Breakeven Brothers

Bennett Bernard (00:11)
Welcome back to the Breakeven Brothers podcast, episode seven, as usual, joined by my co -host, Bradley Bernard. Brad, how's it going?

Bradley Bernard (00:21)
Pretty good, a little busy, lots of lots of updates to share, but not too bad. This weekend has kind of been a bit relaxing. I think the past few weekends has been prepping for interviews and kind of struggling through that whole boat. And so this weekend, almost done getting there. Don't really have too much to share yet on that front, but it's, we're coming to the other side of things. I think that's really exciting. I got some cool stuff lined up, but yeah, a decently chill weekend so far. How about you?

Bennett Bernard (00:48)
Yeah, pretty busy week for work. yeah, for those that don't know, we haven't or those that notice we haven't had a podcast episode in some time. So Brad's a super busy. We're going to get into, know, why exactly and he's gonna give us some interview updates. But yeah, weekend for me has been pretty good so far. F1 is coming back this weekend after a long summer break. So looking forward to that race early tomorrow morning. And yeah, I've been tinkering a lot of tinkering with a lot of

cursor and LMS and a lot of stuff we'll get into here. So I was really excited to be getting back on these podcasts to talk about some of those topics here. So yeah, happy to be back.

Bradley Bernard (01:17)
Hmm.

Yeah, it's been busy. I think we had talked about recording a few times and every time I came down to it, I was just like, Hey Ben, I can't do it. feel like I got to study or do something to get that done. And so, yeah, I think the past probably two and a half weeks, been moving on to the onsite stage for companies. And it's been a lot of work. mean, you're spending five hours, you know, back to back interviews, talking to people. You have a mix of technical, you have a mix of behavioral. So they're asking you, you know,

tell me the most proud project you had. And for me, working at my own company, then working at Meta, usually I have to kind of base my stories off Meta because they're looking for, you know, big company work, multiple teams, multiple orgs, a large impact. But yeah, I definitely have to craft the right story, deliver it, kind of read the room, understand what they're thinking and then be very well adept on the coding side. So iOS coding, algorithms coding, like all that goes into the onsite and you have this one shot, one window to like make yourself look good.

And it's a, there's lots of luck, lots of chance, lots of preparation. I'm happy to say that when I went into this phase, I tried really hard to prepare and like every facet. I think things have gone pretty well. Again, after the interviews, like I usually understand how well I did. You never know if it's going to be a full yes or a full no. There definitely been times where I know I didn't do well. And then I'm like, darn, that sucks, but it kind of helps you study more understand where your gaps are. So it's definitely a learning process, but yeah, I think.

these past two weeks, probably taken, I don't know, seven onsites and I had one in person that was a cool experience, but overall really, really busy coming to the tail end of it. So don't have too many more, but yeah, getting feedback, mostly positive so far, which has been great. But yeah, then kind of wrapping up and choosing where I want to go after all this, but it's been super busy. I've been dying to record with you. I think it's just been a crazy, crazy few weeks.

I'm seeing the other side. think when I was right in the middle of it, I was like, when is this over? Like you start it, you don't really realize when it's gonna end. then now I kind of see the end in sight. And so I'm pretty happy and excited about it.

Bennett Bernard (03:31)
Nice. Yeah. And you have a little bit of a break right now and then you're going somewhere next week, which we'll get into. but yeah, that sounds pretty cool. Any specifics on like the interview updates that you want to share or is just too early at this time and you know, still kind of going through the thick of it.

Bradley Bernard (03:48)
Yeah, probably too early now. think maybe next podcast, probably have more to share there, but yeah, trying to figure out where things landed with feedback, figure out scheduling for other events. then, yeah, probably decide and kind of talk about what that process looks like. There's so much that goes on behind the scenes. Like you spent all this time preparing, then you go through it. Then there's like the, you know, where do you go offer negotiation stage and to do that effectively, you have to understand as much as you can about the company. And there's so many different facets there. So.

Yeah, I think in our next episode, probably dive a little bit deeper into what that thought process looks like. But from where I'm standing today, a little bit harder to kind of outline that still going through it, but almost done with that process. yeah, I traveled for a week, went to Boston, went to Maine, had some fantastic lobster. Then I came back.

And then it was more on sites. then, yeah, next week I'm going to LairCon in Dallas, Texas. So I'm really excited about that. So as I've been trying to schedule this whole thing together, been a little bit of a nightmare. I've been telling recruiters, you I'm free here, not free this week, free when I get back. And then, you know, only free for that gap right there. so thankfully a lot of it has worked out, but I have felt a little bit of, you know, pain trying to get my schedule on, you know, on their side, try to make it work. And I've been filling up a lot of my days. And so a lot of time when I'm

getting down to the thick of it, have like maybe one or two days open and giving them, you know, the small window that's relatively soon. And I don't have much margin for error since I'm trying to kind of collect things at the same time. been a little bit challenging. felt a little bit of a, you know, a problem person to try to get this to work, but people have been really, you know, accepting and accommodating. think it's, it's worked out really well, although I've kind of felt bad trying to get people to schedule it at the right time for me.

Bennett Bernard (05:28)
Yeah, I think, in a weird way, I think being busy is, know, you could say that makes it, you seem like you're not as desperate, like, I have any time. Just please book me anytime to, to me. So, you know, it's good to be busy, I guess, in a sense. Yeah. How do you feel like doing the podcast has, you know, cause obviously you're doing a lot of these interviews, presumably on zoom or on a virtual call. And one thing I've found just by doing this for the

Bradley Bernard (05:38)
Mm -hmm

Yeah.

Mm

Bennett Bernard (05:55)
other six episodes that we've done is I feel better at speaking on camera and on zoom, just in my day to day like work and life. And have you felt that too, like when you've been interviewing, does it feel like it's, you know, maybe last time around when you were interviewing, maybe we weren't in COVID era. So you probably didn't have the same zoom call experience that you're having now, but certainly do you feel that same kind of, you know, we're like, you're feel more comfortable being on camera and like giving

Bradley Bernard (06:03)
Mm -hmm. Mm -hmm.

Bennett Bernard (06:25)
responses to questions because the pod or do you think they're unrelated?

Bradley Bernard (06:28)
I think I definitely felt more comfortable. I think it was a bigger difference for me because, you know, I was working for myself, didn't really have meetings. Like I talked to folks online on Twitter, you know, on text or email had, I don't know, maybe five or 10 like meetings with various companies or people that I wanted to talk to. But more or less, a lot of my interaction was async and then doing the video interviews specifically starting with the podcast. I feel like it definitely helped.

you know, getting comfortable understanding, you know, how to present yourself, especially getting the lighting, the camera, all that set up. Like, I feel like there's a confidence in having the right equipment and then practicing speaking while trying to make sure you don't have lots of these filler words. You have like clear communication. Definitely think it helped. But again, I prepare a lot for these interviews. Once I get down to it, sometimes things just fly out the window, like both the words that I speak and also the technical kind of translation of what I'm thinking. Sometimes

Bennett Bernard (06:57)
Mm

Bradley Bernard (07:24)
you know, I'm like, if I had a less pressured situation, I would do twice as good. Or if I wasn't, you know, coding within this 45 minute window, like it would be way easier. But I think the, the environment definitely creates a bit of difficulty, stress, tension. think that, you know, sometimes I feel really prepared that I get there and it's, you know, definitely hit or miss. thankfully a lot have been going well, but it's just like a different environment. Like I've been on the other side too. And being in these interviews, it's never like super comfortable on each side, but

Bennett Bernard (07:44)
Yeah.

Bradley Bernard (07:53)
I think the companies that I've interviewed with this cycle, they've been really, really great, really, really kind people. And I think the people that I vibe with the most are just like the right energy. you log on, you talk them over video, like they're goofy, they're funny, they kind of meet you where your energy is. And those interviews, I'm like, okay, I'm comfortable. Like it started out well. And I have the skill to get through it. And I understand like where their energy is and it feels right. I think the interviews that are a little bit more challenging are the ones where.

you you kind of bring a different energy. They don't really match it and those are like a little bit harder to work through. But you know, I've had plenty of those and they still usually go well. But I think for me, the easiest ones are kind of having the fun, kind, goofy energy. And I think you can really read it from the other side. So the podcast helped to definitely get some of that interaction because if I didn't, I don't think I would have too much kind of video chat face to face.

Bennett Bernard (08:42)
Yeah, yeah, agreed. Well, cool. Let's keep us updated next week or whenever we have another one of these on the final decision. Yeah.

Bradley Bernard (08:47)
Yeah, big news coming soon. Yeah, I'll definitely share it when I have it. I don't have it yet, but you know, getting through it and gonna think about it. But yeah, kind of talking a little bit about Laracon. I was on Twitter. So again, part of this hiatus for us and me, I haven't really been on Twitter too much. I haven't really posted anything on Twitter. I've been very heads down in the interview space to just make sure I'm well prepared and I'm putting my best foot forward and kind of just doing the rounds that I need to do.

I would hate to go through this process being unprepared. So probably spent maybe too much time on it, but again, I'm happy with what we're looking at now. And part of that was just getting off Twitter and not writing that much code anymore, transitioning over. yeah, Laricon is next week. I hopped back on Twitter. The first thing I saw on top of my feed was kind of Aaron Francis saying, Hey, there's a Laricon Telegram group, go join it. And this is where all the cool people hang out. And so I think I joined it last week.

I was kind of reading the chat for an hour, then got overwhelmed by just the massive amount of messages that were flying in. So I muted it, then I forgot about it. And then I just looked at it earlier today and was like, there's like way more people in here. I definitely see names of people that I know, like really excited to get out there. I think I've been so busy that I didn't really realize like it's in like two days. I'm going to be flying out, you know, from California to Dallas, staying in a hotel then on Tuesday and Wednesday is the conference. So.

Yeah, really looking forward to it. I need to talk a little bit more on Twitter probably over the next two days to kind of remind people who I am. I think when I started my indie journey, I was like very public, kind of networking with lots of Laravel folks and I really enjoyed that. And then the past few months been a little bit more quiet on my end. So people might forget about me. So definitely need to do a bit more networking pre -event so that when I get there, know, people like, who is this guy? You know, I've seen him a little bit before. I think I saw a tweet from Taylor Otwell saying,

If you come up to me at Laricon, like show me your profile picture on Twitter, therefore I'll know who you are. And I really love that one. Cause I think as a, as a follower and as an interactor, you know, you're interacting with someone who maybe has hundreds of thousands of followers and you feel like maybe they respond to you from time to time, but you know, they have way more interactions than you do. And so you might feel like, know, them, you approach him at the conference. Like, you who is this guy? I don't really know who you are. And so it's like, show me your profile picture. That's like your, your ticket to entry. So I really love that.

Bennett Bernard (10:36)
Mm -hmm.

Yeah.

Yeah.

Yeah, that's a good idea. Yeah. Is there any particular talk that I guess have they released the like full schedule of talks and is there any ones that you're particularly excited for?

Bradley Bernard (11:12)
think they have the Laravel team has really been cooking. They hired like eight people, maybe eight's the wrong number, but what feels like a lot, they started with a small team. I think over the past six to eight months, they've been just hiring like mad, like content folks, engineering folks, support. They built out more of like a company for Laravel as far as I can tell from the outside. And with that, a ton of products are being released at Laricon. So I'm really excited. think.

Not every Laricon has a big release. Usually there's like one thing, maybe two. And from what I've heard, at least for this Laricon, Taylor has been tweeting out a bunch of stuff. He'll put maybe like two characters out there and put asterisks for the rest of them. And so he has maybe a list of five or eight things that they're gonna release. And I think partly it's due to good timing. I think sometimes they release things a few months before just because they don't wanna wait. And then partly it's because of all these new folks that have joined that I think are putting in the effort.

having that team synergy, kind of the Laravel brand is coming together. So I haven't looked too much at the talks. Taylor gives one. I think Caleb's giving one. Caleb pours the on flux. I think it's like a live wire data management system. I listened to Aaron and Ian's podcast and he was talking about pricing for that. And I guess it used to be free, but now it's paid. But I think those two are the ones I'm looking forward to the most. Haven't really looked at the schedule. I'll be honest. I think I did.

when it came out, but I think things have changed. So I'll probably take a look at it this weekend and see. But last time I was there, I pretty much attended every single talk, like didn't really step out for anything. So I just kind of try to soak it in as much as I can.

Bennett Bernard (12:42)
Yeah. Yeah. Well, it's going to be hot out there in Dallas. stay inside. Stay inside and appreciate the air conditioning. That's cool.

Bradley Bernard (12:46)
I know.

Yeah, I think it's like 800 people in an auditorium in Dallas and downtown Dallas. So it'll probably be hot outside. Hopefully the AC is blasted in the inside.

Bennett Bernard (13:00)
Yeah, has to be. Yeah. As a Phoenix resident, I will tell you all about the heat and, you know, to stay inside basically. Although I think in Dallas, they've had it especially bad because there's been, you know, they're more humid than Phoenix. you know, we have a friend that lives in Houston, which is obviously further south and closer to the Gulf than Dallas. But he was telling me yesterday or the day before that it was like a hundred and five with like a dew point of 80.

Bradley Bernard (13:14)
Mm

Bennett Bernard (13:28)
And that just sounds horrible. So I was like, yeah, that, you know, the humidity factor being that high is just atrocious.

Bradley Bernard (13:28)
Ouch.

Yeah, not looking forward to that. think what people told me is pack shorts and t -shirts, but also bring one extra layer because the AC is so rampant in that area that like you might enter, you know, warm, but then actually be cold there. So maybe I'll bring my backpack and like just bring like a sweater or something. Cause I love having my laptop out, just able to take notes and things while I'm watching the talks.

Bennett Bernard (13:44)
Mm

Yeah.

Yeah. And if people look closely before I move off of Laricon, if people look closely behind Brad, his badge from last year, think. So yeah, that's it, right? In Nashville. Yeah.

Bradley Bernard (14:07)
Yeah, it's right there. Yeah. It was national. This would be my third time actually. So I went with our dad, I think in 2016 in Kentucky, didn't really talk to too many people. I'll be honest. I was in the college days, kind of, you know, hung out with, me and my dad talked to maybe a few people, pretty, pretty in and out. And then last year definitely tried socializing a lot. That worked out pretty well. I got to talk to a lot of Laravel people, that were up there and kind of just known in the community. And so.

I think this year going into it, my probably top two goals is one, kind of reconnect with the folks that I talked to last year just to say, hey, I'm alive and I'm still kicking. Two, probably expand to meet a lot of the new Laravel people. Again, they've hired more folks, the community has expanded. And I think there's definitely a larger audience in terms of coming to the conference and larger speaker set. So I think last Laracan was very PHP, Laravel specific. This year they're bringing more people that are

kind of like tech influencers or like adjacent to Laravel or software engineering. It's a little bit more general, which, you know, have to see how that ends up being in terms of the conference talks. think it's still mostly Laravel, but they invited Primogen. I'm not sure if you've heard of him, but kind of like a tech influencer. And I think, you know, all the other Laracons, you know, he might not have been invited or like wanted to come, but I think Laravel is growing and I think it's gotten more.

Bennett Bernard (15:17)
Yeah, yeah, yeah.

Bradley Bernard (15:27)
attention and attraction and especially within the Twitter community like levels .io pushes php pretty hard and Laravel is kind of the easier version on top of that. yeah, I'll have to see I'm very excited to attend talk to people and just see what's going on out in the Laravel world. I feel like I've been a little bit unplugged the past few months and so I'm very curious to see what people are working on. I think the Laravel releases from the Laravel core team. Hopefully there's gonna be product that I pick up and buy and use and it makes my life easier. That's that's what I'm looking forward to.

Bennett Bernard (15:56)
Yeah, that'd be cool. Yeah. You'll have a good time. Awesome. So one thing that you kind of mentioned as it was like a natural segue into what I wanted to talk to you about today was when you went in 2016, you said it didn't really talk to any folks and you were kind of just, it was your first time there at a conference, which I think is totally understandable. And like you said, you were younger and now you weren't the professional that you are now. And for me, it's been interesting as I've grown in my career.

Bradley Bernard (15:58)
Yeah.

Mm

Bennett Bernard (16:23)
You think it'd be the opposite, but as I've grown in my career, I've realized that having like a strong, you know, community strong connections and specifically like a mentor or like coach in a sense can be super valuable. Whereas when I first started my career, like, and I think this was a drawback to myself, I'm just being, you know, self -reflective, but like, you couldn't tell me nothing, which is crazy because you don't know, you don't know what you're doing when you get out of school, but like you think that you do. And, you know,

Bradley Bernard (16:46)
You

-huh.

Bennett Bernard (16:51)
something I've again, as I've grown in my career, been able to see the value in like from firsthand experience in terms of like getting mentorship and kind of coaching, you know, in my role and seeking those opportunities outside of my kind of day to day work as well. But, you know, also just how beneficial it can be for someone for you to kind of coach someone or give advice, because, know, it's that old saying where like teaching something to someone, you really have to know it yourself. So, you know, that's something that for me has been

Bradley Bernard (17:05)
Mm

Mm

Bennett Bernard (17:21)
kind of resonating pretty true as of late. And I'm curious, you, know, software engineering is obviously much different from accounting. So I'm curious if you see something similar, if it hasn't been as, you know, firsthand from your perspective and just your general thoughts on, you know, having like a, I don't want to say intimate cause that's probably not the right word, but like having like a formal relationship of like, this person, like I'm learning from, and like, they know that like,

Bradley Bernard (17:47)
Mm

Bennett Bernard (17:49)
they are here to learn from me and so on. Do you have any experience with something like that?

Bradley Bernard (17:51)
Mm -hmm. Yeah, it sounds like this kind of feels like an interview question I've taken over the past two weeks of, hey Bradley, what's it? What's your mentorship experience? And I won't bore folks with the story I have prepared for that, but I guess diving into it a little bit deeper, starting my career, I think being like fresh out of college or, you know, a new person on the job, I think it's really important to get with, you know, understand the leaders, be close to them.

Bennett Bernard (17:58)
There you go, you'll be ready to go.

Bradley Bernard (18:18)
kind of suck up that domain or tribal knowledge and understand how things are done. When I had joined PayPal as a new college grad, I think you don't really understand who might be your official mentor, but you just kind of talk to people and say, like, you know, I'm talking to this person. I find, you know, all their comments or feedback is really helpful. And I just want to get closer to that. So I think, you know, my time at PayPal, I didn't really have a formal mentor or a coach, so to speak.

But I definitely had a great set of people that I could lean on bounce ideas off of, and really they would go out of their way to help me. So I don't think I would coin it as an official mentorship, but really early on in my career, it's how do you get fast track to success? Everyone wants to join a company, put their best foot forward, really onboard onto the systems, be productive and efficient and deliver value really quickly. And I think that was my goal when I joined PayPal. I didn't really have a formal mentorship, but I knew that

These people were helping me get farther in that direction. And then kind of fast forwarding to LinkedIn, the same thing happened. I joined the company, you know, eager, hungry, ready to do some cool stuff. And it was, who can I talk to to help me on board level up faster and get things going? And again, I think when I started LinkedIn, definitely had a handful of folks who were on my side, willing to help out over chat meetings, et cetera. There was actually a time at LinkedIn where I had worked with someone so much that I'd reached out to my manager saying,

you know, I think actually this would be good as a formal mentorship. They weren't really on my team. They're a little bit, you know, out of the way, but I had worked with them on a decent amount of projects. And I realized this person's pretty senior, helped me out a lot and went out of their way really to like give me the advice and guidance that I didn't even know I needed. And then once he had given it to me, I said, you know, this is amazing, fantastic. Like I'd like to continue this even though we're not working together as much, like I would appreciate your feedback. So I think it really depends on

what you're looking for in the job, like what leadership is there. There can be companies that are really busy and don't prioritize coaching and mentorship, but that's kind of on the receiving side. Then I guess fast -boarding to now, me mentoring folks at the end of my tenure at LinkedIn, kind of new hires coming into the organization. And then also Metta, just understanding what's come before me, putting myself in their shoes and trying to deliver the value that I see for a new hire, for a new, you

engineer coming out of college saying, you know, I'm joining a big company like Metta. How do I communicate? How do I manage projects? How do I deliver like technical excellence and quality? There's so many different facets that when I mentored folks, it was like half the technical side and then half kind of the people managing project management, et cetera. And I think seeing that growth in people and meeting them where they are, because people will communicate differently, have different career goals, life goals.

Bennett Bernard (20:59)
Mm

Bradley Bernard (21:00)
If you can understand what they want and you can provide kind of the stories, lessons and values that, you know, maybe I made a mistake in my career and I can share that like quick learning to them or they're running into a, you know, a harder issue at Metta and I'm able to provide, you know, what my best advice would be in like real time. And so I really value those and found those to be really fruitful where I think when I first started my career, I was not, you know, concerned with mentoring anybody because I didn't have the skills or know -how or

Bennett Bernard (21:10)
Yeah.

Bradley Bernard (21:26)
experience and past failures really to do that. And then now as I see things, it's how can I level up others? How can I spread my impact? Like, you know, not only my team, but my organization. And it's really rewarding. I like it a lot.

Bennett Bernard (21:27)
Right.

Yeah. Yeah. And, know, there's like the old adage of like, you know, get around people that are smarter than you and like get around people that are more successful than you. And, you know, it's something that I think it's actually probably at least initially harder to do than you might realize. Like, where do I go find successful people, you know, that I've, that I can learn from like outside of my job, you know, hopefully there's some at your job too, but like,

Bradley Bernard (21:45)
Mm

Mm

Bennett Bernard (22:03)
And if you really want to get like a wide range of viewpoints and perspectives, it's, think in most cases for a lot of people, that won't be enough. Like, cause you'll probably have your direct manager, but you might want to get mentorship from, you know, a skip or like two skips above, you know, that, because like, that's like, you're just thinking long -term and like, you know, wanting to really grow your skillset. And also to just people that have kind of been there and like done it, you know what I mean? And like Larik, Larikon, for example, like, you know, you're

Bradley Bernard (22:11)
Mm

Mm

Mm -hmm.

Bennett Bernard (22:31)
going to be interfacing with people that have built some really cool things. And, you know, the same is true for, you know, the mortgage, accounting conferences that like I've gone to where people are working on really cool companies and they're doing some pretty cool things, pretty cool and interesting things. And it's always good to hear those viewpoints, you know, from people that are doing something similar to you, but like not directly within your org there. It's in the same industry, you know, and then there's like a a shared interest and

Bradley Bernard (22:57)
Mm

Bennett Bernard (23:00)
I found that over, you know, again, over the course of my career, having like that strong network, we did an episode about networking, so I won't like dig too much into that, but just having a community is can be super powerful, especially when you're running into common problems and you can come up with like, you know, shared solutions of like every quarter mortgage companies have to file this quarterly like mortgage financial reporting report. And it is just a horrible like

Bradley Bernard (23:27)
Mm

Bennett Bernard (23:30)
tooth pulling process a lot of times. And, you know, but there's probably people that have done some really cool ways to automate and like make that less painful. And, you know, if you're working at a company, you don't have those things. You don't have the ability to kind of go out and ask and kind of like crowdsource like, what's the best way to really deal with this thing? Cause everyone dislikes it. You know, you're missing out on the opportunity to kind of get some of the other knowledge that might be out there that maybe isn't in your org that you can apply to your own skillset. So

You know, just being aware of, you know, where the opportunities to learn from are. They're not always just in your day to day job yet. They kind of go out and seek them and, go to conferences, join communities, be active on socials, you know, all those things are ways to ultimately learn and take in more information. That's where some of the best, I've gotten great mentorship in my career and in my, kind of my regular, you know, nine to five jobs, but then also to some of my biggest learnings have come just from more informal.

Bradley Bernard (24:06)
Mm

Bennett Bernard (24:26)
interactions that either I sought out or, someone, you know, we kind of just mutually kind of met, met in the middle and, know, kind of spoke and, you know, I just ended up taking a lot away from that in terms of just learning. So it's pretty impressive how once you know what you don't know or know how much you don't know, you kind of open yourself up to learning a bit more when if you think you know everything, surely you don't. And I think it limits your ability to like hear, you know, new viewpoints and hear information, you know.

Bradley Bernard (24:44)
Mm -hmm.

Yeah, I like the way you put that. Cause I think, you know, lots of folks go on this learning curve of I don't know anything. Then you feel like you're leveling up and you feel like, you know, I'm kind of at that peak, like, I know how to do things really efficiently. I think you'd come back down pretty quickly on the other side of that thinking, I really don't know that much. Like there's so much that I don't know. And I think having the like naivety to realize that and go through, you know, kind of this up and down phase really brings you back to reality that

You know, there's so much that goes on in the world. There's so much to learn that you really should always keep an open mind and any thought that you have that I've like completed everything. And I know it all is not the best thought. It's like not going to get you to where you want to be. think having that open mind and kind of continuous learning is definitely where you want to be. And when I, when I talk about mentoring, I think the best mentors that I had were folks that are really close to what I was working on. know.

people talk about, like go reach out to these mentor programs or go try to do this or that. I haven't really found a ton of value in those systems only because I feel like they're kind of giving offhand advice in the general sense. I've always found high value in folks that were kind of plugged into my work, understood me as a person, understood my career goals, like understood the company that I was working at. Cause lots of these times, you know, we're in the same company and really

Outlining opportunities outlining areas of the tech stack that I could invest more time in or own and then translating that to fruition. So yeah, I think when talking about mentorships, it can become, you know, more or less straightforward to go reach out to some mentorship network or some coaching website to say, Hey, I'm a software engineer. I'm an accountant. I want to learn more about this, but I really, you know, you have to have consistent meetings. You have to know them as a person. If you invest that time, I think it's worth it, but

I would definitely take any advice that you get from these folks, if it's like an offhand meeting or a one -off thing, definitely with, you know, caution, because it's really easy to give advice. I think it's hard to give the right advice at the right level at the right time, knowing who that person is. You know, I can sit here and tell you, you know, 50 different stories, but you know, what's important to you? Why should I tell you that? And why do you need it? I think those are the mentorships and also like the mentees that I've really valued. When I try to get feedback,

I make it based on, you know, what they're working on, what they're striving towards. It's not really a time for me to just dump out, you know, life lessons and them to kind of index that it's more. You know, giving them the right thing at the right time. Cause so much information is flowing, you know, at these companies at all times, especially as a new hire, you're kind of firehose with information. If there's anything I can help you with, it's, it's distilling that summarizing that giving you the key points that you can move fast, not really be tripped up by, you know, information that I might give you that you kind of sit there and go.

Okay, that might be useful, but I'm not sure when I'd really use that. And so I think over time, I've really tried hard to make my advice, you know, storytelling, mistake sharing really pointed so that people could use it, you know, that day, that week. And it's not something that I'm really just beating around the bush, giving random advice to folks if they are looking for a mentorship. So really finding that, you know, what I want and what I can give and what they want and meeting right in the middle and delivering like, you know, very timely advice.

Bennett Bernard (27:44)
Mm

Yeah, no, I like a lot of those points. I mean, one thing that I was jotting down as you're talking is, you know, not just giving out random advice, you know, just that doesn't apply to anybody that doesn't do anyone any good. I think a lot of times are you just giving that advice because you want to be seen as a certain, you know, like, I'm a thought leader and I'm trying to give out advice, you know, and there's a quote, I think it's by, by Gandhi that basically says like something to the effect. I'm going to butcher the actual quote, but

Bradley Bernard (28:25)
Mm -hmm. Yeah.

Bennett Bernard (28:37)
some of the effect of like, you know, only speak if it improves upon the silence, you know? And so like, if you're going to say something that doesn't really add any value, that doesn't really like change anyone's perspective, you know, why, why say it at all? You know, he wasn't using it in that context of course, but just where my head went. And one thing that's really good about, I think, mentorships too, and kind of, you know, coaching in a way is, and this is a little bit harder, I think, to set up because it can be

Bradley Bernard (28:50)
Mm

Bennett Bernard (29:07)
I think this kind of thing is sensitive, but like accountability, like certain mentorships and coaching, like if you have someone who you're looking to and you've given them goals and this again could be in your regular job or outside of your normal job and you said, Hey, I wanted to go out and achieve these things. A lot of people can say, okay, great. Yeah. Like, let me know how you do. I'm cheering you on. Like, that's awesome. And then like next quarter you go, I just been too busy and like blah, blah, whatever.

Bradley Bernard (29:11)
Yeah, yeah.

Mm

Yeah.

Bennett Bernard (29:35)
And they're like, okay. Yeah, like that. Totally get it. You know, but then when people are really, you know, able to coach and give you mentorship is when they kind of hold you accountable. It's like, well, really, why did you not get that done? And kind of like, you need to kind of let your guard down, I think, in those kinds of scenarios and be like, I really have a lot to learn from this person. And so I'm not going to get defensive if this person's like, you know, not not they're going to call you out and say you were lazy, but just really kind of question like, okay, why really did you not get that done? Like you're saying you're too busy.

Bradley Bernard (29:44)
Mm

Mm -hmm.

Mm

Bennett Bernard (30:04)
Like how did that happen? I feel like that takes, you know, your, your growth to another level. If you're able to get that from that kind of accountability from someone who's been there and it's done that and says, Hey, I did this, you know, this is what worked for me. I see you want to do something similar or you want to chart your own path to success. And you say you want to do it. And I'm going tell you right now that like,

saying it and doing our different things and like you need someone to kind of keep you on it sometimes that's just that's just the way that you're to be able to get it done so you know i've found that in relationships i've had where the person was able to kind of keep me accountable whether just through you know job titles or just again if someone i was really looking up to and i was like shoot if they say that i need to really get that together then i need to get it together that really makes a big difference too because

Bradley Bernard (30:29)
Mm

Mm

Bennett Bernard (30:54)
you know, everyone can kind of let themselves off the hook a little bit. And sometimes you need someone to kind of say, Hey, you, nah, don't make excuses.

Bradley Bernard (30:57)
Yeah.

Yeah. And, you know, being, like being the mentor, it's definitely a little bit harder to give that feedback as well. Where, you know, sometimes you have to give more critical feedback, things that aren't as easy to deliver. And that's building that trust up over time, understanding how they communicate, how they'd like to receive feedback. It's really easy to give positive feedback, say, you Hey Ben, great job. You landed this project. You did everything we talked about.

I think it's a little bit harder to say, you know, Hey, Ben, maybe this should be a little bit faster. Or maybe this organization could be a little bit better communication, et cetera. And I think those moments are where it's, have to be very mindful and thoughtful about how you deliver it and then understand how they receive it. because again, it's building up that trust, that relationship over time. You really want to make sure that the communication line is open on both sides. there can be times where, you know, you're mentoring somebody and they say, Hey, like, this is how I'd like to receive feedback. Like maybe you gave it one way and they don't.

really vibe with that. And so really being open on both sides, it should be very open, honest, that channel really shouldn't be filtered in ways that people aren't getting the highest value out of it. found the best working relationships I've had with folks, whether that be product, design, even mentorships, it's just really being honest and constructive in ways that we're all trying to achieve the same goal. We're all trying to build a better product in the engineering space or better code base.

And getting that feedback across has a point. I'm not just saying, you know, hey, we could have done this better for the sake of doing it. It's like, Hey, I think this was a meaningful impact. think highlighting this will make you realize it. And then you'll take that feedback and learn from it. You know, everybody makes mistakes. I've made plenty. I think outlining that describing, you know, what might be a better solution to it, getting them to agree and understand and, you know, kind of talk through that thought process. That's where you have these giant level ups where people are, you know,

I didn't know this wasn't handled as well, or I didn't know this could be better. And it can be hard to give that feedback. It can be very challenging. And I think you have to be as a mentor, like honest with yourself, honest with them. I really think deeply about how you deliver it, but I've had a few of those and not super fun, but I think once you start out with that or like as you get through that as a mentor, it just really opens up new doors. I think people really value you more. Cause again, it's really easy to give advice. It's really easy to get positive feedback, but I think

Bennett Bernard (32:58)
Yeah.

Bradley Bernard (33:20)
diving a little deeper in the criticism, with it being constructive and the right delivery, I think that's where things really shine. Again, if all things go well, you don't need to do it. But there are times where you'll push folks to say, take this on or take that on. And sometimes there could be opportunities for just improvements and making sure you deliver that feedback the right way.

Bennett Bernard (33:39)
Yeah. And one thing that, you know, I've, think it's also a challenge too, and just things I've encountered and encountered in my career, but like something I try to wholly welcome in my own circumstances is never being above like inspection and like questioning. Like as an example, like we have this process, it works, you know, it's been around for three years. Don't change it. Like it works.

Bradley Bernard (33:56)
Mm

Bennett Bernard (34:05)
And I'm, you know, I understand that mindset. Certainly like, let's not create work for ourselves. That doesn't need to be there, but like, it's not above questioning, like why we do it the way we do it. So, you know, it's a process. It's not a human being. Like don't get offended. If I say like, Hey, why is it working like this? Could it be better if it was done this way? You'd be surprised at how much pushback that kind of gets when like, you're just asking questions. Like you're at the end of the day.

Bradley Bernard (34:11)
Mm

Mm

Bennett Bernard (34:32)
I think everyone for the most part should want the same thing. Like, let's do this the best way possible, but people really attach part of their pride or their identity and like, this is the way we do it. And, you know, like, I don't want to be, I think the message or the feeling that they might get sometimes is like, if you change this, like you must think that the way I had to set it up was not their best way. And, you know, therefore I feel slighted by that. And, you know, for me, if someone questions, questions something I'm doing,

Bradley Bernard (34:37)
Mm

Mm

Bennett Bernard (35:01)
You know, and it's just like, trying to make it better. Like totally like, yeah, this is why I think we should do it. If they think there's a better way to do it, like, okay, yeah, I'm happy to hear it. Like, let's figure it out. You know, that takes, you know, I'm certainly not perfect. So let me just, I guess, clarify that, but you know, I think it takes, you know, a bit of divorcing yourself from the process to feel like, Hey, this thing is, you know, even though I do this, it probably can be done better. And I I'm totally welcome to someone showing me how it can be done better.

But ultimately, if that's your end goal, to try and make this thing that you're doing the best it can possibly be, don't be afraid to accept help. Don't be afraid to accept that feedback. But again, just in my experience, that can be hard for people sometimes to take feedback that way. They almost feel like it's a personal thing when it's not personal. Nothing is above inspection. You know what I mean?

Bradley Bernard (35:39)
Mm -hmm.

Yeah, I feel like I found a lot of value being deeply curious about things, asking questions, maybe what seemed like dumb questions, especially as joining a company where all these choices have been made, but why is that the case? Is there anything we can do to make that better? And why are we living with what we're living with? And I think that's gone a long way. I've made some pretty large changes at companies just by questioning kind of the status quo. Lots of times people will say, that's just...

You know how things are done, you know, that's just how things are. And to me, especially as a curious engineer, I'm like, okay, you know, I hear that, but maybe I'll, you know, spend a little bit of, 10%, 20 % of my time to understand why that's the case and see if we can build something better. And sometimes there is something better and I'm, and I'm happy to dive into that, but yeah, being deeply curious and then understanding there's flexibility and processes. We're adding something like getting feedback, checking in with folks, make sure it still works. So then if people.

challenge thoughts, ideas, like that should always be open and done in the right way. Again, like you can challenge something, providing suggestion, or you can challenge something saying, you know, I don't like it. Those are two very different approaches and understanding the audience and you know, how to deliver that is super critical because no one wants to be yelled at saying this process doesn't work, but it's more like, hey, I found an inefficiency. Here's something we could do about it. It's kind of like, I think in my career,

I used to be the person that was able to point out problems. And then I quickly transitioned and realized that's not really the best way. You should be able to point out a problem and describe it in a succinct way and then provide a solution because it's really easy to say that sucks, that sucks. Or like, that's not easy to work with or this system's complicated. It's, you know, I understand the system's complicated. It was built, you know, five years ago with this different business requirements.

Bennett Bernard (37:28)
Mm

Bradley Bernard (37:39)
I think we could do better and here's how we could do better because it's really easy to speak into the void and complain about things, but it's a next level to dive into it deeply, understand why it's there, what constraints it had, and then provide a better solution on top of that. I think I've, as a mentor, really pushed that hard on folks to say, find the things that suck and dive deeply into them, but don't just raise the flag saying they suck because people probably already know that. Raise a flag and do your research and then find a better solution or if not,

you know, talk to the folks and see if they're thinking about anything differently. It can be a good way to spark change and start conversations that think people would really ignore if they didn't dive deeper into it. So yeah, I really love doing that.

Bennett Bernard (38:20)
Yeah. What you said about it's easy to point out problems. know, pessimism is so seductive in a way because it's you feel like you sometimes it makes you feel like you're smart or you feel like you're like seeing things in a way that's like quote unquote like realistic. But I found and this is a more of a more recent thing. I'll just be fully transparent. I found that like optimism can be like a real

Bradley Bernard (38:27)
Mm -hmm.

Mm -hmm.

Bennett Bernard (38:49)
a real big difference maker and not that you're pie in the sky and just, you know, idea guy and, you know, not thinking about the important things like actually operate, but, you know, just having a sense of like, can get this done. Like, this is going to be awesome once we finish it. Like it's rare than I feel that people would think if you kind of just said it out loud, but just working, you know, in, in America for the last, you know, 10 plus years, you encounter pessimism, pessimism a lot. And I have been guilty of that too.

And certainly there's been times where, you know, some things happened at work or, I felt like, that wasn't the right call. And, know, that's okay to feel like, I didn't like that. But I think, you know, in hindsight, there's probably times where like, I maybe have dwelled on it too much or like focused on like what the issue was that happened and not been like, okay, well, what can I do now to make this better? Cause certainly I have that ability and yeah, it's, but it, it is so, it's so easy to fall into that trap and

Bradley Bernard (39:32)
Mm

Mm

Bennett Bernard (39:49)
You know, again, I, know, everyone, everyone has a venting session. You know, people, people are human and they need to, you know, get things off their chest, but, know, get it off your chest and then look to the next thing and, and, know, realize that like, Hey, we still have lots of opportunity. You know, we could still make whatever we're working on be brilliant. Like that, that optimism, you know, something that I'm trying to bring more, you know, what's right word? I'm trying to be more.

Bradley Bernard (39:53)
Yeah

Bennett Bernard (40:16)
thoughtful about how I bring that to just, you know, my day -to -day work to things I'm doing outside of work. So it's something that I think can be, you know, underappreciated in a lot of ways.

Bradley Bernard (40:18)
Mm -hmm.

Yeah. And especially in mentorship too, I've been on the side as a mentor, kind of, you know, having that venting session, me, my mentee, et cetera, kind of diving deep into the, maybe the pessimism just to kind of let things go, but then really bringing it back with like the optimism of, yeah, maybe this isn't great. And like, we can both agree on that, but you know, how do you look at it with glass half full? I think, you know,

being a mentor is understanding how to help them, but also to be there for them. That's kind of like the people manager side. Again, I'm not a people manager, I'm an IC, so doing things myself, contributing to the code base and building out systems, but mentorship was kind of my first foray into reading people, understanding what they need and being there for them when, you know, sometimes a great manager is someone who listens, who cares about you and is really able to set you up for success. Might not be super tied to the work that you're doing, but is really there to support you emotionally and mentally.

I think being pessimistic can really bring you down. It can bring others down on the team, like making sure you're not diving too deep into that. I think it's good to have a dose of reality, but at the end of the day, if all your coworkers are saying, all the systems suck and like, we gotta get out of here, like no one's gonna have a good time. So I think it's maybe good to vent once in a while. I think it really helps connect folks. have been times where you kind of slug through a difficult project or time and you say, oof, I'm glad that's over. And everyone kind of...

size and agrees with you. then moving forward, it's, you know, putting that aside and being optimistic, making sure that, you know, you're working on fun stuff. You're talking with cool people. Sometimes it's easy to lose sight of what's good and what's great. It's really easy to focus on what's bad. So yeah, I've definitely been there and had my venting sesh, you know, on both sides. And I think it really helps. It brings people together and, you know, it's great to clear the air once in a while because there can be lots of, you know, internal drama, politics, et cetera, that goes on.

for you to really address that and not hide behind it. I think it's really human -like of the mentorship experience. So, found lots of value in that for sure.

Bennett Bernard (42:26)
Yeah. And you know, playing like this politics everywhere, of course, that's just human nature is politics in a way. But you know, trying to avoid that and being slightly immune to politics in the sense that it doesn't affect you, found is like, just the best way to get through a sticky situation where, you know, someone's getting promoted when they shouldn't, they should be you or, you know, they're doing whatever they're doing to try and make things better. That doesn't really make much sense. Like whatever the issue is, like

Bradley Bernard (42:32)
Mm

Mm

Bennett Bernard (42:54)
Just be immune to it. okay, like play in the sandbox, you know, all you want. Like I'm going to put quality in my work no matter what I do. I'm to put pride in my work no matter what I do. You know, this, when I, you know, Bennett Bernard put my name on something, like this is what it represents. Like focus on those things. The politics will work itself out, you know, and if it's a situation that you, you know, need to exit from or whatever, then you can always make those, those calls. know, it's nothing, nothing is unfixable in a lot of cases. So.

Bradley Bernard (43:01)
Mm

Mm

Yeah. Integrity and quality, think is kind of the mantra of, you know, do your best. continue to do that. Make sure you're putting in the effort, especially when going through these interviews. always tell myself I've prepared, you know, I've done the best that I can. I'm to go into today, even if I stumble on one module. So when I go through these onsights, there's five back to back. I finish up like we're talking or a video today, the next person will join in literally as I'm talking to you and I'm kind of sitting there wrapping up with them. So.

Bennett Bernard (43:26)
Yeah.

Bradley Bernard (43:50)
not a ton of breaks. And if that session didn't go as well as I wanted, it's like shrugging that off and moving forward. And so I think it's being resilient in how you communicate, understanding that sometimes things happen, but pushing forward really quickly and making sure that you're able to kind of drop what was negative or maybe if it was political, just like moving forward in the way that makes the most sense and you're not like dwelling in the past too much. So yeah.

I feel like I've been trying to be very prepared, but also just doing as best you can. And that's like when I walk away from interviews after that, it's, you know, I did what I did and I'm happy about it. And that's all I can ask for.

Bennett Bernard (44:26)
Mm

Yeah. I've told my wife before in interviews, fortunately, I haven't had to really interview in a while, but, you know, I said, I am happy with what I provided and like what I shared and the information I conveyed, how I presented myself and like, if I get it great, I'd be excited for that. If not, I can still rest easy. No, I gave it my all and they felt like someone else just, they're all was better than mine. And that's, that's okay. You know, wishing the best, right? Yeah. Cool.

Bradley Bernard (44:33)
Mm

Mm

Mm

Exactly.

Yeah. Yeah, I agree.

Bennett Bernard (44:58)
Well, since we've had our hiatus, I've been cooking, Brad. I've been cooking on just a lot of stuff. And this is going to go just a little off script here, but we're going to just wing it, all right?

Bradley Bernard (45:04)
Tell me, tell me.

Yeah

I'm in, I'm in. think just to provide a little bit of background of folks, Ben's been texting me about a few things and yeah, I've been really lackluster in responding. So this is the first time I'm going to hear it. I'm ready.

Bennett Bernard (45:22)
You were just saving it for the pod. That's all you were doing. Smart. That was a smart move. So a couple of things. So I'll start with, I've been playing a lot around with a cursor, which is something that you put me onto a while back, a couple of episodes ago, playing around with cursor and with, anthropics, Claude. So I thought I canceled my chat GPT. I thought I did, but literally today I got a little notification from my credit card. I know, I know. I got a notification from my credit card saying, chat GPT. was like,

Bradley Bernard (45:23)
You know me.

Mm -hmm

Mm

you gotta cancel that.

Bennett Bernard (45:50)
So I swear I canceled that, but I yeah, so I got both right now. So, you know, whatever. but I've been really enjoying Claude, first of all. So like it's great. I've been, we'll get into what it's been doing for me, but just great experience. Very intelligent, very fast. and cursor, you know, I signed up for Anthropics, like, I don't know what it's called. where you can use, put in your key to the cursor. Yeah.

Bradley Bernard (45:51)
And you've paid for Claude, right? OK, nice, nice.

Mm

the workbench, the API.

Bennett Bernard (46:17)
Yeah. So did that with cursor and like literally just type in what I want. So I've been using fast API, which has been pretty fun. it's a Python based framework. Yeah.

Bradley Bernard (46:26)
Okay, for the people who, okay, yeah, I was gonna ask for the people who aren't aware, you know, we're, we're Laravel, we're PHP people, give us an intro into Fast API.

Bennett Bernard (46:34)
You're a layer, bro, not me. I try, we'll get into that too. But the reason I want a fast API is because we've been talking a lot about like, you know, LLMs and AI and a lot of the libraries around that topic are, they have a lot of Python in there. or a lot of them have like Python based, libraries. So there's just been a natural fit. So, just, yeah, just typing, Hey, you know, fast API, I want, give me like an index page, get me an about page.

Bradley Bernard (46:36)
Hahaha

Mm

Mm

Bennett Bernard (47:04)
and have it be this kind of color theme, which is always my struggle is I could not interior design a pillow, dude. Like it's nothing. So I'm like, give me give me some cool tailwind CSS classes that make it look nice and modern. And it literally, you know, I'll show you after the after the podcast, like it literally just looks great for like a prototype, you know, for just, you know, something I did today. So I started this part today with cursor.

Bradley Bernard (47:10)
It's okay, takes a while.

Mm -hmm. Mm -hmm.

Mm

Bennett Bernard (47:32)
And just got it up and running and yeah, it runs. can like start my, you know, a fast API server. It's just, it's incredible. So really been enjoying that for those that don't know about cursor and with the Anthropic or whatever system you use, I think it can use multiple. can use chat GPT. You can use Anthropic. it charges you. Like you load up your tokens basically, or you load up a balance with that provider and then like.

Bradley Bernard (47:32)
Nice.

Mm

Mm -hmm.

Bennett Bernard (47:59)
As you request things, it's been, it decrements your balance. So I was like, kind of like not sure like, well, how much do I use, you know, in my little like just requests here and there. And I was like nervous and like, was like coding for like two hours and just like trying things out. And I went and looked and it was like 30 cents. Like it's incredible how much value you get. Yeah. Especially for someone that's not technical like me, you know, like for you in your case, you probably don't, wouldn't need to use it as much if you're just building, you know, a layer of all site you've done that before.

Bradley Bernard (48:07)
Yeah, yeah.

It is so nice.

Mm

Bennett Bernard (48:27)
You know how it all works. You spent time in the docs, you know, for me being a non -technical person that still has an interest in building like technical products, it's just a perfect fit. Cause I can just say, you know, Hey, give me these things and it'll get me like 80 % there. And sometimes a hundred percent, sometimes a hundred percent. Sometimes I don't even touch it, but like in the parts where I would always struggle with the most was like just the HTML, like the HTML code, there needs to be like an industry get together.

Bradley Bernard (48:30)
Mm

Yeah.

Bennett Bernard (48:55)
Because if you look at those boot camp and or not boot camp bootstrap. Yeah, we're just tailwind like does the class names on these elements don't make any sense to the horrible horrible and so Yeah, exactly. So I've really enjoyed that way. I don't have to write I'm gonna look at my code really fast and on the right for my class Div class flex H full transition transform duration 300 ease in out style transform translate X

Bradley Bernard (48:59)
W3

Yeah.

AI is built for you.

That's the beauty of tailwind, you can't hate on it.

Bennett Bernard (49:24)
I am impressed by what they do but the classes are absolute top half left four transform translate y one half bg cool blue hover bg cool green text white p3 rounded full transition duration like dude

Bradley Bernard (49:29)
Yeah.

Yeah, it is a lot. And I think describing cursor a little bit more, there is the first invention of GitHub co -pilot. So you can imagine a text editor. I'm writing code to maybe, I don't know, create a user in my backend system. And as I'm typing, it's filling out the rest of my current line or maybe my current function. As we transition to cursor, cursor is built on top of Visual Studio code. And it has a different experience where you can actually type in

natural language text so I can say, create me a button in HTML with a red font. And when I click on it, it maybe adds a new to do item on my to do app. And so you can literally describe the feature, the button, the UI, like this whole system cursor will take a look at it and generate you code. And I think the latest feature that they came out with, I'm not sure if you played with this yet, but they have a composer mode. I think that's what it's called. Essentially you can give it a prompt so I can describe the same thing of

Create a to -do list app. This to -do list app should have a button on the bottom to create a to -do. You can type in text. You can check a to -do to be complete. You can delete a to -do. You can sort to -dos by some system, essentially describing all the user level requirements. Then it'll go to town. And this is special in the Composer mode because it actually creates a sequence of files. So usually in the chat mode, you're within one file. You've selected something. You've typed in a query to say, improve this code, create a button here.

Composer is kind of like a blank slate. You provide that query and it's going to scaffold and coordinate across multiple files. And so I think it's the next step in AI where people are really excited about this. You know how to describe things, at least in layman terms of what you want things to look like, especially if you're non -technical. then Composer will do the hard work to make your code base a little bit better. And especially if you're non -technical, it's how do I structure the code? How should it be organized? There's all these little bits that

I think are important, aren't really, you don't know how to describe them. And I think cursor does a good job at making it more approachable, making it more digestible where it's going to do that heavy lifting for you. And you're able to focus on just describing that system and they can go find the changes, make those changes across multiple files. You can just sit there and say, yeah, that looks good to me, except, or, you know, that doesn't look good. Let's edit, you know, these three things. And so I think it's a huge level up in terms of developer productivity within.

like the engineering space for people who are established and people who are entering coding for the first time, like across the board, I think it has a huge advantage and I'm really excited to use it. I have cursor. I've probably tried using it maybe four or five times in the past six months. I think I used it when it first came out when people were really hyped on it. I felt a little clunky. I think recently with the advent of Claude 3 .5, fantastic AI model and then just cursor getting better across the board with

You know, more context, better editing features, lots of people on Twitter and YouTube using it, really pushing it forward. They've gotten funding. They've gotten more engineers working on it. It's like really culminating into a breaking point of everyone needs to use it. And like everyone is everyone. Like if you're a big company, I guess big companies can't use it because AI is, know, they have their own AI, but if you're working in a small company or you're like, you're running your own software company, you have to use cursor because you're going to be 10 times more productive and it's only going to get better.

If you're still using Visual Studio code with GitHub Copilot, it's good. It's just not as good. And like Ben said, the cost of these things is stupidly cheap where you can save so much damn time by typing in what you want and just going. Like prototype, it's nine out of 10. Production code, maybe six out of 10, seven out of 10, but still really damn good. I think it's a fantastic experience.

Bennett Bernard (53:20)
Yeah. And just to give like people some context. like just a couple of like ways that I found it to be useful. I would have, had my like index HTML or had like the styles. said, give me some cool blue, cool green, like night mode looking styles. And they did that. So give me all the tailwind classes and all the horrible HTML that it has to put up with. And then when I would make like an about page, when I make like a terms of service page, I would just basically, I would go to Claude.

Bradley Bernard (53:35)
Mm

Bennett Bernard (53:49)
That's why a thousand better ways to do this, this is what I was doing. I go to cloud on my browser and I'd say, Hey, give me just like the content of like a general, general terms of service page, for blank company that I was like kind of tinkering with. And so I'd get that text from Claude, like just text and I'd go put it into a terms of service dot HTML file in cursor. And then, but it was just like raw texts because it wasn't H one tagged and all that kind of stuff.

Bradley Bernard (54:01)
Mm

Mm

Bennett Bernard (54:14)
And I would just tell cursor using the composer thing. I think that you're talking about. can like at files or at folders. And so I would tell the cursor chat to say, Hey, take this text and just format it in the same style at my Mike format in the same style as index HTML. And I bring over those styles and it just did that perfectly. Didn't have to touch it. I did that for like a couple other pages I made. Whereas like before.

Bradley Bernard (54:18)
Mm

Mm

Mm

Mm

Bennett Bernard (54:43)
I had to think about like, templating, like, you know, blade templating, like, so Python uses a lot of ginger templating. I think about that and like what, and I'd have to do any of that. just literally just, it was a super lazy way to do it, but I just literally would go in there, type that in and then boom, it would be totally, totally good. So that was one, use case that was super beneficial. The other one is, you you'd make a new page. like with fast API, you just define routes. And so

Bradley Bernard (54:55)
Mm

Bennett Bernard (55:10)
You define a route and you just go, I won't give like the code on a podcast, but like you say, Hey, I'll have a about route. What's it taking to my about page? So write a little code for the actual route. And that's like in your main Python file. And then of course you need to have the HTML file that like corresponds to what is going to be rendered on your browser when you go to that route. And then also, there's one other, there's one other piece.

Bradley Bernard (55:32)
Mm

Bennett Bernard (55:37)
forgetting. I'm not that familiar with fast API. This is kind of like my demo, but basically a couple, there's like three things that to do when you add a new route. And what I noticed is if once I had add the HTML, cause I was kind of like my starting point, I just get like the visual there that I like, and then I'm trying to add the routes. Once you add the HTML, so I've add the about .html when I would go to my main .py file where it has all the routes, it would like prompt me to like add

like in cursor, it prompt me like, do you want to make a route for the about page and have it, you know, use that HTML template that you just made? Like, so just hit tab and boom, it was there done. And then if you go back to the HTML, you know, where you would usually have like, you know, href links referring to different pages, it would like automatically update for that new about page. So before I'd have like a, just a hash mark while I was like getting things wireframed and all that kind of stuff.

Bradley Bernard (56:05)
Mm -hmm.

Mm

Mm -hmm.

Bennett Bernard (56:32)
And once I added that about the HTML on all the other pages that I had where it was, you know, a href pound sign, but then it said about in the actual a tag, it knew to change that pound sign into the link to. Yeah. So like, you know, again, there's probably a better way to do that, but, you know, that was something that immediately was this one was like, I was like, wow, that just.

Bradley Bernard (56:42)
Mm

That's awesome.

Yeah.

Bennett Bernard (56:59)
Again, there's templating is probably a better way to do it and all that stuff with Jinjo, but I was just like, let me just spin this up. And then again, I was doing this over like the course of an hour, got a full site up and running, like, you know, haven't done the user model and stuff like that yet, but yeah, it's just super impressive.

Bradley Bernard (57:14)
Yeah, I think that's a new feature that cursor rolled out. They had the chat, they have composer, which is like multi file editing. think the one that you described, I don't know what they call it. Maybe it's just more contextual editing, but they're able to pick up what you've done recently. Look at the files that you're opening and then provide like multi -line suggestions that maybe if you go inside your tailwind classes and you've copied and pasted like three different elements, and then you go inside the first one.

change it from text blue to text red. Once you change that to text red, it'll say, you've copied and pasted this element three times. Do you want to do that for the next two automatically? And it's like a no -brainer. And when you think about the feature, you're thinking, duh, I want to do this. But no other text editor is able to pull that off. GitHub Copilot's not going to do that because it's providing more kind of contextual within that line. So maybe if you jump to that line, it might try to suggest it. But in reality, Cursor is like thinking about

how you architect and build applications and translating that into the ID into the AI model. And it gives you this really seamless, robust experience that I think people knew it could happen. And when they finally see it interact with it, it's, you I've been waiting for this forever. And now that it's here, like, I'm not going to look back. I think they're really going to have great adoption. think they announced funding like maybe a week ago. I think it's really a rocket ship that's very competitive. think, you know,

Seeing AI grow this way, I'm sure there's going to be lots of competitors in the space, but they really have an edge. And it's been pushed by like Google. Tons of other people are like really saying this is the best thing out there. You need to learn it. And so I think I saw a tweet yesterday that was like, if anyone can make a class for how to use cursor, like a course, so to speak, of how to use cursor effectively, both for engineers and non -engineers.

That would be a quick, like money ticket because so many people are picking up for the first time. And it's a little bit of a learning curve. I'm sure you've seen it. It's not super easy to use, but once you start seeing the value and building up that kind of. I don't know, muscle memory of using the application, then you're pretty productive. And I think that's been my difficulty in the past six months is I think I haven't spent enough time. And I think the features weren't fully developed in a way that they are today. And I felt like I was kind of fighting with the system more than I wanted to.

Bennett Bernard (59:12)
Mm

Mm

Bradley Bernard (59:29)
So from your usage, definitely want to download it again and give it another run, because I feel like I just want to write code quickly. And if that can get me there faster, I'm happy to pay $20 or whatever it may be, or bring my own API key, like you described, literally anything, to just get stuff done faster.

Bennett Bernard (59:46)
Yeah, and so you mentioned Laravel earlier. So I've you've known I've been trying to like get on the Laravel train. And one thing I will say, it's not Laravel's fault, I'm sure. But like, I'm on Windows and you have to jump through so many hoops to like get it set up on Windows. There's the whole Laravel sale thing. Like it's just like drinking from the fire hose. But but that's my point. That's no, it doesn't. It literally like

Bradley Bernard (59:53)
Mm -hmm.

That's Docker. That's Docker. It works on Windows.

Bennett Bernard (1:00:16)
Don't get me started. I've tried. Again, I'm not a technical person. I've tried, Inlet.

Bradley Bernard (1:00:17)
Well, we'll take this off the pod. I'll help Ben just put that in the notes that, you know, maybe next time when we talk again, we'll have the full setup.

Bennett Bernard (1:00:27)
Yeah, maybe. The more I get in there, I'm like, damn it, this is like just way too much. But, you know, they had me try and download the Windows subsystem for Linux. So I have like Ubuntu and all that. like, but it was just I was like, damn it, let me just make a non fast API and just do this in two seconds. So but yeah, that was my thing on Laravel. I want to want to I want to want it. But it Windows experience again, probably a Windows issue. It's been frustrating.

Bradley Bernard (1:00:36)
yeah, yeah.

Mm -hmm.

Bennett Bernard (1:00:55)
one thing I want to talk with you about, I know we're at an hour, but, it had been a while since we caught up and really want to, and this is more about what I was texting about and harassing you about earlier this week. And this is just a purely a theory question and what makes sense to you. So I'll start by this. I've been really playing with llama or the llama models and I downloaded, think since we last talked about this time, like the seven B and the 70 B or whatever they are.

Bradley Bernard (1:01:18)
Mm

Mm

Bennett Bernard (1:01:24)
I downloaded on my machine, Olamma downloaded the seven B model and been playing with that. And first thoughts right away, just so far way less accurate than Claude, just in general and hallucinating way more. And I, but I know you can like, you can fine tune those things. Like I asked it like, give me like some auto franchises. So it's like big O Jiffy Lube, Valvoline, those auto companies that like our. The franchisor business model.

Bradley Bernard (1:01:36)
Mm

Mm

Mm -hmm.

Bennett Bernard (1:01:53)
And it literally gave me like two and I was, I think it missed Valvoline. I was like, well, why didn't you say Valvoline? And I was like, sorry. I'm like, damn, just that alone. You can't get that right. So really it was a little disappointed in that again. That was just like out of the box seven B model. So, you know, I was like, okay. But then I did the 70 B I downloaded that and I asked it a very basic question about like an API endpoint and it was so slow.

Bradley Bernard (1:02:00)
Yeah, yeah.

Bennett Bernard (1:02:21)
So like it was like, I literally said like, how do I do a get request for like this API? And I gave it like a doc or gave it like a little background and couldn't even finish it. And I just gave up. So yeah, that was my thoughts on those llama models. And the reason I'm bringing that up to get a long story, to make this short story long. The reason I bring this up to you is because I am interested in what do you think is the best way to leverage

Bradley Bernard (1:02:27)
Mm -hmm.

Bennett Bernard (1:02:49)
you know, AI or not AI, but like I'm imagining AI to take someone's question about something, translate it into an API request, go get that information from, you know, the API, like the get request and get it back from that system and then give it back to the user. So thinking of like the Hello Kitty game example I've been like talking about on this podcast, I'm not, I'm not using it for this, but like it's kind of as a similar problem of

Bradley Bernard (1:03:13)
Mm

Bennett Bernard (1:03:18)
Hey, I'm going to ask a question to this, you know, chat. can just call it Claude, you know, Claude open AI, you know, llama. I'm going to ask a question to this chat agent and I need it to go look this information up. Do you think it's more efficient to like look up information from like an API request or to like have a background process that

Bradley Bernard (1:03:23)
Mm

Bennett Bernard (1:03:43)
that's going to hit the API, save that information in a database periodically and have the AI hit that database instead. Like that's a long -winded answer and there's probably no right or wrong or long -winded question. There's no right or wrong, but I'm curious from a technical standpoint, from your experience, like from what sounds to be the better approach. Like, should you just go directly hit the API or would it make more sense to it'd be a lighter technical load to just have a preset like

Bradley Bernard (1:03:54)
Mm -hmm.

Bennett Bernard (1:04:12)
set of API calls that just run every 30 minutes say that get data into the database that you have and then just read off that database with AI. Like what do you think? You know, are are you picking up what I'm, what I'm hitting?

Bradley Bernard (1:04:18)
Mm

Yeah. Yeah. Yeah. I think it depends on how often the data changes. So if you're referencing something that's changing pretty often, let's say, you know, every five minutes, every 10 minutes, I'm not sure exactly what the data source is, but if that's the case, you can easily, you know, have a system such that you can take in a query. that may be a question about Hello Kitty, go reach out to the API and fetch it in real time, get that response back from, you know, their API, their data.

and then send a response back to your clients. And, you know, when we talk about these systems, they can be asynchronous. So maybe the website that you're on that you developed, you send a request to your server. So like the web browser to your server, that could be one whole round trip. Then your server sends a request to the Hello Kitty API. That happens in the background. Once that's done, that could take, you know, one second, that could take five minutes. Once that's done, you could use what's called a web socket, which kind of enables

your browser to talk to the server and your server to talk to the browser. That's kind of a bi -directional stream. Then you could provide that data back to the web browser back to the client. So it depends how often the data changes. I think if it's pretty static, I would just, you know, fetch it once or fetch it every hour, fetch it every day, store in a database. And then if a user had a query, I think the best kind of AI answer to this is to use retrieval augmented generation. rag. So essentially,

Bennett Bernard (1:05:49)
Mm -hmm. Yeah.

Bradley Bernard (1:05:50)
You take all that data, which is probably text. It's, you know, describing how to do things within Hello Kitty. You go to a database that's specialized for rag or I think it's like a vector database is what they call it these days. And you shove in all this text data and you get out all these vectors. Then you take your input query. So how do I jump over this wall in Hello Kitty? And then you vectorize that. And then you essentially have.

know, vector input, vector output, and there's like a similarity matching algorithm such that it'll go find the most relevant text. You'll pull out this text from this large data store, and then that is used as the AI answer when it tries to answer your question. So if you ask it, you know, how to jump over this wall and this map, that phrase will go, you know, similarity match across all the data. Maybe it'll pull out, you know, map, jump, et cetera.

phrases that are relevant. And again, it's not just word matching. It's actually contextually word matching, which I think is a huge deal for AI. Then it pulls out various phrases. You know, you have your question of the user, you know, ask this, here's the documentation from Hello Kitty. Please look at the documentation, only the documentation answer this question. If you don't know the answer, say, I don't know. It's kind of the AI prompting methodology such that, you know, it's not something that you really want to

spend too much time on, but you want to pull in that context, get that answer and move forward on that. So yeah, it really depends. think that's the technical answer for everything is like, it really depends. I think in your case, yeah, how often the data changes. If there's a way to get it asynchronously, you can, but yeah, I'm happy to talk more about that. think it's kind of fun designing those systems.

Bennett Bernard (1:07:24)
Mm

Yeah, I think, you know, just hearing that is, yeah, it makes a ton of sense. I think we'd probably just want the data once a day. Probably it'd be sufficient. and I stand corrected if I said earlier seven B eight B supposedly is the model. Looked it up, but you know, whatever, you know what I'm talking about? I'm not, I'm not technical. I get to say whatever I want. I'm not technical.

Bradley Bernard (1:07:40)
Mm

I think I said 7b2 so it's okay. feel like 7 and 70b is way easier to remember than 8 and 70b so you know, meta's problem.

Bennett Bernard (1:07:59)
Yeah, yeah, it was funny. when I downloaded the 70B one, my daughter was watching a show on my other monitor. And then I was doing all my little tinkering on my other monitor. And when I started up the seven 70B model, her show instantly started like, slowing down or whatever. And it was like taking forever. Yeah. And she was like, like, dad, it's not working. And I was like, there's nothing, honey. Don't worry. It'll be back. Hang on.

Bradley Bernard (1:08:10)
Mm

Yeah, it's a behemoth.

Hehehehe. Hehehehehe.

Bennett Bernard (1:08:26)
Let me quit out of this. Yeah, because I mean, I again, I asked it a simple question and it just sat there and spun its wheels for a minute, but.

Bradley Bernard (1:08:32)
So you're running on the beefy graphics card, right? Nice, nice. Nice, nice.

Bennett Bernard (1:08:35)
Yeah, yeah, I got a 4070 Ti Super. So you'd think that'd be enough, but yeah, I was still struggling for a basic GET request.

Bradley Bernard (1:08:44)
Yeah. And I think Meta has the four five B I think. And that one, you know, four times as big roughly or whatever that may be. And you need these dedicated, you know, thousands and thousands of dollars, Nvidia chips that are humongous. And so, yeah, it's kind of crazy. think that's what these companies are really investing in just the raw hardware from Nvidia to be able to run these really complex models and do it, you know, instantaneously such that you can ask a question to Claude.

Bennett Bernard (1:08:49)
Yeah.

Mm

Bradley Bernard (1:09:13)
Or a cursor and you're getting all this information in the backgrounds running on this, you know, $50 ,000 graphics card. so yeah, kind of interesting.

Bennett Bernard (1:09:19)
Yeah. I think the, to put a bow on, you know, that, rag question and, and using the model for, first of all, it was kind of discouraging how poorly the seven B model operated from the cloud one, just from like a user perspective. Again, I didn't like, I didn't, what's the right word? Like prompt the llama model, like super specifically. So it's kind of general.

Bradley Bernard (1:09:47)
Mm

Bennett Bernard (1:09:49)
But like, was just really surprised by like, if I asked the same question in both, the llama one was just like, so general, you know, so I probably would need to tinker with that. Again, I still think there's like, I, that's what I would want to build on because I think just being open source and being more transparent is super important. So like, you know, that's what I would still try and build something on. But I think initially you'd be like, okay, there's a lot more legwork that needs to go into fine tuning and like parameter rising, if that's even a word.

Bradley Bernard (1:09:56)
Yeah.

Bennett Bernard (1:10:19)
like getting this thing to be specific and like, you know, what I'm after.

Bradley Bernard (1:10:23)
just to clarify slightly, Anthropic does have a similar like dumber model, kind of like Metas. And so when you compare Claude to 8b, I think right now there's like three classes. One is like, you know, dumb and fast, and then one is in the middle. One is that top intelligence, not as fast. And so that comparison of 8b and

Bennett Bernard (1:10:25)
Yeah.

Mm

Bradley Bernard (1:10:45)
you know, what you're looking like at Claude is definitely comparing that bottom class for Meta to the top class of Anthropic. So it definitely will perform worse. And I think you could try comparing Anthropic's worst model. They name them confusingly. It's like Sonnet, Haiku, and Opus, I think, are their three models. And I don't know where they exist in the tier of things, but I think Sonnet's the top. And maybe that would be Opus or Haiku would be compared to Llama 8B.

Bennett Bernard (1:10:59)
Yeah.

Yeah, yeah, you got it.

Bradley Bernard (1:11:13)
And so, yeah, I'm curious what those look like in terms of like left, right, same question and different responses.

Bennett Bernard (1:11:18)
Yeah, no true. yeah. So that was, and that makes total sense. and I started learning, I was looking into like Lang chain. I'm not sure if you're familiar with that. Yeah. So we can get into more of that later, but, yeah, it was just an interesting technical question, but yeah. So running in running the two in parallel, that makes a lot of sense why Lama's base model would be just not as performant as, you know, sonnet, the latest and greatest from anthropic.

Bradley Bernard (1:11:26)
yeah, yeah.

Mm

It sounds like you're having fun though, I gotta say it's the itch of just building something and I felt that deeply.

Bennett Bernard (1:11:49)
Well, you know, and I think what I'm interested in right now is the line between, you know, and you know this, but like, it's crazy to see it firsthand, the line between like human language turning into like programming language is just thinning out because it can just translate. can just take that. Yeah, can just, yeah, cursor, you know, even, even things like Alexa where it's like, Hey, what's the weather? And it knows, you know, I think

Bradley Bernard (1:12:06)
Mm

Yeah, cursor.

Bennett Bernard (1:12:18)
I'm not a big fan of Alexa. it just kind of, we have one and I would never get rid of it. But like I ask it questions and it gives me some weird ass answer sometimes. It just does, you know, again, it's an old model, but my point being is like, that's still a really impressive device. If you think about it, like that's super impressive. And like, I kind of take it for granted if I just don't stop and think about it, you know, but I'm like, it's taking my words, knowing what I want and like 90 % of the time giving me something back that I'm

Bradley Bernard (1:12:31)
You

Yeah.

Mm

Bennett Bernard (1:12:47)
that I wanted, like what's the weather, know, what's the forecast and like, you know, having done some like just tinkering with the Alexa SDK with a Hello Kitty game, know, like write the invocations and like, you know, that's, it's, it's, it's impressive. So just that blurring of, you know, being able to interact with these things on a just much more talking basis. I've, saw someone tweet about how they were like trying to set up a voice to voice, a text thing on their computer.

Bradley Bernard (1:12:48)
Mm

Bennett Bernard (1:13:17)
So they would just talk to Cursor, hey, give me some tailwind classes that are super cool. And the text would go into the chat on Cursor, and then Cursor would do its thing. And then the person would just say, OK, accept. And it would accept it. It was not even typing. And it was just like, that's the world we're getting into. it's just fascinating. Yeah, it's a really exciting place to be, for sure.

Bradley Bernard (1:13:19)
Mm

Yeah.

Yeah.

Yeah, it's typing and then it's voice. think voice doesn't work, you know, in obvious settings. Like if you're, you know, on the subway or in a space where you can't talk, but I think the next step and we'll get there probably eventually, I don't know if it's within our lifetime, but getting that brainwave communication where I can think of something where, know, I can imagine, I can, yeah, I mean, I can imagine like a tailwind button and like, can immediately get cursor to think about it. And then I can immediately say yes or no, like on a blink of a dime. So.

Bennett Bernard (1:13:58)
Neuralink.

Bradley Bernard (1:14:10)
I think that would be super amazing. And by the time we have that, know, who knows what else we'll have. But I think the voice is definitely a step forward where it makes sense.

Bennett Bernard (1:14:19)
Yeah. Have you seen the, or like heard that the neuro link patient, you know, the, the brain chip company that Elon's got that the first patient like is like just a whiz at like CS go like can just play it and like, yeah, I don't know if it's a rumor or what, but like, I think he said it, I think he was on an interview cause he was on the true Rogan podcast. And I think he said it like, it's like, it's like a mod basically. And yeah, I mean the way that, our buddy and I were playing

Bradley Bernard (1:14:25)
Mm -hmm, yeah.

I have not heard of a that.

huh.

Wow, that's pretty cool.

Bennett Bernard (1:14:47)
Rainbow six siege and I thought might be other people that have that chip because it was we were getting whooped Yeah for sure Cool. Yeah, I think there was one other thing I was gonna mention but I lost my train of thought but either way I think what's it's probably a good place to wrap up here. So Why don't we end on our our usual kind of bookmarks or saved readings and then we'll close this one up

Bradley Bernard (1:14:51)
A bunch of cheaters.

Awesome. Sounds good. Yeah. I'll go first. again, I haven't been on Twitter too much, but I did have one saved, I think pretty recently. So let me open this up. Thought it was a little bit fascinating. So this is from Ian Newtall. Essentially this user on Twitter found 84 ,000 Chrome extensions that are going to be extinct in about a month.

I guess Google Chrome is doing a deprecation of specific extensions that aren't on the latest manifest. So I think you have to build your extension on Google Chrome's infrastructure or manifest. And for people that aren't on the latest and greatest version, your extension will be removed from the App Store. And so this user on Twitter found like 84 ,000 of them, I think, ranked them by number of reviews. And it's kind of providing an opportunity for folks to say,

If this is off the app store, you know, come in, create something similar, put it out there and maybe make some money. I think those are really cool opportunities because the old always becomes a new again. And, know, I'm sure tons of people are using these extensions and once they're gone, are going to go look for them, but aren't able to get them. And so I saved it as, you know, half of like, this would be cool to talk about in the pod, but other half of, know, maybe it's time to make an extension, especially with Claude, you know, you could easily imagine.

Bennett Bernard (1:16:04)
Mm -hmm.

Bradley Bernard (1:16:26)
create a Chrome extension that can do X, Y, Z, and then you're off to the races. So I'll share it in the show notes, but I bet there will be a lot of businesses or mini side projects started this weekend or next weekend to try to beat people to being the first one to replace some of these expired extensions.

Bennett Bernard (1:16:45)
Yeah, that's super interesting. And I clicked on the link and the person has a Gumroad page now where you can, I think, pay for his list. I think it's one of those things where you pay what price you think is fair. But just fascinating that he even found an opportunity to kind of monetize some of that research, which is a good thing, right? I hey, went through all the work and did it. So kudos to him. But it's interesting that

Bradley Bernard (1:16:52)
Hahaha

Yeah, respect.

Bennett Bernard (1:17:11)
has 1 ,352 downloads from Gumroad. you know, I'm sure people chipped in some money. You know, not everybody, but I'm sure some people did. That's fascinating. Cool. My link was actually what I sent Brad that he didn't read. it's a, but you know, again, he was just being a thoughtful podcast partner and saving it for the pod. So it's about LLMs and SQL. And basically it talks a little bit about

Bradley Bernard (1:17:21)
Mm

Wow.

Bennett Bernard (1:17:40)
You know, using agents, using Lang chain, like what's the best way to have a model read off data. And I probably got like three quarters of the way through it. hadn't finished it, but what was really interesting to me was basically it comes from a couple of authors. So Francisco in Ingham and John Luo. I don't know their full background, but I believe they're very technical is my impression and

when they wrote about it, they basically were saying that like you need to teach or like inform your LLM like about the tables before you can have it efficiently query them. And it kind of was basically kind of going down the path of, you know, think of LLM like as a human, like, you know, when you're a human, if you're like a data analyst or business intelligence analyst, like, you know, what tables to go look for when you get a question, you don't, every time you get a question from, you know,

a business leader or like a business partner, you don't have to reacquaint yourself with the tables. You know, kind of, okay, this is where that is. That's where this is. And we need to kind of try and give LLMs that same context. And so they basically were kind of going through a couple of different ways to like contextualize your data, you know, with Lang chain and, know, using these different libraries. So there's a Python Lang chain library and then the JavaScript Lang chain library.

Bradley Bernard (1:18:56)
Mm -hmm.

Bennett Bernard (1:19:06)
And, yeah, it was just super fascinating. And again, it was topical to, you know, the conversation that we had just had about like using API calls versus reading a database and talks about like describing the schema talks about describing the data, like what it looks like in there. and all this information that you need to kind of give it for it to be again, efficient and knowledgeable to ultimately get what you're after. If you're trying to have it, you know, do a query for you. So,

Bradley Bernard (1:19:32)
Nice.

Bennett Bernard (1:19:32)
Yeah, just, just very interesting. And again, very topical to what I was looking to, looking to try and do with my own little project.

Bradley Bernard (1:19:42)
Yeah, think Lang chain had a bit of a hype cycle when AI was first out. And then I think it's kind of died down a little bit with a decent amount of backlash, at least on the small echo chamber on X. So I'm not sure where it exists today. I think it's kind of in the middle, hides a lot of complexity, but also gets the job done in ways that makes it easy to get started. So yeah, I'd love to dive into that a little deeper and maybe a future pod.

Bennett Bernard (1:19:48)
Mm

Yeah, one funny thing before we go, when I was looking at, was doing some of that fast API stuff with cursor and all that. It was authentication, know, Laravel, like Django, like the big web frameworks have good like built in authentication. Fast API is a little bit more wild west. And it was using, it was like JavaScript object, something, it was an abbreviation. And that wasn't JSON.

Bradley Bernard (1:20:16)
Mm

JSON or is it JSON web token? JSON web token, J2ET?

Bennett Bernard (1:20:34)
No, it It wasn't JWT. the Claude and cursor. So I had Claude up on one screen that had, you know, it's coding on cursor. They said, Hey, go download or, you know, pip install the, you know, install this library. And it was, I'm being dead serious. was Python dash Jose J O S E. And I'm like, I'm like, what is Jose? Like, is this like for real? Like, is this just some person's like library that they're going to try and like hack me with? It just felt so like.

Bradley Bernard (1:20:52)
Okay.

-huh.

Bennett Bernard (1:21:02)
not legit, but I guess the JOS stands for something. But at first I was like reluctant to install it. I'm like, this doesn't seem legit. Because I've heard about like how, yeah.

Bradley Bernard (1:21:04)
-huh.

Yeah, I just Googled it. JavaScript object signing and encryption. Interesting.

Bennett Bernard (1:21:17)
Yeah. Yeah. So I was like, I'm not, I'm not installing that until I do some more research. But then I looked at it. has a lot of contributors has a lot of, you know, forks and stars or whatever. So it seemed, it seemed legit. And again, Claude and cursor recommending it to me, but I was like, I don't, you know, am I really going to like, is this named after the person? Like did Jose code this or what? Like, but no, it was just confusing.

Bradley Bernard (1:21:34)
Yeah.

It looks like it is JSON web tokens, so I think it might be named differently. But yeah, interesting, interesting. Sometimes you got to fact check the AI, sometimes it's not right. So you definitely got to be careful. I think you did the right thing.

Bennett Bernard (1:21:43)
Mm -hmm, yeah.

Yeah, when I heard, I read an article a while back about how there was some fraudulent libraries out on like PI, think it's PI PI P Y P I there's some, some website that has all the Python libraries. And I read some story that they're a hacker or a of hackers had made libraries that were very commonly named to like an existing library, but just like off by one character. And so people would get confused and they would

Bradley Bernard (1:22:01)
Mm

Hmm.

Bennett Bernard (1:22:20)
PIP install that one and I guess it would cause some security issues. like, you know, pandas, P A N D A S they go PIP install pandas. But like there was like hackers, for example, that would write a package that was like a malware basically under Panda. So, and if someone like typed in PIP install Panda, somehow it was like getting into their system. And so ever since I read that again, years ago, that's always kind of stuck with me. I'm like, it feels like I'm just giving a green light to just downloading a bunch of.

Bradley Bernard (1:22:36)
Hmm.

Bennett Bernard (1:22:49)
code that I don't really fully understand myself. just something to be careful about, I guess. Yeah.

Bradley Bernard (1:22:53)
Yeah, I guess one last note, I feel like I have to build on this. There was a coding conference that I went to that they had talked about a security exploit where they would do a man in the middle attack on the wifi network of the conference. And when folks were actually downloading kind of the real packages like pandas, like you mentioned, they would swap out the download with a malicious package. So they knew developers were going to be working on their apps and you know, this was an iOS coding conference. So.

Bennett Bernard (1:23:16)
Hmm.

Bradley Bernard (1:23:21)
They would hook into the network requests. I would go reach out to the dependency manager for iOS libraries, swap that out with a bad one. And no one would know the difference. When they ran and built their app and distributed it to the app store, they could have this bug and they could steal all these user credentials. So yeah, it was kind of interesting. was like, wow, there's so much little details about working in public and making sure the security is tight and right.

Bennett Bernard (1:23:31)
Hmm.

Bradley Bernard (1:23:45)
Yeah, I think kind of talk about misspelling is one thing, but even getting it right, making sure your Wi -Fi is like secure from end to end or else, you know, you might get a rogue package in there that you'll never know until it's too late.

Bennett Bernard (1:23:52)
Mm

Yeah, that's crazy. that is a scary world out there. Cool. All right, well, let's wrap this up. Good stuff, Brad. And we'll try and catch you again when you're not jet setting around the country. Until next time.

Bradley Bernard (1:24:00)
Cool.

Yeah, we'll talk about Laracon next time, so it'll be fun. Cool. Sounds good. See ya.

Bennett Bernard (1:24:09)
Yeah, for sure. Awesome. See ya.