WEBVTT

00:00:19.693 --> 00:00:22.224
<v Nina>Hello, how are you doing, Matt?

00:00:22.886 --> 00:00:29.121
<v Matt>Doing good. Number two. We've got ourselves no longer a fluke. It's a coincidence, no?

00:00:29.556 --> 00:00:33.428
<v Nina>Yeah. Yeah, number two.

00:00:35.442 --> 00:00:38.124
<v Nina>How is your week going with building? What are you working on?

00:00:38.838 --> 00:00:42.664
<v Matt>It's good. I've been working on my podcast setup and failing miserably.

00:00:43.227 --> 00:00:48.219
<v Matt>I have one of those hub things to connect everything into, and then that would

00:00:48.219 --> 00:00:53.869
<v Matt>connect into my MacBook Pro but it's not it's like works only for power it's

00:00:53.869 --> 00:00:55.417
<v Matt>like not sending any data

00:00:56.433 --> 00:01:02.883
<v Matt>now i have enough i only have two micro usb ports on my air so i can only do

00:01:03.509 --> 00:01:06.064
<v Matt>power a screen and the camera

00:01:06.847 --> 00:01:12.759
<v Nina>Well no i just got a little hub too because i have a macbook air um and.

00:01:12.759 --> 00:01:15.595
<v Matt>Uh yes i mean link tool on your butt, because mine's not working.

00:01:16.511 --> 00:01:23.729
<v Nina>Yeah. Uh, it's an INTPW, that's the name of the brand. It works well for me.

00:01:23.729 --> 00:01:28.409
<v Nina>Like there, there are different separations between like giving powder or power

00:01:28.409 --> 00:01:30.939
<v Nina>and like transferring information.

00:01:31.339 --> 00:01:34.744
<v Nina>I don't know. I'll send you a link to a podcast, but, um.

00:01:35.853 --> 00:01:47.758
<v Nina>Yeah. So this week is programming a lot of side projects and migrated one of

00:01:47.758 --> 00:01:48.984
<v Nina>your projects this week.

00:01:50.738 --> 00:01:53.164
<v Nina>It was Next.js and Python.

00:01:55.132 --> 00:02:04.919
<v Nina>And yeah, it was the podcast search app that you have deployed and migrated it to a JVM.

00:02:05.064 --> 00:02:08.338
<v Nina>And figured out how to host and deploy my,

00:02:10.398 --> 00:02:15.898
<v Nina>first JVM or Java app. Had zero experience with deploying Java before,

00:02:15.898 --> 00:02:21.560
<v Nina>despite working in it for six years, which is very strange.

00:02:22.898 --> 00:02:27.788
<v Matt>Yes. So the concept of this one was, it basically takes a podcast and it turns

00:02:27.788 --> 00:02:30.465
<v Matt>it into a knowledge base like Claude can search against.

00:02:30.999 --> 00:02:37.658
<v Matt>And then like spit out like basically have a nicer search across a podcast more

00:02:37.658 --> 00:02:41.748
<v Matt>than just like it looks for titles and tells you which one it is and at best

00:02:41.748 --> 00:02:43.242
<v Matt>looks in the descriptions

00:02:44.194 --> 00:02:48.037
<v Matt>like this one kind of pulls out timestamps and things and

00:02:51.627 --> 00:02:53.088
<v Matt>What was the architecture before?

00:02:54.687 --> 00:02:57.407
<v Nina>It was Next and Python, and,

00:02:59.631 --> 00:03:01.482
<v Nina>it was hosted on Vercel.

00:03:03.007 --> 00:03:06.004
<v Matt>Was there even a server, or was it just a database?

00:03:07.987 --> 00:03:11.647
<v Nina>I think you just had just the... well, we had Vercel.

00:03:14.391 --> 00:03:17.097
<v Nina>Well, no, you had it set up in Next, I believe.

00:03:17.807 --> 00:03:20.347
<v Matt>Okay, yeah, so it was the next server on Vercel. Mm-hmm, on Vercel.

00:03:22.167 --> 00:03:29.507
<v Nina>It's been a bit. Yeah, so the app was really interesting to poke through just how it worked.

00:03:31.162 --> 00:03:39.614
<v Nina>So like 121 episodes of a specific podcast transcribed with Whisper,

00:03:40.148 --> 00:03:46.069
<v Nina>and then just the way that the search was set up in the app,

00:03:47.550 --> 00:03:52.213
<v Nina>Hosted on Vercel, everything would kind of load at once. It would have to boot

00:03:52.213 --> 00:03:54.295
<v Nina>the model every single time.

00:03:55.851 --> 00:04:04.253
<v Nina>So migrating things to JVM was probably one of the best architecture improvements for it.

00:04:05.013 --> 00:04:08.613
<v Matt>Yeah, I remember now. It was like Vercel's Lambda functions, whatever.

00:04:08.933 --> 00:04:12.773
<v Nina>Yes, it was. It was the internal Lambda functions. And it was totally inefficient

00:04:12.773 --> 00:04:13.913
<v Nina>for when loading models.

00:04:15.672 --> 00:04:17.293
<v Nina>Yeah, expensive.

00:04:19.053 --> 00:04:24.333
<v Nina>Um very expensive um yeah there uh,

00:04:25.756 --> 00:04:31.833
<v Nina>but you know so what moved mainly was the the serving api so just like the search

00:04:31.833 --> 00:04:34.983
<v Nina>uh asking questions caching um,

00:04:36.733 --> 00:04:42.613
<v Nina>streaming answers so there were next js route handlers and that that's what

00:04:42.613 --> 00:04:49.022
<v Nina>was migrated to to java spring um did it migrate the front end um,

00:04:49.864 --> 00:04:54.893
<v Nina>with claude it's it's interesting to like research to research up up front and

00:04:54.893 --> 00:04:58.853
<v Nina>figure out where where is the biggest benefit of migrating different sections

00:04:58.853 --> 00:05:04.753
<v Nina>of this um and i think like having a persistent application um,

00:05:05.633 --> 00:05:10.063
<v Nina>inside a serverless environment with 130 megabytes embedding a model is really

00:05:10.063 --> 00:05:17.901
<v Nina>bad so yeah um yeah uh cold starts are pretty pretty darn gross um,

00:05:18.882 --> 00:05:24.993
<v Nina>so yeah it it definitely was something that needed like a long running process

00:05:24.993 --> 00:05:26.405
<v Nina>and progressive streaming,

00:05:26.997 --> 00:05:34.033
<v Nina>and um in the process of migrating this um it's there's a bit of a bug,

00:05:34.653 --> 00:05:37.371
<v Nina>on my end and it's not actually streaming it's actually,

00:05:38.050 --> 00:05:44.573
<v Nina>um sending everything in like one lump sum still so i'm still learning um going

00:05:44.573 --> 00:05:46.705
<v Nina>through going through with claude but um.

00:05:48.585 --> 00:05:53.948
<v Nina>There were a couple improvements to it. Like, you know, like TypeScripts had

00:05:53.948 --> 00:05:56.764
<v Nina>two versions of the ask pipeline.

00:05:57.304 --> 00:06:02.441
<v Nina>So there was like one regular JSON and then another server sent events for streaming.

00:06:04.195 --> 00:06:07.808
<v Nina>And like there was actually a comment in the code reminding you that if you

00:06:07.808 --> 00:06:09.830
<v Nina>changed one, you had to remember to change the other.

00:06:12.204 --> 00:06:16.448
<v Nina>So it has one pipeline now. um and uh,

00:06:17.638 --> 00:06:21.591
<v Nina>yeah the logic for generating the answer exists once um,

00:06:22.160 --> 00:06:27.640
<v Nina>and it was really cool to also like like go back and forth on potential features,

00:06:28.209 --> 00:06:35.518
<v Nina>like given you have like 130 megabytes of data it's like looking at the app

00:06:35.518 --> 00:06:39.308
<v Nina>and anybody could go go on um i believe the url is,

00:06:39.998 --> 00:06:43.551
<v Nina>we could link it in the podcast, right? ETA brain.vercel.app.

00:06:44.271 --> 00:06:50.388
<v Nina>Um, yeah, it's, you just have one search and then you have the transcripts that

00:06:50.388 --> 00:06:55.225
<v Nina>you could search directly, and then you have, uh, common questions that people ask,

00:06:55.782 --> 00:06:59.631
<v Nina>but with 130 meg meg, uh, megabytes of, of data.

00:07:00.682 --> 00:07:06.837
<v Nina>You could definitely go through and semantically make like a tag cloud of some

00:07:06.837 --> 00:07:10.287
<v Nina>of the most weighted terms um and

00:07:10.287 --> 00:07:15.083
<v Nina>make that like a very interesting experience for the user um to kind of,

00:07:15.687 --> 00:07:22.197
<v Nina>look for look for specific things um in your case matt when you were collecting

00:07:22.197 --> 00:07:25.311
<v Nina>this data how long did it take you to,

00:07:26.199 --> 00:07:29.897
<v Nina>I guess convert 121.

00:07:30.274 --> 00:07:34.523
<v Matt>So did you rerun the conversions or are you using what was already in the database

00:07:34.634 --> 00:07:36.447
<v Nina>I'm using what was already in the database.

00:07:36.447 --> 00:07:47.045
<v Matt>Yeah okay that took about 10 hours and then basically what it did was it went through it took the

00:07:48.566 --> 00:07:54.336
<v Matt>podcast audio downloaded it onto my macbook or not my macbook the other one,

00:07:54.841 --> 00:07:55.973
<v Matt>the desktop Mac version.

00:07:56.350 --> 00:08:01.076
<v Matt>And then did a Hugging Face transcription

00:08:02.719 --> 00:08:06.004
<v Matt>And then on top of that, it did a

00:08:08.502 --> 00:08:14.412
<v Matt>Claude session to do the semantic understanding of it and then upload that as

00:08:14.412 --> 00:08:17.312
<v Matt>like a rag with the transcription timestamps.

00:08:18.322 --> 00:08:25.874
<v Matt>And so I think anything costs like $40 of API tokens and it took like 10 hours.

00:08:27.093 --> 00:08:28.132
<v Nina>Okay, wow.

00:08:30.872 --> 00:08:38.552
<v Nina>Are you speaking $44? like with like a like a pro account or do you mean like you went by usage.

00:08:38.790 --> 00:08:43.942
<v Matt>Or no there's something like i couldn't get it to work with my subscription

00:08:43.942 --> 00:08:49.232
<v Matt>thing i think because the way we're i was doing the lambda stuff in riverside

00:08:49.232 --> 00:08:52.182
<v Matt>i needed to do an api token um

00:08:52.842 --> 00:08:59.114
<v Matt>so i probably could have salt that But I didn't want to waste my time for 40 bucks.

00:08:59.729 --> 00:09:04.872
<v Nina>Yeah, so that's interesting because that's something that I've encountered in

00:09:04.872 --> 00:09:07.830
<v Nina>the most recent updates of Claude when developing with it.

00:09:09.729 --> 00:09:12.552
<v Nina>I have two apps, right? I have one

00:09:12.552 --> 00:09:18.541
<v Nina>app that actually loads the Claude code terminal and output in the app.

00:09:19.922 --> 00:09:24.382
<v Nina>And it's directly connected to my Claude code account. And then I recently tried

00:09:24.382 --> 00:09:34.185
<v Nina>making another app that grills you and gives you tech interview questions with a code editor.

00:09:35.262 --> 00:09:39.702
<v Nina>I'm in the job market right now for people listening, so I wanted to make something,

00:09:41.087 --> 00:09:48.808
<v Nina>that could evaluate my answers and grade them out of 10, so there's Claude underneath the app.

00:09:51.495 --> 00:09:57.865
<v Nina>And I noticed when I went into the Anthropic console that I had like a $5 spending

00:09:57.865 --> 00:10:01.466
<v Nina>limit and was actually directly consuming tokens.

00:10:02.505 --> 00:10:08.725
<v Nina>So it had set it up directly to the API instead of to like-

00:10:09.885 --> 00:10:12.682
<v Matt>It created an API token without even letting you know.

00:10:15.085 --> 00:10:20.465
<v Nina>I think I might have missed in the setup what was going on.

00:10:22.845 --> 00:10:29.625
<v Nina>But you can make a Claude app that just directly connects to your local instance

00:10:29.625 --> 00:10:31.884
<v Nina>of Claude without going through the API.

00:10:32.389 --> 00:10:34.172
<v Nina>That's also possible.

00:10:34.398 --> 00:10:39.745
<v Matt>I am not entirely sure where that stands in the terms of services land.

00:10:39.745 --> 00:10:45.445
<v Matt>Because that was the big thing when the crap thing came out, OpenClaude.

00:10:47.784 --> 00:10:52.045
<v Matt>But they kind of just didn't care. And I mean,

00:10:52.645 --> 00:11:00.368
<v Matt>I have my little chief of staff bot running in my subscription that spends like

00:11:00.856 --> 00:11:02.946
<v Matt>five minutes a day at most running.

00:11:04.060 --> 00:11:08.455
<v Matt>And so they're not going to care. I mean, they really shouldn't have that,

00:11:08.455 --> 00:11:09.964
<v Matt>just kind of small usage.

00:11:10.573 --> 00:11:14.178
<v Matt>But going through all those transcripts for 10 hours straight,

00:11:14.765 --> 00:11:18.207
<v Matt>I'm pretty sure my subscription could have got dinged for that.

00:11:19.484 --> 00:11:24.793
<v Nina>Yeah, probably if you had it directly hooked up to your sub. Yeah.

00:11:26.322 --> 00:11:29.323
<v Nina>Yeah, but those are the things that you learn when you're, I guess,

00:11:29.323 --> 00:11:32.253
<v Nina>developing these kinds of applications.

00:11:35.541 --> 00:11:41.137
<v Nina>Yeah, so I was draining money prepping for interviews to get a job to earn me money.

00:11:41.775 --> 00:11:43.615
<v Matt>The opposite you want to be doing during that time.

00:11:44.004 --> 00:11:45.061
<v Nina>Yeah, exactly.

00:11:47.249 --> 00:11:53.246
<v Nina>But I wonder for the future of AI in apps, I,

00:11:56.353 --> 00:12:03.383
<v Nina>wonder where that's going to go. Do you think that apps are going to use the

00:12:03.383 --> 00:12:07.362
<v Nina>Claude API, or do you think there are going to be some that say,

00:12:07.949 --> 00:12:13.138
<v Nina>sign up with your Claude subscription and and just log in with you oh yeah

00:12:13.504 --> 00:12:17.603
<v Nina>in the app so you can use your pro subscription within our app because there's

00:12:17.603 --> 00:12:20.813
<v Nina>going to be ai and everything and the companies are going to have to make like

00:12:20.813 --> 00:12:23.239
<v Nina>this determination right of.

00:12:23.361 --> 00:12:27.273
<v Matt>Yeah which is that's tough right now that's like the mcp

00:12:28.272 --> 00:12:34.356
<v Matt>like discussion like so you build apis right to have people connect to your

00:12:34.466 --> 00:12:39.696
<v Matt>ai bot or do you have like an ai chat bot that you like custom the agent for

00:12:40.410 --> 00:12:45.393
<v Matt>um and that's like really hard like i mean jira's a good example of this right

00:12:45.393 --> 00:12:50.528
<v Matt>jira has a shitty ai bot and it's useless and nobody likes it

00:12:51.300 --> 00:12:58.114
<v Matt>but someone built a cli around the api for jira and claude is very good at just using it

00:12:58.741 --> 00:13:02.423
<v Matt>and so like you get your tickets there and you can move things over pretty easily

00:13:02.423 --> 00:13:06.733
<v Matt>with that Jira or just like give Claude a link to your Jira to read tickets

00:13:06.733 --> 00:13:09.993
<v Matt>and do updates there instead of actually using the Jira MCP.

00:13:10.914 --> 00:13:11.315
<v Nina>Yeah.

00:13:11.947 --> 00:13:18.983
<v Matt>And I mean, it's puts the cost on the customers, so I'm surprised it's not pushed more.

00:13:20.430 --> 00:13:25.549
<v Matt>Of actually running these agents and keeping the agents like up to date and

00:13:25.549 --> 00:13:31.799
<v Matt>useful where like so it's like do what you're good at kind of mentality so just

00:13:31.799 --> 00:13:35.151
<v Matt>like have whoever like if they want to Claude they want to do

00:13:36.585 --> 00:13:42.179
<v Matt>um chat gpt like make that the client's problem and then you just host an api

00:13:42.179 --> 00:13:47.452
<v Matt>like you would good old web 2.0 style and let them figure it out

00:13:48.206 --> 00:13:54.857
<v Nina>Yeah yeah yeah there's also the side of if you do that you put it on the customer

00:13:55.228 --> 00:13:59.025
<v Nina>is the customer going to even want to use the product like do they even.

00:13:59.222 --> 00:14:05.389
<v Matt>You know yeah so jira like you're a tech user right so So it's no problem for

00:14:05.389 --> 00:14:12.922
<v Matt>me to pick up Claude by Notion, like a Notion thing, integrating.

00:14:13.230 --> 00:14:18.425
<v Matt>Most of those people aren't technical and have enough problems just using the web browser.

00:14:19.011 --> 00:14:22.599
<v Matt>So they're going to want a more comfortable UI to use.

00:14:22.802 --> 00:14:32.709
<v Matt>They tend to use Claude Cowork with MCPs sometimes to have the user interface

00:14:32.709 --> 00:14:33.884
<v Matt>that they're more familiar with.

00:14:34.435 --> 00:14:36.571
<v Matt>It's like you just hand them a terminal, but it's here.

00:14:38.199 --> 00:14:40.739
<v Matt>But it's just, you got to think.

00:14:42.829 --> 00:14:49.691
<v Matt>I still just don't want Notion to have their own agent. I want it to be Claude, or I want it to be

00:14:51.890 --> 00:14:55.692
<v Matt>ChatGPT. That's their job. Their job is to have good agents.

00:14:56.290 --> 00:14:56.731
<v Nina>Mm-hmm.

00:14:58.014 --> 00:15:03.464
<v Nina>Um i i there was a point where i was really frustrated with i don't i don't

00:15:03.464 --> 00:15:07.324
<v Nina>want to talk bad about notion but i was really frustrated about like the the

00:15:07.324 --> 00:15:12.486
<v Nina>paywall behind uh using their agents and um,

00:15:13.449 --> 00:15:18.364
<v Nina>like i did not want to pay you know 100 bucks a year and then on top of that

00:15:18.364 --> 00:15:23.486
<v Nina>extra to use their agents so i made an app that worked around,

00:15:24.467 --> 00:15:30.319
<v Nina>i made an app that worked around that um I think I called it like Novel Notion or something.

00:15:30.870 --> 00:15:38.562
<v Nina>And it was in Markdown and it had like a hierarchy of files,

00:15:38.829 --> 00:15:43.635
<v Nina>but there was a button to sync with Notion.

00:15:44.192 --> 00:15:49.294
<v Nina>So it would near the hierarchy of files and check against every single one that

00:15:49.294 --> 00:15:53.903
<v Nina>was in your Notion and make sure they were perfectly synced and up to date.

00:15:54.344 --> 00:16:00.114
<v Nina>Um and then just it was literally just a copy of notion if i'm being honest

00:16:00.114 --> 00:16:05.494
<v Nina>and then like in the inputs like on the pages you could type a backslash to

00:16:05.494 --> 00:16:07.237
<v Nina>access your local Claude so,

00:16:08.067 --> 00:16:16.037
<v Nina>local yeah so i i spent a whole day making something just so i could have my own a ai agent inside.

00:16:17.251 --> 00:16:25.308
<v Matt>Insight yeah i mean 100 bucks is nuts like that's what second to top pricing for anthropic

00:16:26.231 --> 00:16:34.277
<v Matt>and i i don't i have the top one that's like 200 and i barely hit the limit so there's no way

00:16:36.424 --> 00:16:42.814
<v Nina>With java last night i hit my limit i had to upgrade um no one i didn't realize

00:16:42.814 --> 00:16:44.517
<v Nina>java was such a token monster.

00:16:45.486 --> 00:16:47.640
<v Matt>Yeah, it's all those inputs and proboscity.

00:16:49.363 --> 00:16:50.764
<v Nina>All those types everywhere.

00:16:51.104 --> 00:16:57.797
<v Matt>Yeah yeah but it's just that's so like the margins on that hundred bucks gotta be great

00:16:59.916 --> 00:17:02.984
<v Nina>No it's i i believe i believe the perks are for.

00:17:02.984 --> 00:17:04.009
<v Matt>The business not for us

00:17:04.374 --> 00:17:09.674
<v Nina>Well yeah never mind yeah you could just sync across multiple devices that's

00:17:09.674 --> 00:17:16.784
<v Nina>that's the perk and you get like like two prompts of ai to organize information

00:17:16.784 --> 00:17:18.534
<v Nina>into a table or something Wait.

00:17:18.534 --> 00:17:24.674
<v Matt>And they meter the AI you're using after the 100? That's just greedy.

00:17:25.162 --> 00:17:31.559
<v Nina>Yeah, pretty greedy. But there's an alternative called Obsidian that I heard is.

00:17:31.780 --> 00:17:34.195
<v Matt>Exactly the same. Yeah, I've heard about Obsidian a lot. Yeah.

00:17:34.653 --> 00:17:40.876
<v Matt>Obsidian, Notion's aimed less at tech people, and Obsidian's more like,

00:17:41.358 --> 00:17:43.813
<v Matt>if you kind of know what a graph database is, you can have a better time.

00:17:48.380 --> 00:17:56.031
<v Nina>So yeah, what's something that you've been developing this week or just playing with AI?

00:17:56.414 --> 00:18:01.603
<v Matt>So I got my chief of staff thing, right, and it's giving me reports. It's still

00:18:04.764 --> 00:18:05.690
<v Matt>Mostly useful.

00:18:08.372 --> 00:18:12.308
<v Matt>The thing is, if it's not completely useful, it feels like a pain in the butt.

00:18:14.017 --> 00:18:23.912
<v Matt>I want it to give me more terse information about my PRs, but also useful information at the same time.

00:18:25.125 --> 00:18:29.642
<v Matt>When it's like, oh, that's 7 out of 10, I want to be like, why is it a 7 out of 10?

00:18:31.818 --> 00:18:34.927
<v Matt>Because most of the time something, it goes, and I'm just like,

00:18:34.927 --> 00:18:37.856
<v Matt>oh, you don't know the context of the app.

00:18:38.575 --> 00:18:43.138
<v Matt>So, like, what you're complaining about is something that's taken care of in another repo.

00:18:44.629 --> 00:18:45.987
<v Matt>Like, if I could get it to tell me

00:18:45.987 --> 00:18:50.852
<v Matt>that stuff in, like, the report that's building up, I would prefer that.

00:18:51.072 --> 00:18:56.687
<v Matt>And what I'm having issues with is, like, the division of what gets done on

00:18:56.687 --> 00:19:02.386
<v Matt>box on the client side and then what gets done in the chief of staff box.

00:19:03.817 --> 00:19:08.139
<v Matt>Um so the problem i was doing for a while was it was just trying to like

00:19:09.300 --> 00:19:15.327
<v Matt>get a diff and then like get us some nazis on the box then like in the chief

00:19:15.327 --> 00:19:19.302
<v Matt>of staff was trying to then crunch that into

00:19:20.713 --> 00:19:26.627
<v Matt>a smaller like digestible bit but that digestion needs to be done on the client

00:19:26.627 --> 00:19:29.838
<v Matt>box where it has all the contexts.

00:19:30.436 --> 00:19:37.477
<v Nina>Oh, okay. So it's just two different contexts kind of separated when doing evaluations.

00:19:38.197 --> 00:19:40.949
<v Matt>Yeah, because I'm keeping any code from getting off the box.

00:19:42.617 --> 00:19:45.580
<v Matt>It can only talk about it in abstract terms.

00:19:46.811 --> 00:19:51.697
<v Matt>And that's just two levels of abstraction. By the time it gets to me,

00:19:51.697 --> 00:19:55.170
<v Matt>it's just like, I just don't like this one. It's got some architecture stuff.

00:19:56.575 --> 00:20:01.532
<v Matt>I'm like, no, no, I need to know what the hell is the architecture problem you're running into.

00:20:02.339 --> 00:20:09.734
<v Nina>Yeah, so this is a terse report specifically for improving existing architecture

00:20:09.734 --> 00:20:15.384
<v Nina>in your applications. Like, is it, I guess, what's the...

00:20:15.384 --> 00:20:18.761
<v Matt>So the thing is, like, right, I got PRs on the box that I need to review.

00:20:19.644 --> 00:20:22.624
<v Matt>Like, I could just easily get, like, a little report saying,

00:20:22.624 --> 00:20:24.549
<v Matt>yeah, I have these things, click these links, go to it.

00:20:25.060 --> 00:20:28.310
<v Matt>But that's basically what I get in the email anyways.

00:20:29.544 --> 00:20:35.398
<v Matt>What I kind of want is the quick Claude review. Give this thing a score out of 10.

00:20:37.004 --> 00:20:42.961
<v Matt>Give me what the PR is about. Give me why it's not a 10.

00:20:45.504 --> 00:20:50.246
<v Matt>What's your main objections? So that's basically the whole thing.

00:20:50.879 --> 00:20:56.104
<v Matt>And so getting it's I got everything like structured wise because I'm always

00:20:56.104 --> 00:21:00.344
<v Matt>good about that like I always I never just kind of like send blobs over it's

00:21:00.344 --> 00:21:03.232
<v Matt>got to be like JSON formatted at the very least

00:21:03.923 --> 00:21:08.534
<v Matt>so like Claude always does better when like you switch from agent to agent and

00:21:08.534 --> 00:21:10.790
<v Matt>you give it JSON instead of just text

00:21:12.948 --> 00:21:16.618
<v Matt>So I do it that way. And so it's there, like it's almost there,

00:21:17.007 --> 00:21:22.028
<v Matt>just like where it's getting processed doesn't have the context to build useful

00:21:22.028 --> 00:21:25.355
<v Matt>information in the structure at the moment. Yeah.

00:21:28.849 --> 00:21:34.567
<v Nina>So when you're developing with multiple agents, you know, kind of,

00:21:34.892 --> 00:21:39.168
<v Nina>so you're saying that one kind of generates the prompt for the other agent or

00:21:39.168 --> 00:21:43.813
<v Nina>the format. that yeah yeah so um.

00:21:44.458 --> 00:21:49.148
<v Matt>I guess it does allow for some back and forth like it's allowed three turns

00:21:49.148 --> 00:21:53.678
<v Matt>basically like if it gets something it's just like oh what's this and it can

00:21:53.678 --> 00:21:55.991
<v Matt>ask again for some more information

00:21:56.973 --> 00:22:01.693
<v Nina>Yeah so like follow it three max follow-ups to each other essentially yeah.

00:22:02.255 --> 00:22:02.673
<v Matt>Basically.

00:22:03.696 --> 00:22:06.868
<v Nina>Yeah. So how are you building this? How are you building,

00:22:09.071 --> 00:22:12.988
<v Nina>this terse report system? How is this being confused?

00:22:12.988 --> 00:22:13.796
<v Matt>Yeah, it's

00:22:16.908 --> 00:22:21.093
<v Matt>Java server, right? And then it's the Claude instances on the box.

00:22:21.510 --> 00:22:27.792
<v Matt>Like we said last week, it's just my Claude subscriptions on the box and they're on GPNs.

00:22:28.140 --> 00:22:30.427
<v Matt>And then that's all within my company's VPNs.

00:22:31.786 --> 00:22:37.177
<v Matt>As well so and then that talks to me through the company slack

00:22:38.018 --> 00:22:45.698
<v Matt>and uh how it works behind the scenes like under the code is java server um

00:22:45.698 --> 00:22:50.220
<v Matt>using uh spring ai and that has like special

00:22:51.701 --> 00:22:57.895
<v Matt>components to do different types of talking to Claude back and forth um i rely

00:22:57.895 --> 00:23:04.228
<v Matt>a lot more on just like kind of open pipeline of it and i just do some like

00:23:05.476 --> 00:23:09.493
<v Matt>the open pipeline just to give free text to it that's with json inside of it

00:23:10.242 --> 00:23:14.619
<v Matt>and then i've kind of just wrapped that around some simple java like cron jobs

00:23:15.188 --> 00:23:17.481
<v Matt>have you done like spring boot batch or anything like that

00:23:17.957 --> 00:23:22.625
<v Nina>Um i've done external cron jobs that hit a public API before for.

00:23:22.985 --> 00:23:26.775
<v Matt>Apps. Yeah, it's basically the same stuff, just add scheduled jobs,

00:23:26.775 --> 00:23:29.381
<v Matt>and then you give it a little cron in there for when it's running.

00:23:31.419 --> 00:23:34.549
<v Matt>And it's been, it's been good. Like the structure is fine. Cause that's like

00:23:34.549 --> 00:23:37.919
<v Matt>all like 20 year old technology. Like people have been running jobs forever.

00:23:37.919 --> 00:23:38.946
<v Matt>I don't have to worry about that.

00:23:39.474 --> 00:23:39.933
<v Nina>Yeah.

00:23:40.740 --> 00:23:47.259
<v Matt>And not doing any custom code there, like really laying it on everything but

00:23:47.259 --> 00:23:49.958
<v Matt>the agents to pull out the data.

00:23:50.748 --> 00:23:53.668
<v Nina>Yeah. Yeah. Interesting. Yeah.

00:23:53.795 --> 00:23:59.037
<v Nina>It's the, one of the things that I was encountering when I was migrating the, uh, the podcast search,

00:24:00.198 --> 00:24:05.899
<v Nina>was I, you know, I was torn between doing multi agentic, you know,

00:24:05.899 --> 00:24:09.672
<v Nina>development with it because a portion of this migration was,

00:24:10.009 --> 00:24:14.639
<v Nina>I also wanted to be learning from, you know, working with, um,

00:24:15.339 --> 00:24:17.439
<v Nina>working with a new Java app. So, um,

00:24:22.139 --> 00:24:27.801
<v Nina>I was kind of torn between, do I have an agent for the infrastructure,

00:24:28.248 --> 00:24:33.939
<v Nina>the database, testing, manager, separate them on work trees?

00:24:33.939 --> 00:24:37.569
<v Nina>Is that going to be like setting a fire at multiple parts of the app if they

00:24:37.569 --> 00:24:39.102
<v Nina>have to touch the same thing?

00:24:40.640 --> 00:24:42.986
<v Nina>But it took me a lot oh go ahead.

00:24:43.599 --> 00:24:50.509
<v Matt>I've had no success with multi-agent development I've only had success for multi-agents for reviewing things

00:24:52.831 --> 00:24:58.479
<v Matt>like even being like okay do the front end side with one agent do the back end

00:24:58.479 --> 00:25:02.659
<v Matt>side with an agent it just melts and then it can't

00:25:03.680 --> 00:25:07.488
<v Matt>communicate with each other and I feel it just blows through tokens way faster

00:25:08.075 --> 00:25:09.915
<v Matt>than just taking the time.

00:25:10.606 --> 00:25:16.086
<v Matt>So I just let her run and go get a coffee instead of trying to parallelize the

00:25:17.362 --> 00:25:20.299
<v Matt>agents have you had success with it

00:25:20.920 --> 00:25:28.090
<v Nina>I haven't attempted to do it uh i mean i've i've if the contracts between the

00:25:28.090 --> 00:25:34.131
<v Nina>front and back end are so clear i mean i think then then i've had success with it but,

00:25:35.210 --> 00:25:41.010
<v Nina>um like if one is typing up documentation or one is doing tests that's fine

00:25:41.010 --> 00:25:44.340
<v Nina>because nothing is going to get mutated like right in the middle they're not

00:25:44.340 --> 00:25:45.439
<v Nina>going to touch the same file.

00:25:46.235 --> 00:25:54.670
<v Nina>But I was going to with another personal project where I want to migrate it

00:25:54.670 --> 00:26:00.143
<v Nina>to a more lightweight library like Svelte or framework like Svelte or something.

00:26:00.811 --> 00:26:04.892
<v Nina>I was thinking I do want to have multi-agents.

00:26:06.564 --> 00:26:10.888
<v Nina>And I'm sure there are a lot of people online that have guidelines about how to do that, but,

00:26:13.123 --> 00:26:16.966
<v Nina>I imagine it's very, very difficult unless you have,

00:26:19.237 --> 00:26:25.027
<v Nina>Kind of how you programmed with the terse report where the agents can kind of

00:26:25.027 --> 00:26:29.277
<v Nina>convene and go back and forth for a max of three times before they go off and do their work.

00:26:30.571 --> 00:26:33.253
<v Nina>Like maybe something like that would make it safer, but no.

00:26:34.541 --> 00:26:38.680
<v Matt>Yeah, that's like different agents with different domains. That's not development.

00:26:42.523 --> 00:26:46.947
<v Matt>It's not building software, my multi-agents. They're just reading code.

00:26:47.092 --> 00:26:51.767
<v Nina>Right, they're just doing review. for sure but but they do have the ability

00:26:51.767 --> 00:26:57.177
<v Nina>to like you do set set it up where they can follow up ask each other questions so like maybe,

00:26:57.797 --> 00:27:03.166
<v Nina>maybe there's a way to program with agents and i was going to test this out where you can like,

00:27:03.804 --> 00:27:10.057
<v Nina>um set them set both off at the same time and have like a five minute window

00:27:10.417 --> 00:27:16.567
<v Nina>where they both plan together and then executes and make sure the contract is

00:27:16.767 --> 00:27:18.799
<v Nina>separated while they're building like,

00:27:19.571 --> 00:27:23.884
<v Nina>and then they have to reconvene and stop coding i think,

00:27:24.969 --> 00:27:30.147
<v Nina>if i can figure out a way to get multiple agents to go between plan mode and

00:27:30.147 --> 00:27:34.437
<v Nina>development mode at the same time and they they wait for each other to reach

00:27:34.437 --> 00:27:36.421
<v Nina>certain milestones and like,

00:27:37.373 --> 00:27:42.927
<v Nina>i've never used an application that kind of automates um agent workflow if you

00:27:42.927 --> 00:27:50.371
<v Nina>have multiple workers i'm sure i'm sure those agentic workflow uh applications do that but.

00:27:51.491 --> 00:27:54.179
<v Matt>Have you tried ultra code and

00:27:54.957 --> 00:27:56.367
<v Nina>I i have not.

00:27:56.774 --> 00:28:05.301
<v Matt>I have not so it's it's just a effort level in Claude like you have high max and then ultra code

00:28:06.096 --> 00:28:12.105
<v Matt>uh i forgot it used to be something that was different or they needed a rainbow but anyways um

00:28:14.213 --> 00:28:17.382
<v Matt>It likes to do agent pipelines like that.

00:28:18.259 --> 00:28:23.160
<v Matt>And it does a little bit of that. It doesn't do completely like you're talking

00:28:23.160 --> 00:28:27.969
<v Matt>about with like two agents as peers planning. It more does like an orchestrator

00:28:28.596 --> 00:28:32.770
<v Matt>and then like kicks one off next, kicks one off next more to keep context clean.

00:28:33.800 --> 00:28:36.456
<v Matt>I don't even know if that's as important as it used to be.

00:28:38.053 --> 00:28:42.482
<v Matt>It was back when like context dirtying was a big problem.

00:28:43.312 --> 00:28:47.870
<v Matt>And i think that's the problem that's trying to fix or another tool that i used

00:28:47.870 --> 00:28:55.240
<v Matt>to use was this like orchestrator same pattern but it was you'd have the orchestrator

00:28:55.240 --> 00:28:56.327
<v Matt>right that's always running

00:28:57.215 --> 00:29:01.911
<v Matt>and you that's what you communicate with it was basically like the pm for your project

00:29:02.770 --> 00:29:07.400
<v Matt>and then you it would go and then if you like came up with a feature it would

00:29:07.400 --> 00:29:11.530
<v Matt>then kick off like a project manager

00:29:12.296 --> 00:29:18.856
<v Matt>and then the project manager would spin off like an architect a programmer and a reviewer and a qa

00:29:19.837 --> 00:29:24.684
<v Matt>and like each of those would each do its own thing report back up and then like

00:29:25.217 --> 00:29:29.234
<v Matt>slowly bubble back up to the orchestrator and you'd have something done

00:29:30.378 --> 00:29:36.003
<v Matt>that would just destroy your tokens i mean you just burn tokens left and right doing that

00:29:36.397 --> 00:29:42.272
<v Nina>Yeah that's what i was gonna bring up that sounds like you're burning money at that point.

00:29:42.794 --> 00:29:46.208
<v Matt>And you didn't get much out of it it just looked cool

00:29:46.783 --> 00:29:52.510
<v Nina>Yeah right like uh one one uh workflow automation uh app i really wanted to

00:29:52.510 --> 00:29:55.300
<v Nina>try what was the name of that one again that one was called.

00:29:55.800 --> 00:30:01.480
<v Matt>Uh i forgot the guy like who built it changed the name twice and now i think

00:30:01.480 --> 00:30:08.139
<v Matt>he has something that's not open source anymore. So I'll look for something in the description.

00:30:08.505 --> 00:30:13.149
<v Matt>I can now put links in the description. YouTube is keeping me from last episode doing that.

00:30:13.700 --> 00:30:14.879
<v Nina>Oh, oh, weird.

00:30:15.620 --> 00:30:16.910
<v Matt>Gotta give them my picture.

00:30:18.262 --> 00:30:20.938
<v Matt>Some AI, like, verified it was human.

00:30:22.233 --> 00:30:28.361
<v Nina>Wow. I haven't set up a new YouTube account in many years. Didn't know that was now a thing.

00:30:30.051 --> 00:30:32.807
<v Nina>N8N is the workflow out of me.

00:30:33.509 --> 00:30:39.051
<v Matt>Yeah, that's the pipeline one. It's kind of like Zapier.

00:30:39.111 --> 00:30:42.054
<v Matt>Are you reading Zapier?

00:30:43.002 --> 00:30:44.841
<v Matt>Are they even older ones, Yahoo Pipelines?

00:30:46.908 --> 00:30:52.611
<v Nina>No i never see i never i never uh i wasn't doing a lot of pipeline engineering

00:30:52.611 --> 00:30:55.731
<v Nina>throughout my i was i was mostly doing just mean,

00:30:57.340 --> 00:31:06.087
<v Nina>well i guess i guess um yeah so the the n8n is yeah,

00:31:07.251 --> 00:31:11.081
<v Nina>it's it i like it because it has like this nice visual node canvas so i mean

00:31:11.081 --> 00:31:14.609
<v Nina>i was debating getting that set up to automate like,

00:31:15.630 --> 00:31:21.321
<v Nina>i don't know clearing of contacts um and also like all the mcps that you have

00:31:21.321 --> 00:31:27.251
<v Nina>in there for like checking checking email and i just want to automate my job search,

00:31:28.211 --> 00:31:32.394
<v Nina>yeah that's that's what i want to do i want to i just want to web scrape um

00:31:32.678 --> 00:31:36.069
<v Nina>everything from indeed and linkedin and,

00:31:39.201 --> 00:31:44.321
<v Nina>It seems like that's actually really affecting, AI is really affecting the job

00:31:44.321 --> 00:31:48.481
<v Nina>market right now in terms of competitiveness, and I can't imagine how many recruiters

00:31:48.481 --> 00:31:51.150
<v Nina>are looking at the same AI-generated cover letters.

00:31:52.172 --> 00:31:55.521
<v Matt>I mean, they're just having AI do it, right? Is it doing AI interviews?

00:31:57.321 --> 00:32:00.658
<v Nina>Yeah. I did one AI interview.

00:32:00.833 --> 00:32:02.174
<v Matt>Oh, yeah, you already had two ones?

00:32:02.859 --> 00:32:09.341
<v Nina>Yeah it was it was like for some like tiny annotation gig i was just curious

00:32:09.341 --> 00:32:14.776
<v Nina>because like on linkedin they had a bunch of data annotation gigs and like,

00:32:15.937 --> 00:32:21.321
<v Nina>that's a hole that you don't want to get into right but um i was just curious

00:32:21.321 --> 00:32:25.590
<v Nina>it's and it said ai interview i just wanted to experience it um,

00:32:26.681 --> 00:32:30.971
<v Nina>i hate it i i don't like it um,

00:32:33.415 --> 00:32:38.412
<v Nina>Because all you can really do is just hope to say the right keyword and hope

00:32:38.412 --> 00:32:41.420
<v Nina>that you're not rejected. I don't know.

00:32:45.522 --> 00:32:46.836
<v Nina>That's a little too weird for me.

00:32:50.127 --> 00:32:53.906
<v Matt>It's like experience you expect a person. It's like white chatbots aren't great.

00:32:56.066 --> 00:33:02.572
<v Matt>Kind of okay. but after a while you even like club terminal it's just like you

00:33:02.572 --> 00:33:07.328
<v Matt>notice it's a bot's writing it's got a certain style to it and feel

00:33:09.533 --> 00:33:14.131
<v Matt>no matter what you do it's really hard to get rid of that flavor from now on

00:33:15.089 --> 00:33:22.682
<v Nina>Yeah yeah so it's yeah i like i like talking with a person i don't like customer

00:33:22.682 --> 00:33:26.645
<v Nina>agents and chatbots like that personally,

00:33:27.748 --> 00:33:32.562
<v Nina>if i had a company or something i would definitely have to have them but i don't

00:33:32.562 --> 00:33:35.213
<v Nina>want to deal with them myself um,

00:33:36.351 --> 00:33:42.962
<v Nina>so yeah i'm i'm looking i'm looking at like uh how many people are just auto

00:33:42.962 --> 00:33:48.304
<v Nina>applying with agents to the same jobs in a span of two minutes.

00:33:49.233 --> 00:33:52.772
<v Matt>At least I mean there weren't before there used to be the old just scripts they

00:33:52.772 --> 00:33:57.182
<v Matt>would run through like the LinkedIn easy apply and click the buttons because

00:33:57.182 --> 00:33:58.370
<v Matt>you would already have it all set up

00:34:00.587 --> 00:34:03.455
<v Nina>Yeah I have I have done that before but,

00:34:05.922 --> 00:34:10.452
<v Nina>I have for the easy apply for sure but not anything if there's the option to

00:34:10.452 --> 00:34:13.846
<v Nina>add a cover letter you know I'm going to actually write a cover letter,

00:34:17.391 --> 00:34:23.254
<v Nina>but um yeah everything everything is is bots now and um,

00:34:25.100 --> 00:34:28.594
<v Nina>Yeah, how do you feel about that?

00:34:28.594 --> 00:34:30.964
<v Matt>For the interview stuff? It's not great.

00:34:32.264 --> 00:34:38.312
<v Nina>Really, across all different platforms, all different. How do you feel about that?

00:34:40.954 --> 00:34:42.829
<v Matt>Yeah. I'm using bots a lot myself.

00:34:43.809 --> 00:34:44.163
<v Nina>Right.

00:34:45.844 --> 00:34:49.329
<v Matt>But those are my own personal bots. One being forced upon me.

00:34:50.213 --> 00:34:55.513
<v Matt>I don't even like doing the talk to ChatGPT or top-to-clawed ones.

00:34:56.163 --> 00:34:58.112
<v Matt>We figured those would be the best ones.

00:35:00.524 --> 00:35:07.528
<v Matt>I haven't had much interaction with customer service ones or call center ones.

00:35:08.364 --> 00:35:10.204
<v Matt>I hear those are becoming more and more popular.

00:35:12.254 --> 00:35:16.374
<v Matt>So mine are pretty custom and built by me, and I like mine, and mine are good

00:35:16.374 --> 00:35:19.050
<v Matt>for me. That's where I'm really in.

00:35:19.857 --> 00:35:25.685
<v Nina>Right, yeah. Yeah, personal development bots for efficiency are definitely the,

00:35:26.619 --> 00:35:30.491
<v Nina>definitely the ones that you know i'm okay with that's just like,

00:35:31.426 --> 00:35:35.774
<v Nina>when i'm going you know when i'm calling somebody or i'm trying to talk to a

00:35:35.774 --> 00:35:39.686
<v Nina>person for a specific thing i do not like them um,

00:35:40.517 --> 00:35:45.184
<v Nina>but yeah i i needed to update my skills on linkedin And I was like,

00:35:45.664 --> 00:35:51.744
<v Nina>I am not going to retype the same skills like over again.

00:35:51.744 --> 00:35:59.104
<v Nina>So I set off an agent to transfer my resume skills right in the browser. Um,

00:36:01.606 --> 00:36:07.307
<v Nina>And, you know, it's made things so much easier to do that.

00:36:09.025 --> 00:36:12.565
<v Matt>Yeah, that's just work, right? You're not doing anything creative.

00:36:13.589 --> 00:36:14.029
<v Nina>No, no.

00:36:14.029 --> 00:36:17.470
<v Matt>It's like when creative comes in, it's like it doesn't work out.

00:36:18.526 --> 00:36:21.629
<v Matt>I've been trying to write an article about this chief of staff,

00:36:21.629 --> 00:36:25.440
<v Matt>and it's just garbage. I don't like it at all.

00:36:26.449 --> 00:36:32.731
<v Matt>It just sounds like a bot wrote it. So I got to get it to just write an outline

00:36:33.050 --> 00:36:37.770
<v Matt>of what we did, and then I go in and write everything myself.

00:36:39.117 --> 00:36:43.409
<v Nina>So you mean you've been trying to write a blog article, essentially?

00:36:44.049 --> 00:36:50.209
<v Matt>Yeah, explaining what I did and stuff. It was good at kind of structuring it

00:36:50.209 --> 00:36:54.018
<v Matt>and then generating images to show the architecture.

00:36:55.063 --> 00:36:59.835
<v Matt>But the actual text just it sounds too ai to me i can't publish it

00:37:00.978 --> 00:37:05.249
<v Matt>right for some reason like i have older ones that are just like no this is just

00:37:05.249 --> 00:37:09.941
<v Matt>a new thing so i like ai Claude did it but those like Sonnet 3

00:37:10.578 --> 00:37:14.379
<v Matt>and those don't feel as ai to me as the ones that come out now maybe i'm just

00:37:14.379 --> 00:37:16.465
<v Matt>more experienced maybe i need to go back and read those

00:37:17.556 --> 00:37:21.479
<v Nina>Yeah i mean there there's if you're looking for you know models for create,

00:37:22.309 --> 00:37:26.009
<v Nina>like creative writing or something there are plenty on a hugging face that are

00:37:26.009 --> 00:37:34.089
<v Nina>attuned specifically for like writing blog articles or sounding more human um but,

00:37:35.012 --> 00:37:40.369
<v Nina>yeah i've i've also tried doing that for a blog post or,

00:37:41.356 --> 00:37:46.489
<v Nina>any type of post and it they they have the same they say the same things like

00:37:46.489 --> 00:37:49.745
<v Nina>every single time you know you're trying to talk about like,

00:37:50.250 --> 00:37:53.779
<v Nina>okay I'm just going to say for work experience because I wanted a blurb you

00:37:53.779 --> 00:37:57.982
<v Nina>know for my resume that was formatted in a certain way um,

00:37:59.445 --> 00:38:06.998
<v Nina>Yeah, they'll use the same keywords every time, and the sentence structure is

00:38:06.998 --> 00:38:09.649
<v Nina>just very convoluted and almost run-on.

00:38:10.392 --> 00:38:15.547
<v Matt>It's supposed to be probabilities, right? So it's just using the biggest distribution.

00:38:16.284 --> 00:38:20.448
<v Nina>But you're right. When it comes to structuring, let's say an essay,

00:38:20.808 --> 00:38:26.108
<v Nina>and you have five paragraphs, and the way it segues into each one and kind of

00:38:26.108 --> 00:38:29.003
<v Nina>structures all the themes, like bullet points, bam, bam, bam.

00:38:29.728 --> 00:38:31.534
<v Nina>Um it's very good at that.

00:38:31.940 --> 00:38:36.688
<v Matt>For sure yeah which is good for me because i like staring at a black blank page

00:38:36.688 --> 00:38:43.132
<v Matt>i'll just run away where just like fill in this outline i'll be able to work a little bit better

00:38:43.683 --> 00:38:44.984
<v Nina>And change it afterwards right.

00:38:45.698 --> 00:38:51.478
<v Matt>Yeah like then then have the i like go through grammar just like fix things

00:38:51.478 --> 00:38:54.109
<v Matt>up in spelling and punctuation whenever

00:38:55.335 --> 00:38:58.801
<v Matt>But it's just so bad at writing.

00:38:59.253 --> 00:39:05.984
<v Nina>It is very bad at writing. However, it is really good at,

00:39:08.164 --> 00:39:12.364
<v Nina>you know, I think I mentioned this app last time, and I mentioned this app this

00:39:12.364 --> 00:39:15.874
<v Nina>time. I, like, called this app the right help, right, where you have AI built

00:39:15.874 --> 00:39:17.288
<v Nina>into, like, a document editor.

00:39:18.007 --> 00:39:24.642
<v Nina>Um one of the things that it can do is um it has you know it can generate like,

00:39:25.408 --> 00:39:30.804
<v Nina>you know synonyms and stuff like right right off the bat like and it'll evaluate

00:39:30.804 --> 00:39:36.294
<v Nina>for tone um and make sure everything you know could you can modify the tone

00:39:36.294 --> 00:39:39.864
<v Nina>of a paragraph and that's kind of nice like,

00:39:40.404 --> 00:39:44.768
<v Nina>it It's nice when you're doing very, like, specific things or subtle,

00:39:46.637 --> 00:39:51.104
<v Nina>yeah, just very, very subtle things like changing, like, sentence structure

00:39:51.104 --> 00:39:53.744
<v Nina>when you give it explicit instructions.

00:39:53.744 --> 00:40:01.382
<v Nina>But, yeah, I found, like, this synonym tool that it has is pretty useful.

00:40:03.112 --> 00:40:07.737
<v Nina>Like, just, it's good when you're completely blocked.

00:40:08.579 --> 00:40:15.264
<v Nina>You have a blank page um you know a blank code editing you just need the structure

00:40:15.264 --> 00:40:22.389
<v Nina>of something to get going so whether that's code or writing um it's good for boilerplating,

00:40:23.109 --> 00:40:24.142
<v Nina>that's what i'm trying to say.

00:40:24.937 --> 00:40:28.014
<v Matt>Yeah yeah i mean i just like telling someone how to fix something more than

00:40:28.014 --> 00:40:30.667
<v Matt>actually making something new myself so

00:40:33.575 --> 00:40:40.784
<v Nina>Well i mean i did have i i did enjoy improving the the eta brain app i do like

00:40:40.784 --> 00:40:46.515
<v Nina>improving stuff um but i don't feel like the improvements this round were as,

00:40:47.904 --> 00:40:53.080
<v Nina>where i wanted them to be i still have more i think i want to do to the migration

00:40:53.301 --> 00:40:57.782
<v Nina>um like it would be nice to get rid of Vercel altogether by the way,

00:40:58.897 --> 00:40:59.650
<v Nina>and just have things.

00:40:59.650 --> 00:41:06.848
<v Matt>Yeah, but you got it all in the Hetzner box now. I don't see why I need to keep Vercel.

00:41:07.852 --> 00:41:08.137
<v Nina>Yeah.

00:41:08.613 --> 00:41:13.344
<v Matt>Other than I got some, I got the domains like set up and things like that.

00:41:14.279 --> 00:41:15.921
<v Matt>Yeah. Let me just change that over.

00:41:17.082 --> 00:41:23.760
<v Nina>Yeah, I think pulling everything the UI or the user experience out of Vercel

00:41:23.760 --> 00:41:27.560
<v Nina>is like the next step in a new box and stuff.

00:41:27.940 --> 00:41:31.078
<v Matt>So, It shouldn't be too rough.

00:41:32.160 --> 00:41:35.101
<v Matt>And that's pretty structured. Yeah, I can do that very simply.

00:41:35.647 --> 00:41:38.474
<v Matt>Yeah, I don't do any infrastructure stuff anymore.

00:41:39.540 --> 00:41:44.600
<v Matt>Also, so little new stuff to do in infrastructure work. I'm just like,

00:41:44.600 --> 00:41:48.940
<v Matt>just set up this box and go figure it out. Give it my API token.

00:41:49.660 --> 00:41:50.142
<v Nina>Mm hmm.

00:41:52.667 --> 00:41:53.317
<v Nina>Yeah, so.

00:41:54.983 --> 00:42:02.639
<v Nina>Yeah, I, that, that, uh, yeah, I wish I, I definitely learned a lot from migrating that.

00:42:03.022 --> 00:42:07.094
<v Nina>And I think up front, there's so many things I wish, wish I could have done.

00:42:07.094 --> 00:42:11.079
<v Nina>I wish I could have said, we need the VPN up front or we need, you know.

00:42:12.944 --> 00:42:16.004
<v Matt>Oh, yeah. I mean, that's like regular development, though. You didn't know that

00:42:16.004 --> 00:42:18.515
<v Matt>stuff until you hit the wall. You're just like, why is this not working?

00:42:19.450 --> 00:42:22.631
<v Matt>It's like, ah, yeah, I forgot to tell you about the VPN that exists.

00:42:23.392 --> 00:42:23.839
<v Nina>Yeah.

00:42:24.802 --> 00:42:28.657
<v Matt>Yeah. Yeah, it's normal software. That's a people problem.

00:42:29.724 --> 00:42:34.884
<v Nina>You're right. You're right. Yeah. That's what I would run into that anyway.

00:42:35.724 --> 00:42:39.634
<v Matt>Yeah, you can't do that all up front. If that worked, then people would still be doing waterfall.

00:42:40.377 --> 00:42:46.784
<v Nina>Yeah. And the proxy and the changing the environment files, all that stuff.

00:42:48.852 --> 00:42:52.254
<v Nina>Dealing with a bug in caddy so everything's returning a lump sum.

00:42:52.254 --> 00:42:59.741
<v Nina>Yeah, those are things that you don't run into until you're in it. So,

00:43:01.918 --> 00:43:02.429
<v Nina>yeah.

00:43:04.524 --> 00:43:06.806
<v Matt>Well, what else do you see? See your notes?

00:43:10.264 --> 00:43:13.075
<v Nina>My notes are all over the board. I put together the notes last night,

00:43:13.296 --> 00:43:16.140
<v Nina>and I think next time I'm going to be earlier.

00:43:16.744 --> 00:43:20.134
<v Matt>Nah. You got to keep it fresh. People want the news. Yes.

00:43:23.675 --> 00:43:27.047
<v Matt>Um yeah yeah we got everything mom

00:43:27.918 --> 00:43:34.425
<v Matt>yeah what else oh so one thing it did yeah i did do well is i put um put the

00:43:34.425 --> 00:43:40.770
<v Matt>podcast transcript from last week that's how i got the title and it was good title i liked it

00:43:41.664 --> 00:43:43.365
<v Nina>Nothing is intimidating anymore.

00:43:43.365 --> 00:43:47.643
<v Matt>Right yeah that was sweet and sweet title i was like damn i don't even remember Nina saying that

00:43:48.119 --> 00:43:57.001
<v Nina>I think i yeah i think i did say that i i was i I was having a sky's the limit moment. Yeah.

00:43:58.375 --> 00:43:59.664
<v Matt>It's like so profound.

00:44:03.333 --> 00:44:11.245
<v Nina>It's very it's no i feel like i feel like an optimistic bob the builder or something no it's like uh,

00:44:14.379 --> 00:44:18.815
<v Nina>but no it's it is true though like anybody can build anything now and that's

00:44:18.815 --> 00:44:21.752
<v Nina>what's really nice like when i was going back and forth with um,

00:44:22.361 --> 00:44:29.315
<v Nina>again not never had to think about like memory and ram when it comes to hosting

00:44:29.315 --> 00:44:35.852
<v Nina>a jvm i could just ask it questions like hey is this enough with 130 megabyte models is this,

00:44:36.456 --> 00:44:38.139
<v Nina>what are we going to run into and,

00:44:38.917 --> 00:44:43.085
<v Nina>it actually it spit out ways to optimize um,

00:44:43.898 --> 00:44:49.389
<v Nina>running like on a cheaper cheaper uh instance you know a cheaper um container,

00:44:50.190 --> 00:44:56.425
<v Nina>with less power to it and it was like oh well if you do a a swap of two gigabytes

00:44:56.825 --> 00:45:04.022
<v Nina>of data they will be fine and it gave me all the different options so i would have never been able to,

00:45:04.945 --> 00:45:10.145
<v Nina>do that before unless i googled for hours when i used.

00:45:11.705 --> 00:45:18.889
<v Matt>To be my job is just optimizing JVMs it's like my first joke it was miserable

00:45:19.911 --> 00:45:22.855
<v Nina>You should use the the tool that

00:45:22.855 --> 00:45:30.441
<v Nina>you have to evaluate the code that I put together the report tool that,

00:45:33.100 --> 00:45:37.302
<v Nina>i feel like i feel like it's it's gonna have some things to say um,

00:45:37.999 --> 00:45:43.496
<v Nina>because i was i didn't have much input when it came to like my own input about,

00:45:44.089 --> 00:45:50.429
<v Nina>optimizing like jvm so what was that what was that like like before before claude.

00:45:50.429 --> 00:45:57.739
<v Matt>You just have uh uh you had this like thing called jcat i believe which like

00:45:57.739 --> 00:46:01.543
<v Matt>listened on to the garbage collection and like jvm metrics,

00:46:02.135 --> 00:46:05.539
<v Matt>and you'd run the app, and you'd load test it, and you'd see what broke,

00:46:05.539 --> 00:46:10.181
<v Matt>and then when something broke or didn't break, you'd just adjust the levels.

00:46:10.651 --> 00:46:16.079
<v Matt>You're like, oh, this one needs more heap. This one needs more minimum memory.

00:46:16.079 --> 00:46:19.759
<v Matt>This one needs less maximum memory. It needs more CPU.

00:46:23.991 --> 00:46:27.456
<v Matt>And then you'd add that to the startup script for the app next day.

00:46:28.797 --> 00:46:34.309
<v Matt>And do it all over again you go to a place and you'd spend a week doing it and

00:46:34.309 --> 00:46:37.545
<v Matt>then they give you a new place to go to the next week to do it all over again

00:46:38.300 --> 00:46:40.477
<v Nina>Wow wow,

00:46:42.642 --> 00:46:45.835
<v Nina>that's that's what you did full-time before we worked together.

00:46:46.613 --> 00:46:51.839
<v Matt>No this is like my first job out of college it was a consultant ever had uh

00:46:51.839 --> 00:46:56.729
<v Matt>you just like it wasn't every project some projects was like actually app development

00:46:56.729 --> 00:47:00.973
<v Matt>But when you're between projects, then you need to keep you busy until they found a bigger one.

00:47:01.867 --> 00:47:03.562
<v Matt>You do small engagements like that.

00:47:06.005 --> 00:47:09.352
<v Nina>Yeah. And now that wouldn't even be necessary.

00:47:10.072 --> 00:47:15.932
<v Matt>Now you can do like 100 of them in a day. If someone didn't already just do it themselves.

00:47:18.375 --> 00:47:25.324
<v Nina>How do you feel about that, though? As somebody who had to put in the time to do that.

00:47:26.735 --> 00:47:30.142
<v Matt>Oh, yeah. Like the slow way. No.

00:47:30.485 --> 00:47:33.072
<v Matt>I've never been one of those gray beards that are just like,

00:47:33.072 --> 00:47:38.132
<v Matt>Back in my day, you used to have to manage the JVM by hand.

00:47:38.652 --> 00:47:45.415
<v Nina>Garbage collection. And yeah, it's, yeah, that's, I feel like I'm going to be one of those though.

00:47:46.866 --> 00:47:51.061
<v Nina>I feel like, I feel like I am set up to be one of those because,

00:47:51.608 --> 00:47:54.132
<v Nina>yeah, I'm kind of envious.

00:47:54.532 --> 00:47:56.675
<v Nina>I'm kind of envious of AI sometimes.

00:47:59.131 --> 00:48:03.322
<v Nina>I kind of get mad at it. like oh i put in all this time and this son of a bitch

00:48:03.322 --> 00:48:12.458
<v Nina>can do it in x amount of minutes seconds you know but um no it's,

00:48:14.079 --> 00:48:14.746
<v Nina>new frontier.

00:48:15.407 --> 00:48:18.462
<v Matt>Yeah it just gotta be product focused and that's about the code and more about

00:48:18.462 --> 00:48:20.610
<v Matt>the product yeah just a lot better

00:48:21.248 --> 00:48:22.553
<v Nina>Yeah for sure.

00:48:23.570 --> 00:48:24.927
<v Matt>Build software you don't write code

00:48:28.220 --> 00:48:29.655
<v Nina>I used to love writing code.

00:48:31.216 --> 00:48:35.842
<v Matt>Yeah, I mean, that's the people who have qualms about it are the ones who like the code.

00:48:37.427 --> 00:48:41.731
<v Nina>Yeah. Yeah, I saw, I know we're getting towards the end. Maybe I could table

00:48:41.731 --> 00:48:48.281
<v Nina>this for next time, but I saw a video, a vlog of a guy saying that he was going

00:48:48.281 --> 00:48:51.341
<v Nina>to refuse to use AI for coding it.

00:48:51.341 --> 00:48:52.727
<v Matt>No, that's a bad idea.

00:48:53.430 --> 00:48:55.427
<v Nina>Even if it gets him fired.

00:48:55.879 --> 00:49:00.641
<v Matt>Well, yeah, I mean, that is going to, It's like, it's, well,

00:49:00.641 --> 00:49:04.361
<v Matt>some higher ops who doesn't code that's in charge of hiring,

00:49:04.361 --> 00:49:07.814
<v Matt>firing is going to see they're not using the tokens and they get upset.

00:49:09.132 --> 00:49:14.021
<v Nina>Yes, and he brought that up. He brought up that, like, his project manager was

00:49:14.021 --> 00:49:19.076
<v Nina>upset that he wasn't consuming enough tokens and he thought he wasn't working or something.

00:49:21.101 --> 00:49:24.840
<v Matt>That's exactly what happened when people switched to Agile. like it was

00:49:25.375 --> 00:49:30.506
<v Matt>and people didn't like points and what points don't what points don't mean time

00:49:30.976 --> 00:49:35.051
<v Matt>but then they totally mean time and then people like higher up that didn't know

00:49:35.051 --> 00:49:39.721
<v Matt>what's going on why aren't you getting more points then right and

00:49:40.462 --> 00:49:45.071
<v Nina>So that something recent like when i when i did have a job that happened to

00:49:45.071 --> 00:49:49.899
<v Nina>me i wasn't using enough tokens at a previous place of employment.

00:49:50.265 --> 00:49:50.938
<v Matt>Jesus.

00:49:52.274 --> 00:49:54.694
<v Nina>I don't know if I ever told you that. Did I ever tell you that?

00:49:55.147 --> 00:49:57.649
<v Matt>No. It doesn't make sense if they would track that.

00:49:58.392 --> 00:50:05.151
<v Nina>Right, right. So I was asked, are you using the product? Are you using Claude?

00:50:05.151 --> 00:50:07.854
<v Nina>Are you using? And I was like, yes, I am.

00:50:08.179 --> 00:50:11.593
<v Nina>It's just that the work that I've been doing, I feel like,

00:50:13.055 --> 00:50:18.162
<v Nina>don't need to use it as much. They were just like a bunch of bug changes.

00:50:19.470 --> 00:50:23.208
<v Nina>I know where it is. I'm not going to send an agent off to do it if I know where it is.

00:50:23.742 --> 00:50:27.661
<v Matt>Yeah, it's the metric du jour, so whatever it becomes, you always got to do it.

00:50:28.537 --> 00:50:36.124
<v Nina>Right right but no he he did the the guy in the vlog brought up like skill regression and um,

00:50:37.123 --> 00:50:44.262
<v Nina>you know he he felt like he was forgetting how to actually engineer and then

00:50:44.262 --> 00:50:48.522
<v Nina>therefore if claude's building something he's losing the ability to,

00:50:49.102 --> 00:50:55.158
<v Nina>code review from not from that skill regression of not writing code and that does,

00:50:57.015 --> 00:51:03.872
<v Nina>that does make sense but now Claude reviews code so like you know what's what's

00:51:03.872 --> 00:51:08.341
<v Nina>the what's the train off there like what's the what's the point bun like.

00:51:09.982 --> 00:51:14.312
<v Matt>I mean i don't agree with this guy at all i think it's a very bad idea both

00:51:14.312 --> 00:51:19.692
<v Matt>on the superficial side of it and the practical side of things like if everyone

00:51:19.692 --> 00:51:24.450
<v Matt>starts doing Claude Code like he's not going to be able to keep up reviewing manually

00:51:25.588 --> 00:51:29.512
<v Nina>Right i think he's treating it kind of like join me in this revolution this

00:51:29.512 --> 00:51:34.127
<v Nina>this revolution like and no one's gonna join him that's that's the thing no one is,

00:51:34.910 --> 00:51:39.262
<v Nina>the the way he was phrasing things first they came from my code editor and so

00:51:39.262 --> 00:51:45.085
<v Nina>he was describing it like and it was auto-completing for me then it came for it like,

00:51:46.211 --> 00:51:48.702
<v Nina>it was really it was kind of funny it was kind of.

00:51:48.702 --> 00:51:55.999
<v Matt>I bet this dude writes haskell as well just like does all the high-minded programming things you can do

00:51:56.590 --> 00:52:01.165
<v Nina>No he was actually he's actually just a regular javascript developer surprisingly

00:52:01.537 --> 00:52:08.403
<v Nina>no it even makes it even funnier to me that that makes it really funny to me um.

00:52:08.950 --> 00:52:12.647
<v Matt>Well he better at least be an anti-framework guy that be a vanilla js guy

00:52:14.968 --> 00:52:17.550
<v Nina>Right. If he likes to do it the hard way, right?

00:52:18.170 --> 00:52:18.376
<v Matt>Yeah.

00:52:22.138 --> 00:52:26.416
<v Nina>But, yeah. Yeah, maybe I could pull up this video, like, another time.

00:52:28.111 --> 00:52:29.916
<v Nina>We could talk about it and why it's a bad idea.

00:52:32.441 --> 00:52:34.190
<v Nina>Debate it.

00:52:34.190 --> 00:52:37.828
<v Matt>He better be in a very small company where he has a lot of way.

00:52:38.676 --> 00:52:44.840
<v Matt>Because, I mean, there's a game to this. it's not just do what's right all the

00:52:44.840 --> 00:52:48.983
<v Matt>time you gotta do what people want done sometimes

00:52:51.990 --> 00:52:57.870
<v Nina>Yeah it's this is this is not the place to make make a statement about this

00:52:57.870 --> 00:53:03.050
<v Nina>like we're not going to be able to stop how things are changing you have to

00:53:03.050 --> 00:53:06.294
<v Nina>you have to join or you have to like you know,

00:53:08.709 --> 00:53:12.105
<v Nina>fall fall away but um yeah.

00:53:14.612 --> 00:53:16.150
<v Matt>I think I'm good I

00:53:16.750 --> 00:53:17.770
<v Nina>Guess that's.

00:53:17.770 --> 00:53:19.338
<v Matt>The vibe that's

00:53:19.645 --> 00:53:20.227
<v Nina>The vibe.

