WEBVTT

00:00:17.741 --> 00:00:18.409
<v Matt>Hello, Nina.

00:00:19.233 --> 00:00:19.698
<v Nina>Hello, Matt.

00:00:20.557 --> 00:00:24.445
<v Matt>Well, it's been going good. I got a whole lot of stuff to talk about this time

00:00:24.445 --> 00:00:26.259
<v Matt>around. A little more prepared.

00:00:27.405 --> 00:00:32.965
<v Nina>Same. Same. It was a very, very busy week in the tech world, it seems.

00:00:33.485 --> 00:00:38.078
<v Nina>I could not be on YouTube without seeing something about Jev come up.

00:00:38.587 --> 00:00:39.203
<v Matt>Yeah.

00:00:39.563 --> 00:00:39.946
<v Nina>Yeah.

00:00:40.666 --> 00:00:44.735
<v Matt>Exactly. So I jumped on the hype train for Jev. I jumped on the hype train for

00:00:44.735 --> 00:00:47.324
<v Matt>Amp. I jumped on the hype train for Codex.

00:00:47.874 --> 00:00:52.222
<v Matt>and basically two of them seemed very overhyped for me, and one was real.

00:00:52.783 --> 00:00:58.090
<v Matt>So I guess two truths and a truth, we could play that game.

00:00:59.085 --> 00:01:00.070
<v Nina>Two hypes and a truth.

00:01:00.610 --> 00:01:01.951
<v Matt>You could play two truths and a lie?

00:01:02.282 --> 00:01:04.105
<v Nina>One truth and two lies.

00:01:04.105 --> 00:01:05.730
<v Matt>Yeah, something like that. Whatever.

00:01:06.113 --> 00:01:06.670
<v Nina>Yeah.

00:01:07.393 --> 00:01:12.426
<v Matt>Yeah, so I started off with Jev, because everyone was talking about it,

00:01:13.218 --> 00:01:16.285
<v Matt>and it's just a classifier.

00:01:16.925 --> 00:01:18.204
<v Matt>that does some,

00:01:19.490 --> 00:01:25.455
<v Matt>some JSON outputs, which we kind of already do in our own cloud development.

00:01:26.087 --> 00:01:30.673
<v Nina>Right. You could technically do what Jev is doing with...

00:01:31.276 --> 00:01:36.030
<v Nina>Well, no. Well, okay. I'll let you make your case for it first.

00:01:37.316 --> 00:01:38.070
<v Nina>Why you think that?

00:01:38.670 --> 00:01:42.390
<v Matt>In theory, it can't. But if you put code, like we do it in ours,

00:01:42.390 --> 00:01:47.930
<v Matt>we're doing like a data pipeline, you're going to get you can get json because

00:01:47.930 --> 00:01:51.280
<v Matt>you're going to format it through or you can at least be passing it through

00:01:51.280 --> 00:01:52.793
<v Matt>code that needs it formatted.

00:01:53.444 --> 00:01:59.500
<v Nina>Right right but like in for so for okay for eta brain like we have we have those

00:01:59.500 --> 00:02:03.353
<v Nina>four scrubs right i can see i can see why it might not be necessary

00:02:03.881 --> 00:02:09.371
<v Nina>for the eta brain which is for viewers who don't know that's the podcast app that we have where

00:02:10.147 --> 00:02:14.677
<v Nina>there's like four to six Haiku calls, you know, there's a scrub,

00:02:15.101 --> 00:02:19.934
<v Nina>does the query need rewriting, there's the router, selects one of three of your question types.

00:02:20.352 --> 00:02:24.420
<v Nina>It's, there's certain things that you can just manually program in there,

00:02:24.420 --> 00:02:31.339
<v Nina>but Jev could optimize whether or not some of those calls are useful.

00:02:31.915 --> 00:02:36.460
<v Nina>Like, especially for the first one, does this query need rewriting?

00:02:36.460 --> 00:02:42.635
<v Nina>I mean, you're right you're right it's easy to program a yes or no for that but

00:02:43.252 --> 00:02:50.350
<v Matt>Yeah like why would we introduce a whole nother system to do what save us instead

00:02:50.350 --> 00:02:52.825
<v Matt>of like a tenth of a penny a hundredth of a penny,

00:02:53.579 --> 00:02:58.200
<v Matt>would be the cost of the call because we were using Haiku for it like you just

00:02:58.200 --> 00:03:00.387
<v Matt>like load tested it yesterday and spent like a dollar.

00:03:00.934 --> 00:03:04.950
<v Nina>Yeah that was after optimizing the ask pipeline and a bunch of other things but yeah

00:03:07.464 --> 00:03:15.528
<v Matt>Like we would have to be in just area hitting so many queries for that to be a noticeable cost.

00:03:16.193 --> 00:03:21.661
<v Nina>Right but if there's four to six in and we're both hitting we're hitting a model

00:03:21.661 --> 00:03:27.901
<v Nina>that spoo's tokens and techs jev can make make those decisions early and decide

00:03:27.901 --> 00:03:30.781
<v Nina>if like it's even worth making

00:03:31.408 --> 00:03:34.670
<v Nina>those calls with like less code i don't know

00:03:34.897 --> 00:03:41.400
<v Matt>There's so how much time would you save like and the whole thing takes like at worst a minute.

00:03:41.970 --> 00:03:48.727
<v Nina>Right 1.2 seconds per call i mean i'm gonna have to go through the ask pipeline

00:03:49.052 --> 00:03:53.092
<v Nina>and just use jev because i haven't had a chance to use jev i've just been seeing the hype yeah i

00:03:54.891 --> 00:03:59.901
<v Matt>Tried i tried to use it with the chief of staff stuff and it just was it's not

00:03:59.901 --> 00:04:04.481
<v Matt>useful for me i mean i'm not classifying anything other than like prs which

00:04:04.481 --> 00:04:08.032
<v Matt>you can't really do well it's too much context.

00:04:09.305 --> 00:04:13.595
<v Nina>You're right okay so that that's a limitation of it because there's yes or no

00:04:14.351 --> 00:04:18.385
<v Nina>but there's there's also isn't there two other types there's there's the score

00:04:18.699 --> 00:04:23.301
<v Nina>where i can rank something on a metric like see okay

00:04:23.301 --> 00:04:28.498
<v Matt>All of it is basically a score with categories and confidence levels.

00:04:30.005 --> 00:04:32.687
<v Matt>The binary one is just a category with two.

00:04:34.706 --> 00:04:38.171
<v Matt>They're just over-hyping that for no reason.

00:04:39.171 --> 00:04:43.007
<v Matt>I question how much is their own fault and how much is just the internet's fault.

00:04:43.326 --> 00:04:45.550
<v Matt>I think a lot more of it's just the internet's fault.

00:04:46.531 --> 00:04:53.101
<v Matt>This play makes more sense for like, you have a set pipeline already and you're

00:04:53.101 --> 00:04:56.512
<v Matt>just jamming through terabytes of data,

00:04:57.218 --> 00:05:02.276
<v Matt>And like, you need a category on it. Like I have a past example that I could

00:05:02.276 --> 00:05:03.586
<v Matt>see Jev being useful for.

00:05:03.586 --> 00:05:09.226
<v Matt>Like I talked about where we were generating pictures for like Amazon catalog

00:05:09.226 --> 00:05:13.415
<v Matt>items to like show in your hand, like how big it looked in your hand.

00:05:13.885 --> 00:05:19.386
<v Matt>And every once in a while we'd have an outlier that was like a tuna can the

00:05:19.386 --> 00:05:22.406
<v Matt>size of your head, even though it wasn't, it was just a normal one.

00:05:22.406 --> 00:05:26.600
<v Matt>Someone put in centimeters instead of inches or something or meters instead of inches.

00:05:27.174 --> 00:05:31.816
<v Matt>and so that i could see that because what we had to do is we had to give it

00:05:31.816 --> 00:05:34.663
<v Matt>to mechanical turk and have people go and manually click it,

00:05:35.243 --> 00:05:40.026
<v Matt>like if we could just go through all of them super fast and then like a person

00:05:40.026 --> 00:05:44.756
<v Matt>at the company could spend an hour and through whatever's left or double check

00:05:45.156 --> 00:05:47.305
<v Matt>the false for the positives that came out,

00:05:47.935 --> 00:05:51.686
<v Matt>like that makes sense and you would be doing enough with because those are images

00:05:51.686 --> 00:05:57.296
<v Matt>so the tokens are huge and you you don't want the like you want the speed because

00:05:57.296 --> 00:05:59.284
<v Matt>you just want yes no on that data.

00:06:00.336 --> 00:06:07.226
<v Matt>Or on my unlike pr reviews like you're like oh i see this i want to dive more

00:06:07.226 --> 00:06:11.126
<v Matt>into this oh i need to actually look at something that's not in the pr like

00:06:11.126 --> 00:06:12.661
<v Matt>that you want a full model for that.

00:06:13.319 --> 00:06:20.396
<v Nina>Right right so okay since you worked with it more i have i have an app that

00:06:20.396 --> 00:06:24.040
<v Nina>i made over the course of like i think since we started this podcast

00:06:24.481 --> 00:06:28.616
<v Nina>and what it is is it's similar to your chief of staff in which it's evaluating

00:06:28.616 --> 00:06:30.954
<v Nina>a huge amount of context but it's not a pr

00:06:31.494 --> 00:06:34.309
<v Nina>it's an answer to a coding exercise question

00:06:35.830 --> 00:06:41.498
<v Nina>a very complicated like like this isn't like a like evaluating like a like a

00:06:41.498 --> 00:06:46.132
<v Nina>finite output it's more like abstract like text answers like

00:06:46.637 --> 00:06:52.018
<v Nina>asking how would you build this if you had these constraints and like just very complicated

00:06:52.818 --> 00:06:58.330
<v Nina>i use claude api to evaluate the answer like zero out of a hundred

00:06:59.092 --> 00:07:04.908
<v Nina>now if i were to integrate jev into that based off your experience with chief

00:07:04.908 --> 00:07:09.488
<v Nina>of staff like what would that what would that look like and would it be no because

00:07:09.488 --> 00:07:16.868
<v Matt>You're scoring that's not qualification of yes or no you want context a normal

00:07:16.868 --> 00:07:24.324
<v Matt>model the more general yeah because what you could do is like yes or no like correct or not correct.

00:07:24.969 --> 00:07:31.658
<v Nina>Yeah but it's more it's more on a percentage you want a gradient sir but but

00:07:31.658 --> 00:07:39.896
<v Nina>here's the thing like you could you could have You could be looking for specific fragments of words,

00:07:40.505 --> 00:07:45.144
<v Nina>like fragments of sentences in the answer, and that kind of be your checkpoint

00:07:45.562 --> 00:07:50.168
<v Nina>of, okay, they mentioned A, B, C, D, E, right?

00:07:50.168 --> 00:07:52.853
<v Nina>And that could be a part of the evaluation.

00:07:53.757 --> 00:07:54.291
<v Nina>By just

00:07:57.227 --> 00:08:01.161
<v Nina>parsing that text and seeing that it has a specific amount of mentions of certain

00:08:01.161 --> 00:08:04.931
<v Nina>topics, and that could be a part of the grading, and maybe Jev could come in there.

00:08:06.231 --> 00:08:06.831
<v Matt>You could give it,

00:08:09.047 --> 00:08:15.751
<v Matt>the critiques a little bit, outline, and be like, did it do this? Did it do B? Did it do C?

00:08:16.704 --> 00:08:19.624
<v Matt>And then it would give you a yes or no on each of them.

00:08:20.273 --> 00:08:25.224
<v Nina>Right, and that could be where Jev could be integrated. But is it worth it?

00:08:26.251 --> 00:08:26.411
<v Matt>No,

00:08:28.491 --> 00:08:34.500
<v Matt>I don't see how you make it worth it unless you are using gigantic data models.

00:08:34.854 --> 00:08:35.202
<v Nina>Yeah.

00:08:35.597 --> 00:08:39.940
<v Matt>To a point where you should start to think about building your own model.

00:08:40.527 --> 00:08:46.711
<v Nina>Jev being used to determine data that could be relevant.

00:08:46.711 --> 00:08:52.321
<v Matt>No, at the point you would start using Jev, I feel. You would already have enough

00:08:52.321 --> 00:08:56.004
<v Matt>data to want to do your own model.

00:08:56.614 --> 00:09:01.391
<v Nina>Right, that's what I mean. Let's say hypothetically, on a different topic,

00:09:01.391 --> 00:09:03.175
<v Nina>now I'm jumping to that since you mentioned that.

00:09:03.771 --> 00:09:12.639
<v Nina>Let's say you want to train a model and you want to have Jev like sifle through like a million

00:09:13.278 --> 00:09:19.351
<v Nina>entries of data and determine based off a metric, does this qualify as useful

00:09:19.351 --> 00:09:21.672
<v Nina>to train this model for this specific purpose?

00:09:22.314 --> 00:09:26.501
<v Nina>And you kind of have that iterative approach or not iterative,

00:09:26.501 --> 00:09:27.985
<v Nina>but like recursive approach to...

00:09:28.821 --> 00:09:33.336
<v Nina>Yeah, going through millions of entries to make a more efficient model.

00:09:33.336 --> 00:09:34.847
<v Nina>That would be a good use case.

00:09:35.444 --> 00:09:40.530
<v Matt>Yeah, so you're doing it as making the seed data for the model, right?

00:09:40.860 --> 00:09:41.731
<v Nina>Right, right.

00:09:42.103 --> 00:09:47.262
<v Matt>And yeah, that makes some sense. But that's like a one-and-done thing.

00:09:47.865 --> 00:09:50.796
<v Nina>It is. It is, but...

00:09:50.826 --> 00:09:55.638
<v Matt>Like, I have so much trouble finding, like, this is a good company idea.

00:09:57.552 --> 00:10:05.126
<v Matt>It's just... It's so niche in what it does to a point where someone that needs

00:10:05.126 --> 00:10:07.679
<v Matt>it is good enough to build it themselves.

00:10:08.906 --> 00:10:13.786
<v Nina>Yeah, I think you're right. But I'm still going to use that API token and I'm

00:10:13.786 --> 00:10:17.444
<v Nina>going to see what I can do with it. I'm still going to mess around and see.

00:10:17.787 --> 00:10:21.346
<v Matt>Yeah, give it a shot. I don't know.

00:10:22.890 --> 00:10:26.666
<v Matt>What would you grade it on, right? It's got a RAG for the ETA brain.

00:10:26.666 --> 00:10:28.043
<v Matt>It's got a RAG that looks...

00:10:28.670 --> 00:10:32.147
<v Matt>for like some topics so you can get a confidence.

00:10:32.809 --> 00:10:33.767
<v Nina>Yeah you could load up

00:10:34.074 --> 00:10:34.759
<v Matt>Oh go ahead.

00:10:35.079 --> 00:10:41.106
<v Nina>So here here's where the it could be useful in eta brain so there's like there

00:10:41.106 --> 00:10:44.907
<v Nina>could be a simple retrieval there could be a deep retrieval

00:10:45.354 --> 00:10:50.446
<v Nina>i mean that's that's where it could it could make the decision of like how many

00:10:50.446 --> 00:10:54.326
<v Nina>how many calls to make like determining like

00:10:55.782 --> 00:11:00.397
<v Nina>i mean outside of like does the query need rewording or something? That's silly.

00:11:00.896 --> 00:11:06.048
<v Nina>But in the router, there is a which of, I believe the Ask pipeline has like,

00:11:06.414 --> 00:11:08.823
<v Nina>which of these three types of questions is this?

00:11:09.183 --> 00:11:13.896
<v Nina>So maybe a little bit of Jev right there, right? Without making a Claude call.

00:11:15.700 --> 00:11:20.580
<v Nina>So that would be at the router level. So between scrub, router,

00:11:20.580 --> 00:11:24.880
<v Nina>re-rank, and then you re-rank per hop, I think.

00:11:25.820 --> 00:11:31.240
<v Nina>I was researching last night and asking Claude the same thing again.

00:11:31.240 --> 00:11:33.938
<v Nina>No, I want to specifically know about this.

00:11:34.321 --> 00:11:36.140
<v Nina>Stop generating documentation.

00:11:37.240 --> 00:11:40.033
<v Nina>I just want to have a little conversation here with your context.

00:11:40.946 --> 00:11:45.603
<v Nina>which by the way like outside of plan mode how do you actually like

00:11:46.259 --> 00:11:50.485
<v Nina>like have like a built how do you how do you how do you actually like plan with

00:11:50.636 --> 00:11:52.540
<v Nina>Claude in a way where it just doesn't

00:11:52.540 --> 00:11:56.284
<v Nina>jump the gun and just keep generating documentation upon documentation

00:11:57.310 --> 00:12:01.770
<v Nina>i'm changing subjects too much but definitely the router level i wanted to say

00:12:01.770 --> 00:12:05.380
<v Matt>Yeah so i've also seen videos of the router,

00:12:07.140 --> 00:12:15.320
<v Matt>the model router is what I've heard people call it to decide which level to use and that's okay,

00:12:16.740 --> 00:12:22.300
<v Matt>because you give it the models you can give and then it'll come out with confidence

00:12:22.300 --> 00:12:26.480
<v Matt>level and then in the code you can route the data to the correct one,

00:12:27.606 --> 00:12:33.125
<v Matt>But that's like the whole additional system you're adding to your pipeline just

00:12:33.125 --> 00:12:36.232
<v Matt>to do that routing where you just have Haiku do it.

00:12:36.894 --> 00:12:41.416
<v Matt>And it is more expensive, like noticeably more expensive.

00:12:42.024 --> 00:12:47.647
<v Matt>But again, we're talking like the difference between a penny and like a hundredth of a penny.

00:12:48.321 --> 00:12:54.561
<v Matt>So unless you have the huge throughput, which if you're doing human interactions is not likely.

00:12:55.380 --> 00:13:00.725
<v Matt>like if you're doing like fraud detection through logs of financial data yeah

00:13:00.725 --> 00:13:04.250
<v Matt>okay but they probably already have their own models that do that.

00:13:04.867 --> 00:13:09.575
<v Nina>True but maybe there's somebody who's building an app that doesn't have like

00:13:09.575 --> 00:13:13.681
<v Nina>are you talking about like a proprietary model specifically for

00:13:14.733 --> 00:13:17.613
<v Matt>Yeah again to the point I just don't know who the hell this thing's for.

00:13:18.190 --> 00:13:23.146
<v Nina>Well so customer support that could be one

00:13:24.916 --> 00:13:30.705
<v Nina>does this need a human it could be the choice type should we send this question

00:13:30.705 --> 00:13:35.035
<v Nina>to like billing technical cancellation or fraud and then it could have a score

00:13:35.035 --> 00:13:38.668
<v Nina>for like urgency I mean that's one thing

00:13:39.520 --> 00:13:44.205
<v Matt>I guess I mean you'd have to get such a speed boost to make this worth it.

00:13:44.987 --> 00:13:48.848
<v Nina>Yeah where is it worth it is the question I see the uses but

00:13:49.526 --> 00:13:50.600
<v Matt>Like into like,

00:13:51.508 --> 00:13:54.027
<v Matt>If you're going to do like an agent routing, right?

00:13:54.027 --> 00:13:54.995
<v Nina>Mm-hmm.

00:13:55.795 --> 00:14:01.897
<v Matt>Is adding like a high, a kaku-kaku fast like going to add that much when it's

00:14:01.897 --> 00:14:04.868
<v Matt>going to route to Fable or Astra?

00:14:05.444 --> 00:14:09.557
<v Matt>Like those two are going to eat so much time that you're saving on the wrong

00:14:09.557 --> 00:14:12.689
<v Matt>part of the model timeline.

00:14:13.095 --> 00:14:19.247
<v Nina>Right, but each call does take like about 1.2 seconds and I did read that Jev

00:14:19.247 --> 00:14:25.762
<v Nina>as like milliseconds. So you are saving a little bit of time, too, on the user side.

00:14:26.617 --> 00:14:31.497
<v Matt>Seconds, which, I mean, and sometimes, yes, that's super important,

00:14:32.197 --> 00:14:35.253
<v Matt>like, for, like, media-type stuff.

00:14:36.337 --> 00:14:40.156
<v Matt>Mm-hmm. But when, like, you're a user in an app, like, you can,

00:14:40.783 --> 00:14:43.877
<v Matt>a little spinny thing for an extra second's not going to kill you.

00:14:44.616 --> 00:14:49.117
<v Matt>When you, like, I mean, we talk to AI models, and they spin for,

00:14:49.117 --> 00:14:50.209
<v Matt>like, two to three minutes.

00:14:50.733 --> 00:14:56.827
<v Nina>Right, right. But FYI, also, I added a spinner to ETA brain.

00:14:57.379 --> 00:15:02.767
<v Nina>That's one little thing I did aside from all this, like, the one UI UX thing.

00:15:02.767 --> 00:15:04.641
<v Matt>Yeah, just add a spinner.

00:15:07.668 --> 00:15:11.958
<v Nina>I mean, it made me happy to just see a spinner. So maybe you're right.

00:15:12.542 --> 00:15:19.687
<v Nina>But, like, what about, I don't know, so many times, like, you're working with, like, a chatbot.

00:15:20.052 --> 00:15:25.137
<v Nina>and you're trying to get like a refund for like an order or something

00:15:25.793 --> 00:15:29.907
<v Nina>like Jev and and most people like get really impatient and they just like i'd

00:15:29.907 --> 00:15:33.137
<v Nina>like to talk to somebody i'd like to talk to somebody i don't know if Jev's

00:15:33.137 --> 00:15:36.315
<v Nina>gonna really change that like i don't i don't know

00:15:37.688 --> 00:15:43.754
<v Nina>I don't know what I'm going to build this week, but I'm still going to see the improvements.

00:15:44.154 --> 00:15:49.851
<v Nina>Because every time I keep messing with ETABrain, I do a probe evaluation where

00:15:49.851 --> 00:15:55.571
<v Nina>I compare where Master was at before based off how many calls made.

00:15:55.571 --> 00:15:58.815
<v Nina>The Ask pipeline hasn't changed, but how many calls have been made?

00:15:59.475 --> 00:16:06.424
<v Nina>How long does it take to go through each step of the calls and go through the pipeline?

00:16:07.021 --> 00:16:09.848
<v Nina>And I got it from,

00:16:12.589 --> 00:16:22.375
<v Nina>on the old jar, it was 17 to 19 seconds for that first call to 0.8 seconds, it said, 0.8.

00:16:23.101 --> 00:16:29.651
<v Nina>And then when listing relevant moments to part of the app where you click relevant

00:16:29.651 --> 00:16:35.635
<v Nina>searches, and it got from 17 to 19 seconds to 8.6 seconds.

00:16:36.276 --> 00:16:39.851
<v Matt>Yeah, I mean, you're going from serverless to dedicated server, so that makes sense.

00:16:40.170 --> 00:16:42.721
<v Nina>Yeah. Yeah, but it was still nice to see.

00:16:43.743 --> 00:16:49.934
<v Nina>But maybe with Jev, I don't know. Maybe I could try to see if, like, I'll see.

00:16:51.059 --> 00:16:53.829
<v Matt>You can put me wrong. I will accept it if I see it.

00:16:54.899 --> 00:17:00.958
<v Nina>I don't know until I actually go in and try, but I need to also go in without Claude in the equation.

00:17:01.347 --> 00:17:04.255
<v Nina>Like, this is a moment where I have to... What?

00:17:04.985 --> 00:17:05.327
<v Matt>Why?

00:17:07.459 --> 00:17:13.839
<v Nina>Because sometimes I don't want Claude to come up with the creativity side of

00:17:13.839 --> 00:17:18.611
<v Nina>looking at a pipeline and seeing how I can use like a third like another third party

00:17:19.540 --> 00:17:24.429
<v Nina>tool I want to go through and see what I could think of just looking at the

00:17:24.429 --> 00:17:26.482
<v Nina>code myself I mean do you ever do that

00:17:27.075 --> 00:17:30.599
<v Matt>Just no no more why wouldn't you have it just this why would you have it just

00:17:30.599 --> 00:17:34.449
<v Matt>tell you what's in there diagram it and then you tell it what to do.

00:17:34.449 --> 00:17:38.822
<v Nina>I did I did diagram it I'm gonna diagram that's that's a much better idea but

00:17:39.502 --> 00:17:43.669
<v Nina>I'm sorry I've been doing so many interviews this week and code exercises and

00:17:43.669 --> 00:17:50.119
<v Nina>i'm telling you i do miss i do miss the code editor i do it's

00:17:50.976 --> 00:17:58.919
<v Nina>yeah no yeah no yeah i i have i in other news i have a lot of interviews this

00:17:58.919 --> 00:18:01.250
<v Nina>week and an offer so that's good congrats

00:18:01.907 --> 00:18:07.259
<v Nina>thank you i just got two more to finish up and i really want one of them I

00:18:08.541 --> 00:18:16.510
<v Nina>really want one of them yeah but yeah in this process I have to go back to coding a little bit because

00:18:17.203 --> 00:18:19.734
<v Matt>So what's the process they giving you?

00:18:21.187 --> 00:18:28.548
<v Nina>Well I guess this is interesting for the podcast I had to write unit tests in

00:18:28.738 --> 00:18:32.701
<v Nina>Java it was like a pair programming thing

00:18:33.862 --> 00:18:40.275
<v Nina>So it was, they drive and I tell them what to write for certain cases. I was

00:18:42.535 --> 00:18:47.158
<v Nina>expecting algorithms and, I don't know, building a little chess game.

00:18:48.135 --> 00:18:52.255
<v Nina>Remember what you made me go through? Not made me go through, that was fun.

00:18:53.862 --> 00:18:57.705
<v Nina>But no, this was very straightforward. And then just a lot of technical verbal

00:18:57.705 --> 00:19:02.435
<v Nina>questions about systems engineering, more so than the coding.

00:19:02.435 --> 00:19:07.285
<v Nina>they want to see how you would think about if you have an app at a small scale,

00:19:07.285 --> 00:19:11.076
<v Nina>a medium scale, a large scale how would you build it differently?

00:19:12.019 --> 00:19:12.855
<v Matt>How big is the company?

00:19:14.217 --> 00:19:19.897
<v Nina>There's four companies one's very small the react i'm just going to call them like react shop small

00:19:20.466 --> 00:19:26.272
<v Nina>broadband cable company like i think like broadband cable software company they

00:19:26.272 --> 00:19:29.626
<v Nina>make software for broadband cable companies maybe like 50 employees

00:19:30.247 --> 00:19:35.872
<v Nina>the one i really want they have like 10 000 employees big big corporate and then

00:19:36.983 --> 00:19:44.512
<v Nina>the fourth one small like maybe 100 so most they're mostly small but i would

00:19:44.512 --> 00:19:49.070
<v Nina>get to work in rust in one of them so i'm cool

00:19:50.152 --> 00:19:58.332
<v Nina>that's pretty cool and it's rust typescript Vue 3 aws like that's that's what

00:19:58.332 --> 00:20:04.124
<v Nina>i worked in for six years plus minus Java Spring and just swap that with rust

00:20:04.972 --> 00:20:10.262
<v Nina>but yeah today it's probably going to be algorithmic exercises for for one of

00:20:10.262 --> 00:20:16.419
<v Nina>them they're going to make me do front end and then i also believe back end which i did not prep for

00:20:16.912 --> 00:20:18.253
<v Matt>Yeah we'll see.

00:20:18.805 --> 00:20:19.333
<v Nina>We'll see

00:20:19.936 --> 00:20:25.027
<v Matt>Yeah i'd be surprised if your large one doesn't include many more steps.

00:20:25.602 --> 00:20:29.992
<v Nina>We have to build a pokemon game in java which is going to be pretty fun that's

00:20:29.992 --> 00:20:32.503
<v Nina>the next one like object-oriented programming

00:20:33.216 --> 00:20:37.038
<v Matt>Tweaked this is why i'm freelance Yeah.

00:20:38.362 --> 00:20:43.030
<v Nina>I don't know how to go about telling one company, can you wait a week?

00:20:43.768 --> 00:20:47.309
<v Nina>You know, like, I don't know how to do that,

00:20:47.599 --> 00:20:49.114
<v Matt>But just tell them.

00:20:50.125 --> 00:20:51.228
<v Matt>tell me you got other things.

00:20:51.940 --> 00:20:56.344
<v Nina>I i did i said hey i'll let you know friday and then i got another interview

00:20:56.344 --> 00:20:58.684
<v Nina>for the one i really want next week oh

00:20:58.684 --> 00:21:03.809
<v Matt>Yeah well you know okay so you already told them one thing and i can backtrack okay.

00:21:04.419 --> 00:21:08.424
<v Nina>I could verbally say yes and then well whatever i'll figure it out i want to

00:21:08.424 --> 00:21:11.089
<v Nina>be respectful and honor everybody's time but

00:21:11.333 --> 00:21:13.208
<v Matt>I mean the fire at will goes both ways.

00:21:14.934 --> 00:21:15.224
<v Nina>Yeah.

00:21:17.254 --> 00:21:20.570
<v Nina>They said they put aside money to bring me on.

00:21:21.482 --> 00:21:27.920
<v Nina>I don't want to talk about the specifics of it though, because this is a podcast where it goes out.

00:21:31.386 --> 00:21:36.611
<v Nina>Why are people doing the algorithmic exercises when we have these AI tools anymore?

00:21:41.105 --> 00:21:44.815
<v Matt>Java just has dot sort. You just let it do what it does. I never understood.

00:21:44.815 --> 00:21:47.349
<v Matt>Ten years ago, I made people do it.

00:21:48.665 --> 00:21:50.878
<v Matt>No one's writing sorting algorithms.

00:21:51.517 --> 00:21:56.265
<v Nina>Right. I learned how to do... I forgot how to... Well, not learn.

00:21:57.228 --> 00:22:04.595
<v Nina>I forgot how to do a quick sort. And I had it open, and I was like, oh my god, I hate this.

00:22:04.595 --> 00:22:06.682
<v Matt>And it turns out you never do,

00:22:08.765 --> 00:22:12.091
<v Matt>a quick search. So it's just like, why?

00:22:12.561 --> 00:22:19.805
<v Nina>They should let people use AI tools outside of Claude for coding exercises,

00:22:19.805 --> 00:22:23.175
<v Nina>because that's what the day-to-day is going to be anyway. And they should judge

00:22:23.175 --> 00:22:25.326
<v Nina>and evaluate you by what you search.

00:22:25.963 --> 00:22:31.655
<v Nina>i think that would be that would be to interview people because if you're if

00:22:31.655 --> 00:22:36.185
<v Nina>you're searching like a regex like cool if you're if you're searching like a

00:22:37.245 --> 00:22:43.955
<v Nina>like like a higher higher level decision or something that you should know then red flag

00:22:44.441 --> 00:22:49.409
<v Matt>I mean if someone just shot from the hip a correct regex i would hire them no matter what it was.

00:22:50.032 --> 00:22:51.414
<v Nina>If they shot oh

00:22:51.565 --> 00:22:55.555
<v Matt>Oh if they were just like straight off the dome like this is the regex you need

00:22:55.555 --> 00:22:59.332
<v Matt>to do the correct search and it worked i'd be my mind would be blown.

00:23:01.886 --> 00:23:04.435
<v Nina>I know like two or three i i yeah

00:23:04.435 --> 00:23:08.945
<v Matt>Exactly no one knows that off the top of their head other than complete insane people.

00:23:09.629 --> 00:23:14.805
<v Nina>I always i always mess up like just starting a regex like i can remember i can

00:23:14.805 --> 00:23:19.765
<v Nina>remember passing the regex but i can never remember which way it starts,

00:23:19.765 --> 00:23:22.539
<v Nina>which way you escape, like actually, oh,

00:23:23.803 --> 00:23:28.430
<v Matt>If they took the time to learn that, they know everything else. So you just hire them.

00:23:29.551 --> 00:23:33.208
<v Nina>I'm just going to make flashcards of every regex I can think of and just...

00:23:34.294 --> 00:23:38.787
<v Nina>Yeah, it seems that companies are having a really hard time coming up with creative

00:23:38.787 --> 00:23:41.326
<v Nina>ways to interview people because of AI.

00:23:41.732 --> 00:23:42.037
<v Matt>Yeah.

00:23:44.524 --> 00:23:50.520
<v Matt>There's a lot of people on the other side of the equation, too.

00:23:51.997 --> 00:23:55.540
<v Matt>It was not as easy to find people, but now you're swamped with people.

00:23:56.167 --> 00:24:00.010
<v Matt>Everyone's resume is just AI-generated to be perfect.

00:24:01.137 --> 00:24:05.201
<v Matt>It wasn't useful before, but it's even now it's a false positive.

00:24:05.776 --> 00:24:12.761
<v Nina>Yeah i mean i i i used Claude design for mine but i embellished it

00:24:13.663 --> 00:24:19.317
<v Nina>i embellished it as much as i could to not flag ats readers you know like just

00:24:19.317 --> 00:24:21.241
<v Nina>so just so it stands out but

00:24:22.168 --> 00:24:27.407
<v Nina>did have you used Claude for like resumes and invoices and stuff I mean,

00:24:27.407 --> 00:24:31.480
<v Matt>I do my own invoices, yeah. I've helped other people's resumes.

00:24:32.119 --> 00:24:34.162
<v Matt>So, done that with Claude.

00:24:34.764 --> 00:24:37.086
<v Nina>And they're like, wow, this looks great. And do you tell them you used Claude?

00:24:37.698 --> 00:24:38.837
<v Matt>Yeah, I mean.

00:24:38.837 --> 00:24:45.227
<v Nina>Okay. I've seen friends that recently had a few friends in IT that were all

00:24:45.227 --> 00:24:46.413
<v Nina>laid off at the same time.

00:24:47.114 --> 00:24:51.067
<v Nina>And they were asking me to help proofread their resume and everything.

00:24:51.067 --> 00:24:57.047
<v Nina>and they there are people who are extremely anti-ai but looking at like their

00:24:57.047 --> 00:25:02.177
<v Nina>resumes it was very i think it was a case where i think they needed to use ai

00:25:02.177 --> 00:25:04.325
<v Nina>for like layout stuff i don't know

00:25:04.963 --> 00:25:09.127
<v Matt>Yeah that's where your resumes used to be useful in helping see how good the

00:25:09.127 --> 00:25:11.112
<v Matt>person is at human communication,

00:25:11.861 --> 00:25:16.267
<v Matt>they would like basically they have to describe their life in one sheet of paper

00:25:16.267 --> 00:25:19.537
<v Matt>like what do they pick that's important what do they leave out how do they present

00:25:19.537 --> 00:25:21.630
<v Matt>it what's the layout yeah,

00:25:22.297 --> 00:25:27.257
<v Matt>um but like i usually did that more than like did they work at google or they

00:25:27.257 --> 00:25:33.158
<v Matt>work at some small mom and pop shop like that was less important to me yeah yeah.

00:25:34.877 --> 00:25:40.209
<v Nina>I want to move on to the orbing with Amp, but I also want to talk about,

00:25:40.720 --> 00:25:45.997
<v Nina>there was a video about the tech industry where I think it was like,

00:25:45.997 --> 00:25:49.970
<v Nina>it was like one of those like kind of vice format, like documentary style, like short,

00:25:50.477 --> 00:25:56.284
<v Nina>short things where they were interviewing three people who had been laid off from meta,

00:25:57.542 --> 00:26:02.445
<v Nina>Intel, and then the other one, Microsoft, I think.

00:26:02.961 --> 00:26:09.467
<v Nina>these were all people with like anywhere from eight to 15 years of experience

00:26:09.467 --> 00:26:12.347
<v Nina>and they're all complaining and how they couldn't find a job like one was a

00:26:12.347 --> 00:26:14.917
<v Nina>product product engineer one was

00:26:15.500 --> 00:26:20.267
<v Nina>a software engineer at meta who claims that she submitted 8 000 applications

00:26:20.267 --> 00:26:24.517
<v Nina>and she can't find a job and it's been 15 months and then another one was

00:26:25.254 --> 00:26:29.777
<v Nina>i think like a cloud engineer of some kind But yeah, they were all complaining

00:26:29.777 --> 00:26:34.427
<v Nina>that they couldn't find jobs and they all came from these big companies.

00:26:35.057 --> 00:26:37.956
<v Matt>Yeah. I mean, they're overpriced and they're underskilled.

00:26:38.577 --> 00:26:38.607
<v Nina>Okay.

00:26:39.997 --> 00:26:40.821
<v Nina>You know what? That answer is...

00:26:42.836 --> 00:26:46.371
<v Nina>It was bugging me, but if you do stay at a company for many,

00:26:46.371 --> 00:26:48.345
<v Nina>many years and you just work

00:26:49.052 --> 00:26:55.261
<v Nina>in maybe a department or a team that still uses archaic things and they haven't

00:26:55.261 --> 00:26:59.024
<v Nina>migrated and they earn like $180,000,

00:26:59.635 --> 00:27:00.245
<v Nina>you're right.

00:27:00.829 --> 00:27:04.721
<v Matt>Yeah, I mean, there's a sweet spot between the evergreen beginner,

00:27:04.721 --> 00:27:10.521
<v Matt>someone who switches jobs every year, and the giant corporation 15 years.

00:27:11.316 --> 00:27:15.801
<v Matt>They've gotten very good at handling the beasts of the Fortune 500 company they're

00:27:15.801 --> 00:27:18.831
<v Matt>in, but they can't learn a new system.

00:27:19.453 --> 00:27:21.427
<v Matt>They spent 15 years learning one system.

00:27:22.011 --> 00:27:23.761
<v Nina>Well, Matt, you just said that.

00:27:24.481 --> 00:27:28.111
<v Matt>And now they want half a mil to join a new job where they're not going to do

00:27:28.111 --> 00:27:32.341
<v Matt>well because they're used to getting their stock options that were like $400K

00:27:32.341 --> 00:27:34.109
<v Matt>because they've been there for 10 years.

00:27:34.411 --> 00:27:38.891
<v Nina>So wait, what's that term? what's the term with like a silver spoon what's the

00:27:38.891 --> 00:27:44.292
<v Nina>term golden handcuff golden golden handcuffs silver spoon okay so they're spoiled

00:27:45.241 --> 00:27:46.501
<v Nina>yeah can't find a job if they've

00:27:46.501 --> 00:27:51.357
<v Matt>Been at facebook for 15 years even eight years i'm pretty sure they got a lot of money.

00:27:51.913 --> 00:27:55.371
<v Nina>Yeah i kind of like that they're going through like you know there's like these

00:27:55.371 --> 00:27:58.981
<v Nina>sad montages of like them looking out a window and

00:27:59.888 --> 00:28:03.551
<v Nina>and meanwhile they're in like these huge houses and walking like there's three

00:28:03.551 --> 00:28:09.641
<v Nina>dogs and it's like okay anyway you answered my question pretty quickly i was

00:28:09.861 --> 00:28:11.870
<v Nina>complaining about this particular

00:28:12.589 --> 00:28:15.951
<v Nina>this particular doc to like two or three friends and they were like they didn't

00:28:15.951 --> 00:28:20.449
<v Nina>have anything like finite or any opinion to say on it but you just nailed it

00:28:20.722 --> 00:28:24.251
<v Matt>Yeah probably don't want to be mean okay,

00:28:25.899 --> 00:28:33.597
<v Matt>Anyways, back to Orbin. Okay, so I tried Orbs, and I signed up for Codex to do it this week.

00:28:34.471 --> 00:28:42.148
<v Matt>So basically what it lets you do is it's got a Visual Studio interface, basically.

00:28:42.148 --> 00:28:47.058
<v Matt>And then you got your projects on the side, and then you'd be like,

00:28:47.058 --> 00:28:51.267
<v Matt>okay, this project is about this repo, so you connect the repos.

00:28:51.748 --> 00:28:55.690
<v Matt>And then it's like, okay, I want to start a tab. So you start a tab,

00:28:55.998 --> 00:28:59.230
<v Matt>and that's like now your chat interface with the LLM.

00:28:59.904 --> 00:29:04.548
<v Matt>And then you go, okay, I want to do a facelift on the UI.

00:29:04.548 --> 00:29:10.248
<v Matt>And it's like, okay, I'll go and spin up an Oracle, and then I'll spin up to

00:29:10.248 --> 00:29:15.006
<v Matt>learn about it, and then I'll spin up some workers to do it all on a different machine.

00:29:15.586 --> 00:29:18.108
<v Matt>And it goes and does that. And then you can spin up another tab.

00:29:18.108 --> 00:29:22.488
<v Matt>It's like, okay, make the backend more efficient. And then it spins up another one.

00:29:22.877 --> 00:29:28.102
<v Matt>So basically, you're spinning up VMs for each of these tests that then have...

00:29:28.960 --> 00:29:33.467
<v Matt>code that have ai on it and can run and code and then i'll bring comes back

00:29:33.467 --> 00:29:37.557
<v Matt>to you with prs and things like that it kind of actually doesn't want you to

00:29:37.557 --> 00:29:42.696
<v Matt>do prs it wants you to be a vibe coder and not do that stuff but you can do prs if you really make it.

00:29:43.401 --> 00:29:51.447
<v Nina>Okay so it's it's not built off of just one single ai model right it's it's

00:29:51.447 --> 00:29:58.254
<v Nina>um it switches between various models for like specific tasks like intelligently

00:29:58.783 --> 00:30:03.547
<v Matt>It has its own models which i'm pretty sure are Anthropic but they say they

00:30:03.547 --> 00:30:07.797
<v Matt>didn't i don't know if they said 100 it is and then you have your own ChatGPT,

00:30:08.752 --> 00:30:13.317
<v Matt>account and then basically it has like a knob which for some reason i don't

00:30:13.317 --> 00:30:18.817
<v Matt>know why to like turn up the intensity with different models and stuff and it's

00:30:18.817 --> 00:30:22.525
<v Matt>a lot of configuration there that i did not find enjoyable,

00:30:23.517 --> 00:30:29.360
<v Matt>and but then like it takes forever just for it to write some code i had to go

00:30:29.511 --> 00:30:33.459
<v Matt>and like update a blog post and that took like 10 minutes,

00:30:34.266 --> 00:30:39.961
<v Matt>and like before i even started writing anything just to like spin up the box and load up the model.

00:30:40.463 --> 00:30:42.501
<v Nina>Was it just the first call because

00:30:42.983 --> 00:30:49.007
<v Matt>No i did i did about five before i gave up so i did like five within one because

00:30:49.007 --> 00:30:51.047
<v Matt>i thought that was slow and then i was I was like, oh, maybe it's just this

00:30:51.047 --> 00:30:55.187
<v Matt>box is iffy. And then I started another model and tried a few times,

00:30:55.187 --> 00:30:57.405
<v Matt>and then another, and another, and another. And then I was like, yes.

00:30:57.980 --> 00:31:04.679
<v Matt>I was like, is this Codex or is this Amp? And so I downloaded Codex.

00:31:05.318 --> 00:31:09.451
<v Matt>And it rewrote the blog post awesomely in like two minutes.

00:31:10.377 --> 00:31:14.452
<v Matt>And that was the first go. And I think most of that time was me logging in.

00:31:15.249 --> 00:31:17.896
<v Matt>And so I like Codex. Amazing.

00:31:19.617 --> 00:31:20.917
<v Nina>I've never programmed with Codex.

00:31:21.317 --> 00:31:24.798
<v Matt>Yeah, I finally bit the bullet, tried it out. Well, it wasn't coding.

00:31:24.978 --> 00:31:29.122
<v Matt>The coding, I find, is okay. It's the same as Claude to me.

00:31:29.424 --> 00:31:35.812
<v Matt>But the human language writing is like head and shoulders above Fable.

00:31:36.621 --> 00:31:39.464
<v Matt>And I didn't even try Astra. I was on a lower model.

00:31:40.161 --> 00:31:43.633
<v Nina>That's the only reason I would like to try Codex.

00:31:44.114 --> 00:31:49.577
<v Nina>And it doesn't surprise me that the human language coming from Codex is much

00:31:49.577 --> 00:31:52.809
<v Nina>better, because ChatGPT in general is much better than...

00:31:53.976 --> 00:31:58.486
<v Nina>human language and just writing natural flowing sentences claude is horrible

00:31:58.486 --> 00:32:03.601
<v Nina>at it yeah awful at it what's the pricing point on that out of curiosity

00:32:03.740 --> 00:32:12.236
<v Matt>Because so i used the 200 bucks a month one because i wanted to mess around

00:32:12.236 --> 00:32:16.834
<v Matt>with Astra later um i think you could do the 20 bucks one and probably be fine.

00:32:17.446 --> 00:32:20.052
<v Nina>I'm on a 200 Claude subscription

00:32:20.238 --> 00:32:28.408
<v Matt>Yeah i am too and i'm gonna keep that or i think I'll do coding in Claude and writing in Codex.

00:32:28.878 --> 00:32:30.248
<v Matt>That's kind of where I'm going to go.

00:32:30.875 --> 00:32:33.870
<v Nina>When you try Astra, what are you going to use it for in particular?

00:32:34.164 --> 00:32:42.305
<v Matt>I might redo some UIs. Let's see, it's better at the UI design than Claude.

00:32:43.826 --> 00:32:48.749
<v Nina>I do use ChatGPT for generating UI mocks over...

00:32:48.964 --> 00:32:54.746
<v Nina>I mean, I'll use both. Okay, so I've used Claude design before,

00:32:54.746 --> 00:33:00.486
<v Nina>but typically it works better for UI mocks if you use ChatGPT,

00:33:01.564 --> 00:33:07.903
<v Nina>generate the mocks, and then feed that to Claude, develop a design system with some inputs.

00:33:08.153 --> 00:33:13.906
<v Nina>Just bringing them back and forth, typically results get some good results.

00:33:13.906 --> 00:33:17.706
<v Nina>Like, one of these apps that I maintained that I ran into this week,

00:33:18.908 --> 00:33:22.506
<v Nina>my first production bug, because I actually have users on this app, Thank you.

00:33:23.796 --> 00:33:27.837
<v Nina>I had to like come up with threads on comments and I was like,

00:33:27.837 --> 00:33:29.837
<v Nina>man, this UI is not working for anything.

00:33:29.837 --> 00:33:33.997
<v Nina>And I had to, you know, go, I went back and forth developing like what it would

00:33:33.997 --> 00:33:38.937
<v Nina>look like and coming up with a design system. But yeah, starting in ChatGPT

00:33:40.040 --> 00:33:43.251
<v Nina>and then switching to Claude and back and forth.

00:33:43.663 --> 00:33:48.591
<v Nina>And sometimes even like just feeding the HTML and CSS that Claude generates

00:33:49.282 --> 00:33:54.791
<v Nina>to ChatGPT with an image with some input is really interesting.

00:33:55.957 --> 00:34:03.957
<v Nina>The latest model on Codex called, is it, I'm sorry, is it Astra or is it, it's Astra, right?

00:34:03.957 --> 00:34:04.764
<v Matt>Yeah, Astra's the new one.

00:34:05.508 --> 00:34:10.537
<v Nina>That's not available in like browser ChatGPT. no

00:34:10.537 --> 00:34:13.809
<v Matt>You gotta like yeah pay the hundred or more for it.

00:34:14.349 --> 00:34:22.457
<v Nina>Yeah I've seen it generate 3D models and like full games and I'd be really interested

00:34:22.457 --> 00:34:23.971
<v Nina>in seeing how it does there

00:34:24.801 --> 00:34:25.713
<v Matt>Because I I've.

00:34:26.997 --> 00:34:31.037
<v Nina>Used I've used a lot of AI for like for game engine stuff

00:34:31.950 --> 00:34:38.017
<v Nina>but I'm finally like I'm finally like actually making like I'm making like a

00:34:38.017 --> 00:34:42.258
<v Nina>like a socket a socket based game with like multiple players for this

00:34:42.911 --> 00:34:47.807
<v Nina>this website that i have with multiple users we i have like 15 to 20 but like

00:34:47.807 --> 00:34:49.431
<v Nina>they're on it like every day

00:34:50.163 --> 00:34:51.683
<v Nina>and yeah i ran into

00:34:52.504 --> 00:34:58.007
<v Nina>i ran into a bug where i accidentally coupled comments on documents because

00:34:58.007 --> 00:35:00.003
<v Nina>i was just not paying attention

00:35:01.077 --> 00:35:05.497
<v Nina>like it should have been a separate table and so we had like one one person

00:35:05.497 --> 00:35:10.267
<v Nina>who had a document open and then another person who was also in the document

00:35:10.267 --> 00:35:14.517
<v Nina>commenting on the document and the reads and writes were concurrent and it was

00:35:15.343 --> 00:35:18.251
<v Nina>duh so I had to decouple decouple them

00:35:18.954 --> 00:35:24.520
<v Nina>And yeah, it was easy with Claude and doing some research, but I should have

00:35:24.520 --> 00:35:28.420
<v Nina>never let Claude make the decision to couple them in the first place.

00:35:28.950 --> 00:35:32.640
<v Nina>I was like, oh yeah, that'll be fine. And then not thinking that documents could

00:35:32.640 --> 00:35:35.894
<v Nina>have multiple people viewing them and writing to them.

00:35:36.462 --> 00:35:39.446
<v Matt>What the hell is this game that has 20 people constantly using it?

00:35:39.922 --> 00:35:46.620
<v Nina>So what it is is it's for a discord community and it's a round robin writing

00:35:46.620 --> 00:35:50.150
<v Nina>game where people can write fiction together online

00:35:50.741 --> 00:35:54.890
<v Nina>so it's for a specific fandom i'm not going to mention the fandom but yeah so

00:35:54.890 --> 00:35:57.663
<v Nina>how it works is you sign up you get it you you make an account

00:35:58.093 --> 00:36:01.780
<v Nina>you mark whether you're a part of a certain discord community and then you get

00:36:01.780 --> 00:36:05.770
<v Nina>certain access to certain features like sharing round robin writing games to

00:36:05.770 --> 00:36:08.151
<v Nina>that discord and to specific channels

00:36:09.196 --> 00:36:14.660
<v Nina>And you have a profile and you have you have like an about section where people

00:36:14.660 --> 00:36:21.512
<v Nina>can link like their their websites and it's also a solo document editor for for fiction too

00:36:22.075 --> 00:36:24.570
<v Nina>so people can share their stories

00:36:24.570 --> 00:36:29.511
<v Nina>with with others in this fandom add beta readers to it mark it done

00:36:29.918 --> 00:36:36.290
<v Nina>have line by line annotations but the round robin writing part is the fun part

00:36:36.290 --> 00:36:40.285
<v Nina>so every week people get on it you have a host of a game

00:36:40.952 --> 00:36:46.070
<v Nina>and that host you can have up to like i think it's five or six writers in one one session

00:36:46.740 --> 00:36:51.800
<v Nina>and a prompt generator and so the prompt generation generates writing prompts

00:36:51.800 --> 00:36:56.520
<v Nina>based off like characters, location, tags, call to action.

00:36:56.520 --> 00:37:03.252
<v Nina>Like it's, it's a pretty, it's a pretty rope, like big, big prompt generator.

00:37:03.822 --> 00:37:06.160
<v Nina>And so then everybody then gets like four prompts.

00:37:06.783 --> 00:37:10.835
<v Nina>And in the session, they can all vote, vote on the prompts.

00:37:11.451 --> 00:37:16.536
<v Nina>So the one that wins is the one people write and you can, you have timed rounds.

00:37:17.122 --> 00:37:22.532
<v Nina>So every person gets like 60 seconds to write a line or two minutes to write a line.

00:37:23.685 --> 00:37:29.073
<v Nina>And when the timer starts counting down, there's like music that stresses them out.

00:37:29.888 --> 00:37:33.382
<v Nina>There's also gimmicks to mess up their writing.

00:37:34.011 --> 00:37:34.836
<v Matt>And this is free?

00:37:35.515 --> 00:37:36.948
<v Nina>Yeah, this is something I made.

00:37:37.328 --> 00:37:38.877
<v Matt>Why are you charging people for this?

00:37:39.675 --> 00:37:41.068
<v Nina>Because I hate making money.

00:37:43.099 --> 00:37:47.220
<v Nina>Because it was just for a Discord community, for people who wanted to write

00:37:47.631 --> 00:37:49.158
<v Nina>essentially fanfiction.

00:37:49.622 --> 00:37:55.388
<v Nina>and I just wanted to like give some friends like a nice little space for people

00:37:55.388 --> 00:38:01.498
<v Nina>to write because everybody was paying for like ellipses which is like a writing

00:38:01.498 --> 00:38:02.988
<v Nina>platform that you have to pay.

00:38:02.988 --> 00:38:03.879
<v Matt>How much is that?

00:38:04.520 --> 00:38:08.428
<v Nina>I'm going to open it up for multiple fandoms. So this would be for like,

00:38:08.428 --> 00:38:10.408
<v Nina>you know, the Tumblr cretins.

00:38:14.475 --> 00:38:18.068
<v Nina>It's $6 per month but an annual plan of $60 per year.

00:38:18.846 --> 00:38:21.228
<v Matt>Yeah. And how many uses do you have?

00:38:22.548 --> 00:38:27.208
<v Nina>I have... It's growing. Because people outside of the Discord are like,

00:38:27.208 --> 00:38:28.734
<v Nina>Hey, my friend wants to join.

00:38:28.968 --> 00:38:33.448
<v Matt>God, Nina! You're just sitting on this thing? Yeah, it's okay.

00:38:33.448 --> 00:38:35.306
<v Matt>So how many uses do you have? A hundred?

00:38:35.537 --> 00:38:38.368
<v Nina>I can send it to you and you tell me if I should start charging.

00:38:39.049 --> 00:38:42.088
<v Nina>So I'm a little embarrassed about it because it's for fanfiction.

00:38:42.088 --> 00:38:44.648
<v Nina>It's for weirdos who like writing fanfiction,

00:38:45.028 --> 00:38:48.710
<v Matt>You know? Yeah, but you could just make it about marketing and then you could do B2B.

00:38:49.266 --> 00:38:53.568
<v Matt>and have like b2b ad making scram sessions.

00:38:54.167 --> 00:38:56.795
<v Nina>Oh i could have ad revenue on it can i

00:38:57.197 --> 00:39:02.568
<v Matt>Well you can add ad revenue as well like you do the free version for the fandoms

00:39:02.568 --> 00:39:08.058
<v Matt>and they could be like getting ads and then you can do a paid version for businesses

00:39:08.447 --> 00:39:11.221
<v Matt>who want to like have writing bashes and their marketing team,

00:39:11.830 --> 00:39:14.054
<v Matt>and they could pay like 100 bucks a month.

00:39:14.639 --> 00:39:16.568
<v Nina>Writing bashes for marketing Thank you.

00:39:17.189 --> 00:39:22.367
<v Nina>I guess it's it's style. There's 20 different themes for the specific fandom.

00:39:22.877 --> 00:39:27.230
<v Nina>And like, if I show it to you, you'll probably be like, oh, my God,

00:39:27.230 --> 00:39:29.410
<v Nina>how much time did you spend on this?

00:39:29.770 --> 00:39:33.222
<v Matt>Yeah, I'm already thinking that. So it can't get worse.

00:39:33.559 --> 00:39:36.760
<v Nina>Yeah, that's a good idea. So this.

00:39:36.760 --> 00:39:40.836
<v Matt>So what do you what do you have? You had 50. You had 100. You had 10.

00:39:41.419 --> 00:39:42.640
<v Nina>50, 100, 10.

00:39:42.640 --> 00:39:43.381
<v Matt>What do you mean?

00:39:43.991 --> 00:39:48.430
<v Nina>No i've kept it just to a single discord because every time i tell somebody

00:39:48.430 --> 00:39:50.200
<v Nina>they want to jump on it and yeah

00:39:50.700 --> 00:39:56.470
<v Nina>i capped it at a hundred just in case so i can handle traffic so if the hundredth

00:39:56.470 --> 00:40:00.800
<v Nina>user joins they will or tries to sign up or the hundred and first they get a

00:40:00.800 --> 00:40:05.513
<v Nina>nice little animated banner that says hello i'm one person join the waiting list

00:40:06.069 --> 00:40:12.125
<v Matt>Oh my god Nina so you have 600 a month in revenue that you're not even trying?

00:40:12.757 --> 00:40:15.400
<v Nina>Show me how to do, I don't know how to market. I don't know.

00:40:15.400 --> 00:40:19.400
<v Nina>This is why, this is why I'm, I'm, I'm a contractor with your consulting.

00:40:19.400 --> 00:40:24.130
<v Nina>Maybe you could show me how to market myself here because I,

00:40:24.130 --> 00:40:30.281
<v Nina>I'm very good at marking, marketing, creative, like things that I just make,

00:40:30.512 --> 00:40:34.330
<v Nina>but I'm not good at like coming up with a plan to make money at them.

00:40:34.330 --> 00:40:37.687
<v Nina>So yeah, I know.

00:40:39.729 --> 00:40:43.549
<v Nina>What would you do if I, because I'm starting to open it up and I'm starting

00:40:43.549 --> 00:40:46.949
<v Nina>to like post it on, like, I'm going to start posting.

00:40:46.949 --> 00:40:49.466
<v Matt>You got to start charging. You got to start charging like now.

00:40:51.269 --> 00:40:56.851
<v Matt>At least throw ads on there and then have a free no ads. Pay me six bucks a month.

00:40:57.643 --> 00:41:09.381
<v Nina>Okay. I put a lot of time into caching and snapshots and seeding like content data for prop prompts.

00:41:09.847 --> 00:41:14.083
<v Nina>If something goes wrong, I have version history and rollbacks now.

00:41:14.707 --> 00:41:19.705
<v Matt>Yeah. So there's this thing called Stripe. You should spend like a day on it and add Stripe.

00:41:20.706 --> 00:41:22.346
<v Nina>Stripe for payment?

00:41:22.974 --> 00:41:27.252
<v Matt>Yeah. And then you get Google Ads, and then you add that into it as well.

00:41:27.833 --> 00:41:29.481
<v Nina>But Google Ads are so ugly.

00:41:31.646 --> 00:41:37.956
<v Matt>Facebook Ads? Which ads do you prefer? Maybe you can get some authors to sell

00:41:37.956 --> 00:41:39.345
<v Matt>their books or something. I don't know.

00:41:39.891 --> 00:41:41.760
<v Nina>It could compete with AO3.

00:41:42.399 --> 00:41:43.956
<v Matt>I don't know what that is. What's AO3?

00:41:43.956 --> 00:41:49.756
<v Nina>Okay. It's a fan fiction writing website. But here, after this call,

00:41:49.756 --> 00:41:52.666
<v Nina>I'll send it to you. It's on my resume because it looks really nice.

00:41:53.106 --> 00:41:55.930
<v Matt>Yeah, you have a working business. Of course that sounds nice.

00:41:57.506 --> 00:42:02.085
<v Nina>I am answering to bugs in the Discord because somebody was like,

00:42:03.086 --> 00:42:06.728
<v Nina>the other day, this is hilarious to you, but no, someone the other day,

00:42:06.933 --> 00:42:09.278
<v Nina>and this is something I learned coding with AI, by the way.

00:42:09.585 --> 00:42:11.218
<v Nina>Somebody the other day was like,

00:42:11.725 --> 00:42:20.494
<v Nina>help, my story keeps having amp and amp ampersand, and it kept appending.

00:42:21.180 --> 00:42:25.906
<v Nina>And I was like, immediately, I was like, sanitizer. The sanitizer screwed up.

00:42:25.906 --> 00:42:28.118
<v Nina>Like when you switch from HTML to rich text.

00:42:28.673 --> 00:42:32.219
<v Nina>And it kept saving and doing that on their document.

00:42:33.003 --> 00:42:37.786
<v Nina>Which, it's corrupt data, but it kind of is like a memory leak.

00:42:37.786 --> 00:42:42.146
<v Nina>And if it's spam saving because there's changes every 30 seconds on the page,

00:42:42.896 --> 00:42:45.647
<v Matt>That could... And they can pay you $6 and you will fix it.

00:42:47.136 --> 00:42:48.982
<v Nina>I fixed it. I did fix it.

00:42:51.686 --> 00:42:57.156
<v Nina>It was because I, instead of having a third party like dom sanitizer,

00:42:57.156 --> 00:43:01.558
<v Nina>I coated my own with Claude because I was like, why not? And there was one bug in it.

00:43:02.447 --> 00:43:05.518
<v Nina>So I'll charge. I'll charge. I will. I'll figure it out.

00:43:06.365 --> 00:43:11.778
<v Nina>That's the thing about, like, these fan spaces is that, like,

00:43:12.231 --> 00:43:16.811
<v Nina>they're typically, like, anywhere from, like, college kids to, like,

00:43:17.764 --> 00:43:22.364
<v Nina>a middle-aged mom who doesn't have a job, like, that are in,

00:43:22.364 --> 00:43:27.124
<v Nina>like, these groups, like, you know? So it's kind of, I do have a...

00:43:27.564 --> 00:43:31.154
<v Matt>And they pay for, like, Wordle stuff, or they'll do ads, like,

00:43:31.154 --> 00:43:32.499
<v Matt>for those little games they play.

00:43:33.142 --> 00:43:35.667
<v Matt>Like, you're not breaking the mold, too.

00:43:35.934 --> 00:43:37.489
<v Nina>I was going to add little games, too.

00:43:37.698 --> 00:43:44.177
<v Matt>It already is a game. They're playing these things. You could have leaderboards. They're sharing data.

00:43:44.827 --> 00:43:50.744
<v Nina>Leaderboards, but the only way that, because there's only 25, I host

00:43:52.844 --> 00:43:57.477
<v Nina>a round-robin writing once a week on a Saturday for two and a half hours.

00:43:57.896 --> 00:44:05.428
<v Nina>But it is going to have a calendar where people can actually see who's writing at which times.

00:44:05.771 --> 00:44:09.109
<v Nina>because there is an active now and you can see who's online.

00:44:09.689 --> 00:44:15.462
<v Matt>But first, it will have ad revenue. Alright, I'll accept ad revenue.

00:44:16.200 --> 00:44:21.998
<v Nina>Maybe I should just ask Claude, how do I make this make me money? Yeah. Just do that.

00:44:22.520 --> 00:44:26.514
<v Matt>I mean, so many people are using it and there's a waitlist too? Yeah.

00:44:27.598 --> 00:44:30.192
<v Nina>Only if we hit 100 users, it's pre-programmed.

00:44:30.192 --> 00:44:31.992
<v Matt>Okay, so you haven't hit 100 yet. Okay.

00:44:31.992 --> 00:44:35.806
<v Nina>No, it's pre-programmed because I'm trying not to share it too much.

00:44:36.079 --> 00:44:41.252
<v Nina>But I'll implement the ad revenue, and then I'll start sharing it to other discords

00:44:41.252 --> 00:44:42.520
<v Nina>and see if they want to use it.

00:44:43.113 --> 00:44:46.514
<v Nina>So I will. I will, I promise.

00:44:47.309 --> 00:44:52.112
<v Nina>Is there a way to have Google Ads where it's not, it's not like in your face

00:44:52.112 --> 00:44:57.097
<v Nina>every two seconds and you have to like play a little game where you're hitting X every, you know?

00:44:57.952 --> 00:45:02.023
<v Matt>It's a discord app right so i don't know you know i'll figure that one out sorry.

00:45:02.644 --> 00:45:08.279
<v Nina>Okay okay i do have somebody subscribing like a few like a couple people subscribing

00:45:08.418 --> 00:45:12.145
<v Nina>like ten dollars on like a personal like kind of patreon page

00:45:13.019 --> 00:45:18.169
<v Nina>That's something, but yeah. Yeah, help me with marketing, because I don't know how to market.

00:45:18.715 --> 00:45:21.757
<v Matt>I don't know- Dude, what do you mean to help you market? You already have marketing.

00:45:21.757 --> 00:45:25.187
<v Matt>People are trying to share this thing already for you. You have viral marketing.

00:45:25.810 --> 00:45:31.037
<v Nina>But I need to know how to make, outside of ad revenue, a subscription plan might

00:45:31.037 --> 00:45:33.386
<v Nina>turn people off. I don't know, it might.

00:45:34.100 --> 00:45:37.897
<v Nina>I have a support, support me at the bottom. I do have that.

00:45:39.362 --> 00:45:44.272
<v Matt>You just have it do ads for free and pay for not ads.

00:45:44.824 --> 00:45:48.278
<v Matt>And then you can add more stuff later for a higher pay point.

00:45:48.684 --> 00:45:53.616
<v Nina>And maybe if they watch ads, they could unlock cool things and that's more incentive to...

00:45:54.029 --> 00:45:58.394
<v Matt>I want to get that. It's not illegal, but it's just a pain in the butt.

00:45:58.987 --> 00:46:04.207
<v Matt>Get some ads in there and then it just keeps popping up and you get a few cents

00:46:04.207 --> 00:46:06.677
<v Matt>for every hour or something they play.

00:46:06.677 --> 00:46:12.078
<v Matt>it's better than nothing and then see how many people move over to the subscription,

00:46:12.623 --> 00:46:17.117
<v Matt>or if you have new features if it'll make you feel better you can put them behind

00:46:17.117 --> 00:46:20.992
<v Matt>the paywall and then be like oh I got some cool new stuff,

00:46:21.740 --> 00:46:25.237
<v Matt>and so they're paying for the new stuff not the free stuff you've been giving

00:46:25.237 --> 00:46:27.388
<v Matt>to them for god knows how long.

00:46:28.071 --> 00:46:32.355
<v Nina>Yeah there's even like a like a quick command for writer's reference too

00:46:32.756 --> 00:46:37.847
<v Matt>You don't need to tell me about hotkey stuff you got in there just make them pay for this.

00:46:38.356 --> 00:46:40.017
<v Nina>I'm sorry, it's painful. It's

00:46:43.097 --> 00:46:44.597
<v Nina>painful. Yeah. Okay.

00:46:46.116 --> 00:46:52.369
<v Nina>I didn't even get to talking about how orbing technically sucks with Replit, because...

00:46:52.596 --> 00:46:53.622
<v Matt>Oh yeah, what are they doing?

00:46:54.157 --> 00:46:59.236
<v Nina>The agent takes forever. Just like you said with Amp.

00:46:59.929 --> 00:47:05.087
<v Nina>There's no stream of what it's doing. Like, you know, you have with Claude or

00:47:05.296 --> 00:47:06.469
<v Nina>anything in your terminal.

00:47:06.829 --> 00:47:13.832
<v Matt>Yeah, same with Amp. You can get a little bit of it, but it's definitely meant to be vibed coded.

00:47:14.749 --> 00:47:18.151
<v Matt>you're just supposed to let Amp take the wheel and hope for the best.

00:47:18.862 --> 00:47:24.459
<v Nina>It's broken if I've pushed up from my local machine to Replit and I've tried

00:47:24.459 --> 00:47:29.139
<v Nina>to use the Replit agent it's broken my app before so I just stay away from it

00:47:30.529 --> 00:47:34.980
<v Matt>Replit was the first one on the train, you would think it would be the best one.

00:47:35.450 --> 00:47:42.439
<v Nina>It's not good I don't even let it deploy I deploy from my command line because

00:47:42.439 --> 00:47:47.015
<v Nina>I just don't I don't want to touch it. It's the worst agent I've ever used in my life.

00:47:47.701 --> 00:47:50.779
<v Matt>I feel like if you're at the point, right? The same thing with Jev.

00:47:50.779 --> 00:47:53.262
<v Matt>Like, if you're at the point, you need this stuff.

00:47:53.869 --> 00:48:00.068
<v Matt>Like, I want to be vibe coding at, like, a multi-agent level on multiple boxes.

00:48:00.989 --> 00:48:06.170
<v Matt>You might be technical enough to spin up your own box and have your local agent

00:48:06.356 --> 00:48:11.789
<v Matt>SSH into these and kick off, like, what I do for the chief of staff stuff.

00:48:11.789 --> 00:48:16.489
<v Nina>I did that with Hetzner. That was something that I never had, I mean,

00:48:17.109 --> 00:48:21.534
<v Nina>I knew I could do it, but I just, if I have to write all that out by hand,

00:48:21.662 --> 00:48:29.421
<v Nina>like, or not by hand, but myself, I'm just, I'm so happy I have Claude to SSH. Yeah. Yeah.

00:48:29.740 --> 00:48:30.204
<v Matt>Makes sense.

00:48:30.778 --> 00:48:31.057
<v Nina>Yeah.

00:48:32.302 --> 00:48:36.604
<v Nina>True. But do we have time for your other point about small tickets?

00:48:37.428 --> 00:48:41.361
<v Matt>I haven't even gotten to that. I will do it next week. Okay.

00:48:41.361 --> 00:48:44.457
<v Matt>I'm spent. I've ranted too much today.

00:48:44.949 --> 00:48:47.944
<v Nina>Yeah. Well, this was a good one. Lots to talk about.

00:48:48.558 --> 00:48:53.249
<v Matt>Yeah. So, I want you to have ads or payments by next week.

00:48:53.843 --> 00:48:54.650
<v Nina>Okay.

00:48:55.412 --> 00:48:56.115
<v Matt>Make some money.

00:48:56.706 --> 00:49:00.352
<v Nina>I'll write it down. I swear. I promise.

00:49:01.431 --> 00:49:08.581
<v Nina>it'll just be one one little little banner in the corner that's fine all right

00:49:08.581 --> 00:49:09.912
<v Matt>All right that's the pod I guess.

