipkemboi Tony Kipkemboi webcam 00h 00m 00s 215ms StreamYard: Yeah, thank you. Thank you for having me. This was really fun. Nifemi Aluko: For sure, for sure. Thank you.