Oil Analyst Saves HOURS Analyzing News (Live Build)

WEBVTT
00:00:00.000 --> 00:00:08.080
Uh it would take about one to two hours throughout the day just to screen and look through different uh news articles like CNN, uh CNBC.
00:00:08.480 --> 00:00:11.039
Let's run this dangerously, you know, cowboy mode.
00:00:11.359 --> 00:00:11.599
Oh.
00:00:11.919 --> 00:00:16.559
And since we're talking about oil trading, I guess uh cowboys will be listening.
00:00:16.879 --> 00:00:20.160
One article is saying no vessels are passing through the street at all.
00:00:20.320 --> 00:00:24.079
And then the second article is saying, oh actually an Indian vessel passed through.
00:00:24.399 --> 00:00:26.079
I think we are setting the new standard, right?
00:00:26.160 --> 00:00:31.199
Like Brian's gonna be the forerunner of what it means to be uh a market analyst in 2026.
00:00:31.359 --> 00:00:38.560
So you make sure everyone is as good as you laugh crazy that they actually went through so many articles in just a few seconds.
00:00:39.359 --> 00:00:40.000
Yeah.
00:00:40.320 --> 00:00:47.200
And for me, it's like one hour max I have to read like what five, six, seven articles.
00:00:53.520 --> 00:00:56.880
Yeah, hi everyone, uh welcome to the Code Rift Podcast.
00:00:56.960 --> 00:00:59.359
I'm Eric, AI Vibecoder.
00:00:59.439 --> 00:01:08.480
Along with me, we have Yao Hong, who is an AI orchestrator, and we have Brian, who is uh oi market analyst.
00:01:08.640 --> 00:01:09.200
Is that right?
00:01:09.519 --> 00:01:10.560
Yeah, correct.
00:01:10.879 --> 00:01:11.840
Alright, cool.
00:01:12.000 --> 00:01:22.079
And today we're gonna be figuring out how to automate some part of the research process when it comes to reading the news.
00:01:22.560 --> 00:01:29.120
Brian spends about one or two hours a day reading news and presenting that to traders, I think.
00:01:29.359 --> 00:01:37.439
So uh Brian, do you mind just sharing a bit more about what you do in a day and what we are hoping to solve today?
00:01:37.840 --> 00:01:51.040
Okay, yeah, so for context, this part of my work is more um consolidating data or news uh throughout the day, you know, and presenting it in a concise manner to the traders.
00:01:51.200 --> 00:02:05.280
So as we had mentioned, um I will take about one to two hours throughout the day just to screen and look through different uh news articles like CNN, uh CNBC routers, or just across many news sites.
00:02:05.760 --> 00:02:15.120
So I think the nuance here is that some of these news articles are uh left-leaning or right-leaning, or they have very flashy headlines as well.
00:02:15.360 --> 00:02:26.479
And it's a bit difficult to uh look through the articles and sift out the facts in a nice concise manner to the traders for them to make better trading decisions.
00:02:26.719 --> 00:02:29.360
So that's why I reached out for um some help.
00:02:29.439 --> 00:02:29.759
Yeah.
00:02:30.240 --> 00:02:34.000
In a way it's similar to the ESG episode.
00:02:34.560 --> 00:02:35.680
Um I think yes.
00:02:36.240 --> 00:02:36.560
Yeah.
00:02:37.120 --> 00:02:37.520
Yeah.
00:02:37.759 --> 00:02:49.439
So if you haven't watched that, for those who are who are listening, uh our first episode was on scraping news headlines uh for an ESG analyst in a bank.
00:02:49.680 --> 00:02:58.479
And today's episode is a bit different in a sense that we we do want to get into the details of the news.
00:02:58.960 --> 00:03:03.120
And there is sort of uh a sifting process behind that.
00:03:03.599 --> 00:03:14.879
Where uh do you do you mind explaining a bit more, Brian, about the left slash right leaning and maybe other factors that you are looking out for when you read a piece of news?
00:03:15.360 --> 00:03:38.400
Okay, so um perhaps the on the point on left or right leaning, so some news sites or news articles like for example, Fox or um maybe CNN or CNBC, some of them are uh you know catered towards like the the political right in the US or the political left.
00:03:38.639 --> 00:04:03.599
So some of them um they have headlines that are quite biased, or uh news uh or the content might be biased, and I think the other factors will include, for example, uh very flashy headlines, but um the article might say otherwise, or there's some facts or very hard numbers in the articles that I would uh very much prefer to sift out over the headlines.
00:04:04.400 --> 00:04:04.960
Yeah.
00:04:05.280 --> 00:04:05.919
Yeah.
00:04:06.319 --> 00:04:13.840
So it's kind of stripping out the facts from the headlines, you know, because sometimes headlines are for the clicks and the views.
00:04:14.080 --> 00:04:14.400
Correct.
00:04:14.719 --> 00:04:18.720
You really just want the cool hard facts that leads to something.
00:04:19.040 --> 00:04:27.199
Do you mind explaining what the facts usually leads to and how your your thought process usually goes from the facts?
00:04:27.839 --> 00:04:50.959
So after pulling out the facts throughout the day or throughout the past few days or the week, then um it might be good or it might be nice for AI or to have this sort of this thing that we are building, uh come up with a structured uh approach on what can happen to prices.
00:04:51.360 --> 00:04:58.959
So maybe an example would be, or very current example would be the Iran war or uh closure of the Homus Strait.
00:04:59.040 --> 00:05:03.839
So that is like a global oil choke point, which is 20% of global oil flows.
00:05:04.079 --> 00:05:16.480
So maybe a very brief example would be okay, headline uh war closure of Homus Strait, then the impact would be okay, then the Strait of Homus is closed.
00:05:16.639 --> 00:05:18.079
So what would that lead to?
00:05:18.319 --> 00:05:24.959
The impact would be supply from the Arab Gulf stops, and there's not enough supply to meet existing demand in Asia.
00:05:25.279 --> 00:05:25.600
Yeah.
00:05:26.000 --> 00:05:31.519
So the short term would be, so maybe there would be a short-term and long-term impact.
00:05:31.680 --> 00:05:48.399
So the short-term impact would be depletion of inventory, and then they would buy more from refineries in the West, and then um in the long term, for example, it would uh incentivize refineries to increase production if the war persists.
00:05:48.720 --> 00:05:58.399
Yeah, so how that would maybe impact prices in the near term would be um prices because oil trades on a curve, right?
00:05:58.480 --> 00:05:59.680
They call it futures.
00:05:59.920 --> 00:06:05.120
So the prices now would be a lot more expensive than in the future, and that's called backwardation.
00:06:05.360 --> 00:06:08.480
Yeah, so I think that can be traded as well, yeah.
00:06:08.720 --> 00:06:11.920
Or they can also trade like location spreads like east-west.
00:06:12.240 --> 00:06:16.399
So, for example, the prices in the east would be a lot more expensive than the west.
00:06:16.480 --> 00:06:18.079
So they trade that differential as well.
00:06:18.240 --> 00:06:18.480
All right.
00:06:18.720 --> 00:06:29.279
So, in the process, do you have to analyze to that degree to provide to the traders, or do you just take out the facts and maybe some related facts?
00:06:29.519 --> 00:06:37.279
Right now it's just some related facts, but you know, I'm just hoping that eventually he can give you okay, okay, okay.
00:06:37.519 --> 00:06:37.920
Yeah, that's true.
00:06:38.560 --> 00:06:45.360
But I think for now, maybe if it we are not so ambitious or something that is realistic, then the hard facts would be very good already.
00:06:45.600 --> 00:06:45.759
Yeah.
00:06:46.240 --> 00:06:47.759
Well no, we should try, we should try.
00:06:47.839 --> 00:06:51.439
Uh provided that we have a certain structure, I think it's possible.
00:06:51.680 --> 00:06:52.160
Yeah, yeah.
00:06:52.399 --> 00:06:53.040
Yeah.
00:06:53.439 --> 00:06:58.319
Alright, so we can do it a few ways.
00:06:58.560 --> 00:07:00.720
I see it as three main things.
00:07:00.879 --> 00:07:03.199
The first part is really just getting the news.
00:07:03.360 --> 00:07:14.000
Step two is sifting the information out um according to uh the facts, and of course, maybe putting in some of the the biases.
00:07:14.399 --> 00:07:15.519
Yeah, remove the bias.
00:07:16.000 --> 00:07:20.720
The last stage is where you you output it or you do the scenario analysis.
00:07:20.800 --> 00:07:25.680
It's like three things you mentioned the impact, the response, and the price of the event.
00:07:26.079 --> 00:07:35.680
And would the impact, response, and price be tied to one piece of fact, or would it be tied to one article, usually?
00:07:36.079 --> 00:07:41.279
Um ideally it would be across multiple pieces of facts across multiple articles.
00:07:41.439 --> 00:07:41.839
Yeah.
00:07:42.240 --> 00:07:46.079
Because not all articles hold the same facts.
00:07:46.319 --> 00:07:48.240
There might be multiple facts in one article, right?
00:07:48.560 --> 00:07:49.120
Yeah, correct.
00:07:49.360 --> 00:07:50.560
Yeah, okay.
00:07:51.360 --> 00:07:51.680
Okay.
00:07:52.319 --> 00:07:56.959
Okay, so I'm just going to start sharing my screen and we are going to code today.
00:07:57.600 --> 00:07:58.000
Nice.
00:07:58.399 --> 00:08:03.680
Um because I still think it's important for people to try to code their own stuff.
00:08:03.839 --> 00:08:09.920
Ya Hong, are you okay if I if I go dangerously skip permissions or go ahead, go ahead.
00:08:10.000 --> 00:08:10.560
Yeah, go ahead.
00:08:10.720 --> 00:08:12.240
You don't need to keep pressing button.
00:08:12.720 --> 00:08:12.879
Okay.
00:08:13.360 --> 00:08:15.680
Anyway, it's not my laptop, so it's okay.
00:08:16.480 --> 00:08:21.199
Okay, I am going to do this thing called um dangerously skip permissions.
00:08:21.439 --> 00:08:24.639
Uh, for for Brian's uh benefit, right?
00:08:24.800 --> 00:08:40.480
So essentially when you use plot code or codecs, this these um coding agents usually we run them in a sandbox because they they have a lot of um permissions to do a lot of things, right?
00:08:40.799 --> 00:08:46.399
But you can actually pass in a command to say just run this dangerously, you know, cowboy mode.
00:08:46.720 --> 00:08:52.000
Oh and since we're talking about oil trading, I guess uh also cowboys to be listening.
00:08:54.240 --> 00:08:58.320
Yeah, so uh you could type clot and then you type this thing, of course.
00:08:58.480 --> 00:08:59.440
Um dangerously.
00:08:59.919 --> 00:09:08.000
Once again, uh we are using the more advanced version of clot code, um, which you can access starting from 30 bucks, uh Sync dollars.
00:09:08.240 --> 00:09:14.639
But yeah, so basically I have a command where when I press X it helps me skip permissions.
00:09:14.799 --> 00:09:16.320
Uh they call it an alias.
00:09:16.559 --> 00:09:20.720
So I am actually recording our conversation.
00:09:21.600 --> 00:09:28.559
Um and I'm going to I'm going to copy what we just talked about.
00:09:28.879 --> 00:09:29.519
Yeah.
00:09:29.840 --> 00:09:37.919
So for brilliant sake, uh because Eric is trying to be lazy, doesn't want to type out what we talked about.
00:09:38.159 --> 00:09:38.480
Oh, okay.
00:09:38.559 --> 00:09:40.159
So you just like copy and paste the whole.
00:09:43.360 --> 00:09:47.440
Actually, one thing I needed to do was to make it in plan mode.
00:09:47.600 --> 00:09:49.840
So I'm gonna just stop this whole thing.
00:09:50.080 --> 00:09:51.840
Um and I'm gonna copy this.
00:09:51.919 --> 00:09:55.120
This is a transcript of a user interview with an oil market analyst.
00:09:55.360 --> 00:09:57.919
Please plan out the approach and let's build.
00:09:58.240 --> 00:10:06.639
And then I will have the transcript below and we'll let it uh run.
00:10:07.519 --> 00:10:13.360
So now it's just going to kind of spend a few minutes uh planning out how it's going to build it.
00:10:13.600 --> 00:10:21.200
Um but I think we we gave it enough uh context to to think about yeah how it's going to approach it.
00:10:21.919 --> 00:10:31.519
Brian, uh would you be able to share any articles that you have just in case you know some of these articles are behind a paywall?
00:10:32.159 --> 00:10:39.120
Anyways, the plan right now is to build.
00:10:39.360 --> 00:10:40.080
Oh you know what?
00:10:40.159 --> 00:10:43.039
I think it's actually reading some stuff from the episode one.
00:10:43.679 --> 00:10:45.039
Oh, that's pretty smart actually.
00:10:45.440 --> 00:10:46.240
So why not?
00:10:46.799 --> 00:10:47.279
Sorry about that.
00:10:47.519 --> 00:10:51.039
Uh this is building on episode one, so if you haven't watched that, please do.
00:10:51.279 --> 00:10:53.279
Then you uh get more context.
00:10:53.519 --> 00:10:55.039
So that's that's very smart.
00:10:55.279 --> 00:10:57.039
It's on fetching, classifying, generating.
00:10:57.120 --> 00:10:59.679
And it's when you do dangerously, man.
00:11:00.480 --> 00:11:02.799
Yeah, we can read a lot of things on my computer.
00:11:02.960 --> 00:11:03.120
Yeah.
00:11:03.679 --> 00:11:06.480
Um so it's always good to sandbox, right?
00:11:06.559 --> 00:11:07.759
Which I'm actually not.
00:11:08.240 --> 00:11:12.799
But anyways, um for you guys, I will risk my my computer.
00:11:13.039 --> 00:11:19.759
Um, what we're building, step one is to fetch oil news uh for those that are just listening.
00:11:19.919 --> 00:11:24.399
Uh stage one is to fetch oil news from Reuters CNN.
00:11:24.960 --> 00:11:36.080
Stage two is to extract the facts, stage three is scenario analysis, and then the output format would be a morning briefing and an evening update.
00:11:36.399 --> 00:11:38.879
Oh, which it did understand as well.
00:11:39.600 --> 00:11:48.480
Um and then in terms of the format, actually it's it's mentioning that it wants a word doc.
00:11:49.039 --> 00:11:51.360
Do you want a word dog or do you want Excel?
00:11:55.279 --> 00:11:58.639
So there is actually a particular format that we want, right?
00:11:58.720 --> 00:11:59.759
Uh Brian.
00:12:00.000 --> 00:12:01.600
So uh yes, correct.
00:12:01.919 --> 00:12:05.600
But um but I think it's fine, we can do that later.
00:12:05.840 --> 00:12:08.080
Uh let's just proceed.
00:12:08.240 --> 00:12:16.080
I'm just going to I'm just gonna clear context and and bypass permissions so that we start building this.
00:12:16.480 --> 00:12:19.039
Okay, everything was built.
00:12:19.279 --> 00:12:29.840
Uh we have so they built a Python project as uh configuration with fetches, okay, quite similar to what we did the first round with fetch and extract.
00:12:30.399 --> 00:12:34.399
But now we have an additional analyze uh path.
00:12:34.480 --> 00:12:34.879
Yeah.
00:12:35.200 --> 00:12:41.440
Yep, so we do need to install requirements and then create an n file.
00:12:41.679 --> 00:12:48.559
I think you need to do that in the file that actually, not in the this open uh open.
00:12:49.200 --> 00:12:50.960
So I just maybe go back here.
00:12:51.519 --> 00:12:52.559
Cd dash, yeah.
00:12:52.639 --> 00:12:52.799
Okay.
00:12:53.200 --> 00:12:53.759
Enter.
00:12:54.639 --> 00:12:56.799
Go back to your last known folder.
00:12:57.120 --> 00:12:59.919
Okay, and then install requirements.
00:13:00.720 --> 00:13:07.360
Okay, so we're uh for Brian's sake, we're just installing a bunch of stuff that you don't know what installing onto your computer to run your project.
00:13:08.639 --> 00:13:10.639
Well, there's so much stuff to download.
00:13:11.279 --> 00:13:18.799
Yeah, because these are all libraries, you know, open source libraries that people have created and we are building on top of on the shoulders of giants, right?
00:13:19.039 --> 00:13:19.840
Oh, that's crazy.
00:13:20.240 --> 00:13:20.480
Okay.
00:13:21.120 --> 00:13:24.960
Yeah, and and it's uh yeah, how how people build stuff these days now?
00:13:25.120 --> 00:13:30.320
We've been installing things on uh from the internet for the last I don't know how many years.
00:13:30.559 --> 00:13:30.879
Yeah.
00:13:31.200 --> 00:13:40.399
So there were somewhere in the world there is this one small library that everybody references, which uh if it goes down, then half the internet will go down.
00:13:40.639 --> 00:13:42.399
Yeah, and it has happened before.
00:13:42.799 --> 00:13:44.960
But then no one will add to the library anymore, right?
00:13:45.039 --> 00:13:48.159
Since we've like uh not really.
00:13:48.320 --> 00:13:51.440
I mean people maintain the existing libraries and all that, right?
00:13:51.519 --> 00:13:52.879
Because it's always been there.
00:13:53.120 --> 00:13:53.679
I see.
00:13:53.840 --> 00:13:57.679
And yeah, everything is open source, so it's it's pretty cool how how things work.
00:13:58.080 --> 00:13:58.960
Interesting.
00:13:59.279 --> 00:14:03.360
Okay, so I'm just gonna paste my key into the.env file.
00:14:04.080 --> 00:14:07.120
Um but let me not expose your key.
00:14:07.200 --> 00:14:12.559
Okay, so for Brian's sake, uh these API keys are quite sensitive.
00:14:13.279 --> 00:14:25.519
And there are sniffers out there on the internet that looks for people who expose their keys and then they get free access to your Entropic or your Gemini account and they can do crazy things.
00:14:25.840 --> 00:14:26.639
Oh shit.
00:14:27.120 --> 00:14:28.080
Okay, yeah.
00:14:28.320 --> 00:14:40.879
So part of the security reason that we have to be a bit more careful because if some people they save the keys into their repository, which is and then they put it on public uh access.
00:14:41.759 --> 00:14:46.639
Yeah, so quite I mean it's not super dangerous, it's just be more careful.
00:14:46.720 --> 00:14:48.879
Yeah, like a password, right?
00:14:48.960 --> 00:14:50.960
Like a key to correct, correct, correct.
00:14:51.039 --> 00:14:52.240
Yeah, yeah.
00:14:52.480 --> 00:14:58.559
So I just asked uh Clock Code to open the.env file for me to paste it in.
00:14:58.720 --> 00:15:04.080
Uh paste the API key in, and then now I will say, okay, I have put it in.
00:15:04.399 --> 00:15:13.679
Um this is also an example GOAT standard output, which is based off this article that I will paste uh below.
00:15:14.320 --> 00:15:40.960
This is so that you would return the facts in this um or you would process the facts and return it in this format, which is what we call a scenario analysis, so that every fact uh that is retrieved from an article would be able to be um be returned in this format.
00:15:41.200 --> 00:15:43.440
Okay, hopefully that's clear enough.
00:15:43.840 --> 00:16:00.080
I'm just gonna paste the article um below here, and then I am going to paste uh the picture of the ideal output that Brian sent me, and then I will paste what I just said.
00:16:00.960 --> 00:16:04.639
Um, let's go.
00:16:04.799 --> 00:16:06.559
Okay, so it's probably going.
00:16:06.799 --> 00:16:08.559
I think it was already kind of done.
00:16:08.639 --> 00:16:10.639
Uh we should have checked how it looked.
00:16:10.720 --> 00:16:14.720
Um but now we're just updating the scenario prompt.
00:16:14.960 --> 00:16:17.039
Uh just assume that it's quite good.
00:16:17.360 --> 00:16:22.639
Yeah, I think based on the previous episode we saw it was decently feasible.
00:16:23.679 --> 00:16:26.399
Of course, uh in reality, we would spend more time on this.
00:16:27.200 --> 00:16:27.440
Yeah.
00:16:28.000 --> 00:16:34.720
We would spend more time to make sure that uh whatever you put in is is correct and what you intend it to be.
00:16:34.799 --> 00:16:36.879
Yeah, just from my understanding, right?
00:16:36.960 --> 00:16:43.360
This this tokens thing that I see are sort of like uh like you know, the more tokens you use, the more you have to pay.
00:16:45.039 --> 00:16:46.080
Yeah, yeah.
00:16:46.240 --> 00:16:58.320
If you're using the API for now, because um Eric's running it on the max plan, so uh he he's paid a fixed fee for the month and they give you a set limit.
00:16:58.639 --> 00:16:59.200
Oh, okay.
00:16:59.440 --> 00:17:00.159
Yeah, yeah.
00:17:00.240 --> 00:17:03.519
Unfortunately, mine is almost 99% already.
00:17:04.799 --> 00:17:08.880
I I I hit my limit like two days into the second day of the week already.
00:17:10.640 --> 00:17:12.319
I've been struggling for the last couple of days.
00:17:12.960 --> 00:17:14.880
Definitely uh tomorrow morning it resets.
00:17:16.559 --> 00:17:19.039
So it's uh bit of a challenge.
00:17:19.440 --> 00:17:23.039
I need to definitely upgrade my plan to the$200 one.
00:17:23.359 --> 00:17:24.480
Time to go for it.
00:17:24.720 --> 00:17:28.079
Okay, I'm just going to test this thing.
00:17:28.400 --> 00:17:31.119
Um with the the Python.
00:17:31.680 --> 00:17:33.039
Okay, let's see what's happening here.
00:17:33.119 --> 00:17:33.680
That's very cool.
00:17:33.759 --> 00:17:37.039
Like Oil News Intelligence 2, maximum eight articles.
00:17:37.359 --> 00:17:38.640
It's fetching the news now.
00:17:38.720 --> 00:17:43.359
Okay, see all the uh different news feeds that it's fetching from.
00:17:43.920 --> 00:17:46.720
So interesting that it's in the terminal, as in no.
00:17:50.240 --> 00:17:50.720
Yeah.
00:17:50.960 --> 00:17:51.519
Yeah.
00:17:51.920 --> 00:18:03.119
And actually, I guess it's ironic that the UI UX now is going back to the terminal after decades of our graphical user interfaces.
00:18:04.400 --> 00:18:08.400
I mean, for someone like me, it's like, oh man, so back to the roots.
00:18:08.799 --> 00:18:12.160
When we started playing computers, it was all black screens like this.
00:18:12.480 --> 00:18:12.799
Yeah.
00:18:12.960 --> 00:18:15.680
And also because it's low latency, it's very fast.
00:18:15.839 --> 00:18:18.000
You can see whether you're getting what you need.
00:18:18.720 --> 00:18:28.880
It has so for those who are listening in, um, it already processed step one, which was fetching the news, and it fetched about 748 articles.
00:18:29.200 --> 00:18:32.960
Um, and then after duplication, it's about 600 articles.
00:18:33.279 --> 00:18:35.119
Wow, I don't know if it's bluffing us.
00:18:35.279 --> 00:18:43.200
It sounds a bit crazy, but uh I think maybe we forgot to put in a filter for the date to only process the last 24 hours.
00:18:43.519 --> 00:18:43.759
Yeah.
00:18:44.799 --> 00:18:47.920
So as I'm as we're reading this, I think we can start giving feedback, right?
00:18:48.079 --> 00:18:50.319
Oh, it's about 48 hours and it's not for six.
00:18:50.720 --> 00:18:51.279
48 hours.
00:18:51.359 --> 00:18:52.400
Feel free to give feedback.
00:18:52.559 --> 00:18:55.039
We we'll feed this back into the system.
00:18:55.599 --> 00:18:55.759
Okay.
00:18:56.000 --> 00:18:56.160
Yeah.
00:18:56.400 --> 00:18:58.640
So so what's your feedback so far looking at this?
00:18:58.880 --> 00:18:59.039
Yeah.
00:18:59.200 --> 00:19:05.359
Oh, I think it's it's it's quite crazy that they actually went through so many articles in just a few seconds.
00:19:06.160 --> 00:19:06.720
Yeah.
00:19:07.119 --> 00:19:08.000
And yeah.
00:19:08.799 --> 00:19:14.960
For me, it's like one hour max, I have to read like what five, six, seven articles.
00:19:16.160 --> 00:19:17.759
Your job is at risk.
00:19:18.880 --> 00:19:20.720
No, I think we are setting the new standard, right?
00:19:20.799 --> 00:19:26.000
Like Brian's gonna be the forerunner of what it means to be uh a market analyst in 2026.
00:19:26.240 --> 00:19:28.559
So you make sure everyone is is as good as you, lah.
00:19:29.039 --> 00:19:35.920
Hopefully you probably will be able to figure out something similar or even better because you know your process the best.
00:19:36.160 --> 00:19:39.200
You have your own data in your stored in your computer, right?
00:19:39.279 --> 00:19:42.000
So that's also yeah.
00:19:42.240 --> 00:19:47.200
So okay, maybe for those who are listening in, step one, it fetched the news.
00:19:47.440 --> 00:19:51.920
Step two, it extracted the facts from about 600 plus articles.
00:19:52.160 --> 00:20:01.279
Step three, it ran the scenario analysis, um and grouped by and it grouped the facts by different events.
00:20:01.359 --> 00:20:02.160
Uh whatever that means.
00:20:02.400 --> 00:20:03.839
Uh event groups, that's pretty interesting.
00:20:03.920 --> 00:20:07.279
So event group one, Malaysia's Prime Minister's business dealings.
00:20:08.000 --> 00:20:20.319
Although I I don't know how relevant that is to your work, but uh maybe we we were too open-ended on the filtering part.
00:20:20.400 --> 00:20:20.720
Yeah.
00:20:21.200 --> 00:20:39.359
Yeah, maybe it's a little bit broad, but perhaps we could like um narrow it to macroeconomics or geopolitics, and also just oil uh physical fundamentals like storages or oil flows, anything physical related to oil.
00:20:39.680 --> 00:20:40.319
Yeah.
00:20:40.880 --> 00:20:47.440
Um another thing was were you okay with the 48 hours, or do you feel like it's too much?
00:20:48.160 --> 00:20:49.920
Well, maybe one day would be fine as well.
00:20:50.160 --> 00:20:52.240
I think 24 will be 24 will be better.
00:20:52.799 --> 00:20:54.000
Yeah, let's do 24 hours.
00:20:55.759 --> 00:20:57.359
Any any other feedback?
00:20:57.680 --> 00:20:59.119
Uh no, no more.
00:21:00.079 --> 00:21:00.400
Okay.
00:21:01.440 --> 00:21:03.119
I'm gonna paste this in.
00:21:03.599 --> 00:21:06.079
Here's some feedback for you.
00:21:07.680 --> 00:21:10.480
Um friendly.
00:21:11.039 --> 00:21:11.839
Thank you.
00:21:12.079 --> 00:21:12.400
Okay.
00:21:12.880 --> 00:21:15.200
So now it has done this file.
00:21:15.440 --> 00:21:18.559
Uh which I will read this file and see what happens now.
00:21:18.880 --> 00:21:20.960
I will find it in my desktop.
00:21:21.359 --> 00:21:26.720
Okay, so we have uh okay, very weirdly written, but we have uh oil market intelligence briefing.
00:21:27.039 --> 00:21:29.279
Six articles analyzed, six scenarios.
00:21:30.160 --> 00:21:33.759
Brian, is there any event that you you think this event is okay to read out?
00:21:33.839 --> 00:21:38.480
Like, or maybe the uh the Indian one or Indian one.
00:21:38.799 --> 00:21:39.119
Okay.
00:21:39.680 --> 00:21:44.559
So for India, I'm just gonna read out uh India's gas diversion measures.
00:21:44.720 --> 00:21:55.119
So it has a TLDR, which is the emergency gas diversion to priority sectors, is tightening domestic supply, likely boosting sport LNG demand and prices globally.
00:21:55.359 --> 00:21:59.519
So there's an impact on the supply-demand, trade flow policy, infrastructure.
00:22:00.079 --> 00:22:01.440
It viewed it as a short term response.
00:22:01.519 --> 00:22:04.640
I think I think it classified it as a short term uh response.
00:22:05.680 --> 00:22:22.720
And then the impact on prices uh on the global L and G, sport prices on Indian domestic gas prices, on diesel prices, and there's a confidence level which is medium as well.
00:22:23.119 --> 00:22:25.119
Oh, that's quite a nice uh addition.
00:22:25.279 --> 00:22:25.519
Yeah.
00:22:26.000 --> 00:22:28.640
So what's your your thought on this?
00:22:29.440 --> 00:22:45.440
Um I think it's it's pretty good, like the way that it actually knew how to separate or follow the format that I gave you, which is split into impact, response, and uh how the price reacted to the impact, which was also based on the facts from the article.
00:22:46.000 --> 00:22:50.000
Um yeah, so also it's very good in this part, but then I think it's a bit too broad.
00:22:50.079 --> 00:22:58.880
It's like anything energy, but I wanted more like flows or um yeah, anything related to physical uh oil.
00:22:59.359 --> 00:22:59.599
Yeah.
00:23:00.319 --> 00:23:00.640
Okay.
00:23:02.079 --> 00:23:02.319
Cool.
00:23:02.400 --> 00:23:03.599
I'm gonna copy that in.
00:23:04.480 --> 00:23:17.440
Um so I'm gonna copy in your your feedback as a prompt, um, which is how we want more physical uh we want more physical oil.
00:23:17.599 --> 00:23:18.400
How do you say that again?
00:23:19.200 --> 00:23:21.440
Um flows like f L O W S.
00:23:21.920 --> 00:23:22.880
All flows.
00:23:23.599 --> 00:23:24.880
More feedback for you.
00:23:25.359 --> 00:23:30.720
Sorry, one more one more thing is macroeconomics and uh geopolitics.
00:23:31.200 --> 00:23:31.519
Okay.
00:23:32.079 --> 00:23:34.240
Macroeconomics and geopolitics.
00:23:34.480 --> 00:23:34.880
Yeah.
00:23:35.039 --> 00:23:37.920
And then of course you also sent me two articles just now, right?
00:23:38.079 --> 00:23:39.039
Like correct.
00:23:39.599 --> 00:23:42.000
Was that what you wanted, Brian?
00:23:42.319 --> 00:23:49.920
Like were those the articles you you wanted to analyze or like were those the the likes of you wanted the likes of those, is it?
00:23:50.160 --> 00:23:50.720
Yeah, correct.
00:23:50.799 --> 00:23:52.799
So maybe perhaps something similar.
00:23:53.200 --> 00:23:56.240
Um these kind of topics, will you say that?
00:23:56.559 --> 00:23:57.920
Something else similar.
00:23:58.480 --> 00:23:58.799
Okay.
00:23:59.599 --> 00:23:59.759
Okay.
00:24:00.640 --> 00:24:00.799
Okay.
00:24:01.200 --> 00:24:10.799
So yeah, for context, the two articles that I sent was uh one of it was not a litter of oil to pass straight of hormuz, expect$200 price tag.
00:24:11.200 --> 00:24:18.240
And then the second article was first India bound vessel to cross Hormuz amid crisis docked at Mumbai port.
00:24:18.559 --> 00:24:23.440
So I think that was this was what I was trying to explain in that um the headlines are flashy.
00:24:23.519 --> 00:24:28.000
Like they are saying one article is saying no vessels are passing through the strait at all.
00:24:28.160 --> 00:24:32.079
And then the second article is saying, oh, actually an Indian vessel passed through.
00:24:33.279 --> 00:24:35.599
So I think that was the nuance that I was trying to capture.
00:24:36.000 --> 00:24:36.240
I see.
00:24:36.400 --> 00:24:38.160
Okay, I'm just showing that on the screen now.
00:24:38.480 --> 00:24:40.079
Okay, three changes made.
00:24:40.240 --> 00:24:42.640
Um okay, how do we test again?
00:24:43.039 --> 00:24:51.440
Okay, so caveat is that this is currently still a script that is running on my computer.
00:24:51.839 --> 00:24:52.319
Right.
00:24:52.640 --> 00:24:56.720
So my computer basically needs to be remain on for this to even run.
00:24:56.880 --> 00:25:03.440
And it's running in my terminal, which is obviously not what you want you wanna be looking at, uh Brian.
00:25:03.759 --> 00:25:20.240
So there there is still probably a few, if not tens more, tens tens of hours more required to kind of make this production great, you know, to be able to share with your your team or be able to use with multiple people on it.
00:25:20.400 --> 00:25:21.119
Uh definitely.
00:25:21.359 --> 00:25:32.400
But but what we are doing now is just kind of show showing that we can get um get to the value of what we are um we are promising or what we are we are talking about.
00:25:32.559 --> 00:25:32.720
Yeah.
00:25:33.279 --> 00:25:33.680
Yeah.
00:25:34.319 --> 00:25:41.119
It was already impressive that you could just pull out the the facts, even if it was yeah.
00:25:41.440 --> 00:25:44.960
Yeah, so I'm just going to ask for the command to run.
00:25:45.200 --> 00:25:47.440
So just run these commands again.
00:25:47.759 --> 00:25:48.160
Short.
00:25:48.720 --> 00:25:49.200
Short?
00:25:49.359 --> 00:25:50.480
What is short?
00:25:50.799 --> 00:25:53.599
I'm just gonna ask it, you know, it's like it's an intuition teacher.
00:25:54.160 --> 00:25:57.039
Just read the code and not waste your tokens, my friend.
00:25:58.240 --> 00:25:59.359
That is true too.
00:25:59.519 --> 00:25:59.839
Yes.
00:26:00.480 --> 00:26:05.359
I am at 40 almost 40% now of the context window.
00:26:06.000 --> 00:26:06.640
Oh, that's okay.
00:26:07.039 --> 00:26:12.000
Limits it to eight articles max, medium is 20 articles, long is all articles, no limit.
00:26:12.319 --> 00:26:12.960
I see.
00:26:14.079 --> 00:26:15.039
Interesting.
00:26:15.359 --> 00:26:15.839
It's okay.
00:26:16.480 --> 00:26:16.559
Okay.
00:26:17.519 --> 00:26:19.839
Wait, it fetched a thousand articles now.
00:26:20.000 --> 00:26:20.640
Oh my gosh.
00:26:21.200 --> 00:26:24.160
I think it expanded the stuff that it wants to look at.
00:26:25.039 --> 00:26:25.599
Okay, wait.
00:26:25.680 --> 00:26:26.160
So short.
00:26:26.400 --> 00:26:27.359
Oh, you see, okay, what okay?
00:26:27.519 --> 00:26:29.839
Let's uh for those listening in, what's happening now?
00:26:30.240 --> 00:26:36.000
Right first it has um RSS feeds, which are all your news articles from different sources, right?
00:26:36.799 --> 00:26:42.319
Then it's uh fetching also from Google News RSS with specific keywords.
00:26:42.799 --> 00:26:43.680
So that's part two.
00:26:43.839 --> 00:26:44.400
You see that, right?
00:26:44.559 --> 00:26:49.279
Fetching Google News physical flows, crude oil shipment tanker.
00:26:49.839 --> 00:26:52.880
Or then it has an oil cargo reroute shipping.
00:26:53.759 --> 00:26:58.160
So it's not say very smart, but pretty smart at the same time.
00:26:58.799 --> 00:27:05.680
Because you are just brute forcing all the different types of uh scenarios that can happen.
00:27:06.000 --> 00:27:09.920
See, there's even uh OPEC, Saudi Arabia, Russia, oil.
00:27:10.480 --> 00:27:14.240
So if something happens outside of these keywords, then you will not pick up.
00:27:14.720 --> 00:27:16.559
What do you think of that, Brian?
00:27:16.880 --> 00:27:19.519
Um actually, I think this is actually what I what I want.
00:27:19.599 --> 00:27:21.279
Like this is what uh we see on the table.
00:27:21.519 --> 00:27:25.279
It's good enough, uh it's good enough to fill it covers most of the things we see.
00:27:25.519 --> 00:27:27.759
Yeah, correct, like almost 80, 90 percent.
00:27:27.839 --> 00:27:28.000
Yeah.
00:27:28.319 --> 00:27:35.200
The keywords that I just saw, it was also very relevant to what I try and shift out when I read through articles, yeah.
00:27:35.519 --> 00:27:36.240
No, but that's great.
00:27:36.319 --> 00:27:38.880
We uh 90% is good for for just one hour anyway.
00:27:39.039 --> 00:27:40.559
Yeah, I think I think 90% is okay.
00:27:41.039 --> 00:27:44.160
So um to scroll 100%, yeah.
00:27:44.319 --> 00:27:50.720
Yeah, so we we fetched the news, step one, fetch the news, step two, extracted the facts from the articles.
00:27:50.960 --> 00:28:00.000
Step three, we did a scenario analysis, and then step four generated the briefing, uh, which I can find somewhere.
00:28:00.240 --> 00:28:00.880
Yeah.
00:28:01.759 --> 00:28:02.559
Oh, okay.
00:28:02.799 --> 00:28:05.119
Wait, did it just replace the previous one?
00:28:05.279 --> 00:28:06.640
Yeah, yeah, replaced the one.
00:28:07.039 --> 00:28:08.319
Oh, but it's still South Korea.
00:28:08.400 --> 00:28:09.119
No, it's a bit weird.
00:28:09.200 --> 00:28:09.599
Oh, South East.
00:28:09.759 --> 00:28:10.559
South Korea will come out again.
00:28:10.640 --> 00:28:11.599
Yeah, investment bill.
00:28:11.680 --> 00:28:12.720
Uh probably thought it was quite.
00:28:13.119 --> 00:28:16.160
But Middle East is not India guest diversion, Cathay.
00:28:16.400 --> 00:28:17.920
Oh, close this one again, open again.
00:28:18.000 --> 00:28:19.039
Uh close, close, close.
00:28:19.440 --> 00:28:20.480
Oh, your API key.
00:28:21.440 --> 00:28:23.759
Uh this is weird, actually.
00:28:23.920 --> 00:28:24.799
This is weird.
00:28:25.200 --> 00:28:26.640
Today at 1022.
00:28:26.720 --> 00:28:27.920
Let me open it again.
00:28:28.160 --> 00:28:32.640
Okay, Middle East, India, South Korea, Pyongyang.
00:28:33.039 --> 00:28:35.200
Okay, so this is the new one.
00:28:35.519 --> 00:28:40.000
Um, I think the the first one was is is good.
00:28:40.079 --> 00:28:41.039
It's very targeted.
00:28:41.440 --> 00:28:43.279
First one, okay, on Middle East.
00:28:43.440 --> 00:28:44.319
And what makes you say that?
00:28:44.720 --> 00:28:53.440
Okay, so I think um what we generally want to see would be some keywords like Middle East, which is where most of the oil uh exports are.
00:28:53.599 --> 00:28:56.720
Uh maybe 20% of the world's uh oil exports.
00:28:56.799 --> 00:29:00.720
We also like to see, for example, product grades instead of just oil.
00:29:00.880 --> 00:29:04.400
We want to see, for example, jet fuel or diesel.
00:29:04.640 --> 00:29:10.000
So I think very specific product grids would be very important uh for us as well.
00:29:10.240 --> 00:29:10.720
Yeah.
00:29:11.440 --> 00:29:12.319
Which we have here, right?
00:29:12.400 --> 00:29:13.039
It's talking about jet.
00:29:13.359 --> 00:29:13.759
Yeah, correct.
00:29:13.839 --> 00:29:15.200
So that's what we have here.
00:29:16.880 --> 00:29:21.599
Anything else when you read through this that you will improve on or that you think is good?
00:29:22.400 --> 00:29:34.480
I think for now this is very good already, as in the fact that it can just scrape out uh the data from the articles, but um I think it's still a little bit too uh broad also.
00:29:34.720 --> 00:29:38.079
But this this first example right is is perfect, yeah.
00:29:43.279 --> 00:29:47.759
Yeah, source-wise, it's not really showing the source, right?
00:29:48.000 --> 00:29:49.279
Yeah, some bug there.
00:29:50.000 --> 00:29:50.640
Yeah.
00:29:51.039 --> 00:29:53.200
So I think that's one thing to improve on.
00:29:53.759 --> 00:29:55.920
It doesn't show the sources.
00:29:56.720 --> 00:30:01.920
Um so actually, I don't even know whether this is good, but you probably know because you've you might have read this already, right?
00:30:02.079 --> 00:30:03.279
Yeah, I think I read this somewhere before.
00:30:03.440 --> 00:30:05.440
Jet fuel prices actually shot up.
00:30:05.599 --> 00:30:08.160
Um because of the conflict.
00:30:08.480 --> 00:30:09.119
Yeah.
00:30:09.440 --> 00:30:13.039
So I know we have maybe like 10-ish minutes.
00:30:13.440 --> 00:30:23.920
I think what we could do to wrap it up, maybe is just to um talk about how um the sources are not showing up clearly.
00:30:25.119 --> 00:30:27.920
Give an example of what's the wrong thing that is showing.
00:30:28.480 --> 00:30:30.960
So not showing up clearly in the event analysis.
00:30:32.000 --> 00:30:34.240
Uh sis example this.
00:30:34.640 --> 00:30:42.079
So for these kind of things, we uh sometimes need to be a bit more pointed so it doesn't go around wasting your tokens searching for things.
00:30:43.519 --> 00:30:44.160
Yeah.
00:30:44.480 --> 00:30:45.119
Yeah.
00:30:45.359 --> 00:30:52.799
And then because we also want this to be okay, we also want this to be kind of usable, right?
00:30:52.960 --> 00:30:56.559
Um so if okay, I'm gonna prompt.
00:30:56.720 --> 00:31:08.319
Uh today we were really trying to see whether it could copy the thought process when it comes to you know looking through articles, and actually behind this thought process, there might have even been a prompt, right?
00:31:08.559 --> 00:31:12.880
And we go to analyze.py on the left side in the the code, yeah, yeah.
00:31:13.039 --> 00:31:13.839
Take a look at that.
00:31:14.000 --> 00:31:15.759
Yeah, analyze.py.
00:31:16.000 --> 00:31:19.279
So as you can see, there is a few prompts here already, right?
00:31:19.440 --> 00:31:20.720
You can see that.
00:31:20.960 --> 00:31:29.279
So it's running through Gemini to analyze the news as uh as a toy analyst.
00:31:29.839 --> 00:31:32.319
Yeah, then group the facts.
00:31:33.039 --> 00:31:34.400
Okay, down, down, down.
00:31:34.880 --> 00:31:35.279
Right.
00:31:35.519 --> 00:31:39.839
So this is why it appears in the format that we talk about, right?
00:31:40.240 --> 00:31:43.039
Yeah, yeah, because you see you give it quite quite good.
00:31:43.119 --> 00:31:47.200
So if you want to make changes, you edit the instructions here.
00:31:47.680 --> 00:31:52.319
Well, I think for a very brief evaluation of an article, it's already very good.
00:31:52.480 --> 00:31:55.119
Yeah, it will take some time to think about it also.
00:31:55.440 --> 00:31:58.559
This is the immediate correct, correct, correct, correct.
00:31:58.640 --> 00:31:59.759
Yeah, yeah.
00:32:00.079 --> 00:32:04.480
And in fact, the the confidence seems a bit iffy to me.
00:32:04.960 --> 00:32:16.720
Because you ask the the model to come out with a high, medium, low, but you don't give it what it means to be high, medium, low, then it will definitely hallucinate there.
00:32:17.200 --> 00:32:25.759
Yeah, so here's where you would actually manually go in, and it would be very valuable to type in what what a high thing would look like, you know.
00:32:25.920 --> 00:32:32.799
And then when it comes to of course, you know you might you might change your thought process someday, correct?
00:32:32.960 --> 00:32:36.400
Whether you want it to be a three-part framework of impact response and price.
00:32:36.640 --> 00:32:57.680
Um so here's where you would really want to like inspect and and be very careful about how the system is creating its own prompt, which is fantastic because one year ago, you know, putting prompt pipelines um together was very, very difficult to do um with just vibe coding.
00:32:58.079 --> 00:33:00.319
Not even one year ago, just a couple of months ago.
00:33:00.559 --> 00:33:06.559
Oh, like yeah, it's in doing yeah, you know, these kind of prompt pipelines are actually quite quite tough to do.
00:33:06.880 --> 00:33:08.720
We spent a lot of time working on the prompts.
00:33:09.279 --> 00:33:23.119
Last year we were working on some projects and most of the time was spent trying to tweak the prompt to give us certain output because the LM is just not behaving, and now it's like we literally we didn't do anything.
00:33:23.599 --> 00:33:24.720
Yeah, we were just there.
00:33:26.000 --> 00:33:26.720
Exactly.
00:33:26.960 --> 00:33:27.599
Yeah.
00:33:28.799 --> 00:33:31.200
So I mean I also have no job, so I don't know.
00:33:31.359 --> 00:33:32.720
That's why I'm on a podcast.
00:33:33.039 --> 00:33:35.680
Hopefully to become a full-time comedian or something.
00:33:36.160 --> 00:33:42.400
Yeah, I mean, of course, Brian, I can share this, uh, share whatever all the code that we made with you.
00:33:42.640 --> 00:33:45.839
Yeah, and you can you can kind of figure out how to change it.
00:33:46.160 --> 00:33:53.920
So you need to first get uh Cloud um subscription, the$30 one, and ask it to explain to you how this works, how to use all this thing.
00:33:54.160 --> 00:33:58.400
Or you could just use your your$30 like Chat GPT.
00:34:00.559 --> 00:34:04.799
And then just ask like, how do I run how do I run this uh Python script?
00:34:04.880 --> 00:34:08.239
And then how do I how do I put in my API key?
00:34:08.320 --> 00:34:10.000
And how do I do that safely?
00:34:10.239 --> 00:34:11.840
Because you don't want it to be leaked.
00:34:12.239 --> 00:34:12.880
Yeah.
00:34:13.280 --> 00:34:19.039
So I think I think that's how you could um theoretically take it forward from there.
00:34:19.280 --> 00:34:24.320
And then of course you could also play around with with uh clock code as well.
00:34:24.559 --> 00:34:35.199
So I just wanted to conclude this with a mermit flowchart, which I find quite helpful in just understanding you know the flow of data sometimes.
00:34:35.360 --> 00:34:40.800
Sometimes it gets very complicated, but so I'm not sure whether this will be hard to understand.
00:34:41.119 --> 00:34:45.360
So yeah, I mean maybe maybe Yao Hong if you don't mind just explaining this briefly.
00:34:45.679 --> 00:34:51.280
Alright, so this explains the code, uh what the how the code works, right?
00:34:51.360 --> 00:34:59.440
So earlier we talked about Google News, RSS feeds, the few sources we have for Reuters Energy, Reuters commodities, etc.
00:35:00.400 --> 00:35:02.079
If filter and dedupe.
00:35:02.159 --> 00:35:06.719
Okay, so dedupe is short form for D because a lot of the headlines repeat again, right?
00:35:06.800 --> 00:35:08.320
So we need a step to clean up.
00:35:08.480 --> 00:35:17.360
So in the last um 24 hour, I assume that it just filters by that, and they use title similarity with an 80% threshold.
00:35:17.519 --> 00:35:18.880
So how similar are they?
00:35:19.199 --> 00:35:22.239
Are they uh in in terms of the title?
00:35:22.480 --> 00:35:22.880
Right?
00:35:23.119 --> 00:35:28.880
And from here, Gemini would extract uh them with the hard facts only, right?
00:35:29.199 --> 00:35:34.559
Uh with some guidelines on the claims, the source bias label.
00:35:34.719 --> 00:35:36.000
I don't know what that means actually.
00:35:36.639 --> 00:35:38.400
So you see price and cost, yeah.
00:35:38.480 --> 00:35:38.800
Okay, anyway.
00:35:38.960 --> 00:35:39.760
Yeah, and then it goes in.
00:35:40.000 --> 00:35:42.079
Finally, they analyze the problem that we saw earlier.
00:35:42.480 --> 00:35:46.800
So this is actually quite helpful for someone who doesn't really understand code, right?
00:35:47.760 --> 00:35:50.880
You can kind of visualize what's happening at each step.
00:35:51.199 --> 00:35:54.159
Yeah, and I think that's still quite important, right?
00:35:54.239 --> 00:35:56.320
Because it's very important, yeah.
00:35:56.719 --> 00:36:05.440
You don't want to just like rely on the system and blindly accept things without knowing how it works or without even seeing the prompt that is going on behind it.
00:36:05.840 --> 00:36:09.119
Um so I thought it was just good to end off with this.
00:36:09.360 --> 00:36:14.960
So that's like a very nice um uh this is like sort of a mind map, right?
00:36:15.280 --> 00:36:15.679
Yeah.
00:36:16.000 --> 00:36:16.639
Correct, correct.
00:36:17.199 --> 00:36:18.320
On how the code works.
00:36:19.119 --> 00:36:20.239
Yes, yeah, yeah.
00:36:20.320 --> 00:36:22.320
So that's you can explain to your boss.
00:36:22.800 --> 00:36:24.559
Oh, like why I did this.
00:36:24.880 --> 00:36:26.639
Yeah, what how each step works, right?
00:36:26.960 --> 00:36:28.719
You will sound so much smarter.
00:36:28.960 --> 00:36:29.840
Yes, yeah.
00:36:30.079 --> 00:36:30.639
Yeah.
00:36:31.679 --> 00:36:32.480
Yeah, yeah.
00:36:32.960 --> 00:36:34.639
It is the new skill, right?
00:36:34.719 --> 00:36:40.400
To to be able to uh orchestrate, like or to be able to picture what the flows look like.
00:36:40.639 --> 00:36:43.119
But maybe I'll just stop sharing my screen.
00:36:43.440 --> 00:36:48.400
So maybe Brian, we had one hour just kind of jamming on this.
00:36:48.559 --> 00:36:55.440
Uh it was quite unconventional from our past two episodes because we skipped a lot of permissions.
00:36:56.000 --> 00:37:00.159
Yeah, and we just jumped right straight without really planning.
00:37:00.719 --> 00:37:13.199
Yeah, and maybe to caveat like that requires some experience so you wouldn't jump straight to to dangerously skipping permissions uh on your first day of like vibe coding.
00:37:13.440 --> 00:37:21.519
But maybe, maybe Brian, if there's any um any reflections on what you saw so far that you didn't manage to mention just now?
00:37:21.760 --> 00:37:22.800
I think overall it was very good.
00:37:23.119 --> 00:37:37.599
Genuinely, I thought it wouldn't work, but then like in such a short span of time, they managed to scrape out so many news articles, like hundreds over, and from there pull out um actual hard facts and from there also draw a conclusion.
00:37:37.760 --> 00:37:41.840
So I believe this was also quite different from the previous two episodes, right?
00:37:42.079 --> 00:37:47.199
So I think that was very interesting, and um I'll definitely take a look and try on my end also.
00:37:47.360 --> 00:37:49.360
Yeah, so thank you for thank you for that.
00:37:50.000 --> 00:37:50.559
Yeah, yeah.
00:37:51.840 --> 00:37:53.840
Anything you would like to close with?
00:37:54.800 --> 00:38:00.559
Uh there are quite a lot of few things that I would improve in the process for scoring and and evaluation.
00:38:00.639 --> 00:38:03.440
I'm we didn't get time to go through.
00:38:03.519 --> 00:38:09.280
Uh that makes the difference between vibe coding and actually making something in production.
00:38:09.440 --> 00:38:09.679
Yeah.
00:38:10.559 --> 00:38:12.239
But it's okay, one hour quite short.
00:38:12.320 --> 00:38:15.199
But yeah, so we got pretty far already.
00:38:15.280 --> 00:38:15.440
Yeah.
00:38:15.679 --> 00:38:18.719
Maybe if we had like another half an hour, right?
00:38:18.800 --> 00:38:24.639
Like the the eval evaluation process that Yao Hong's talking about.
00:38:24.880 --> 00:38:25.119
Correct.
00:38:25.360 --> 00:38:29.199
Did you do you mind sharing a bit how that half how that time would look like?
00:38:29.280 --> 00:38:42.880
Like in terms of how that time would look like uh we will probably get one of those or a few out of those articles and uh give it more quantitative in terms of how it calculates the scoring.
00:38:43.760 --> 00:38:48.960
And also like specifically which part of the text that we want to see, right?
00:38:49.599 --> 00:38:56.320
And then using those evals, we run the model through it and say, okay, this is the output that we want.
00:38:56.639 --> 00:39:03.360
So in case you make any changes prior to the steps, we should actually still come back to the output that we desire now.
00:39:04.639 --> 00:39:14.400
So it's a bit more involved and maybe too advanced for a beginner, but so maybe yeah, maybe not the right thing to do.
00:39:14.719 --> 00:39:15.039
Yeah, it's not a good idea.
00:39:15.199 --> 00:39:16.880
Just tell people that there's a possibility.
00:39:17.360 --> 00:39:18.480
We'll get there, we'll get there.
00:39:18.719 --> 00:39:36.639
If you didn't catch that, uh yeah, it's it's just you know, realizing that you know, I I think a lot of the outputs that AI will produce will take days, if not weeks, if if not even months, to improve sometimes.
00:39:36.800 --> 00:39:45.679
Uh maybe this one would be a bit would lie on the shorter time frame, but you would not immediately trust everything.
00:39:45.760 --> 00:39:59.440
I mean, you will be extremely skeptical until you do enough kind of evaluations and test that you know the outputs are actually something you could trust, and then you could go and have fun with your the extra time you have.
00:39:59.840 --> 00:40:01.360
Trying to tweak your models.
00:40:01.840 --> 00:40:02.400
Yeah.
00:40:02.960 --> 00:40:04.639
There's so much more to be done now, actually.
00:40:05.679 --> 00:40:05.920
Yeah.
00:40:06.079 --> 00:40:11.119
But but anyways, um Brian, uh um I'll send over the files, of course.
00:40:11.280 --> 00:40:15.679
Uh sorry we didn't get to do a bot, but I think the purpose was quite different today.
00:40:15.920 --> 00:40:25.199
Um we basically tried to see if uh we could build something that replicated the way that you would analyze uh news articles.
00:40:25.440 --> 00:40:30.960
So it seems like it it performed uh pretty up to expectations, if not more.
00:40:31.840 --> 00:40:34.960
Uh I will give it a six out of ten just plus.
00:40:35.840 --> 00:40:36.880
I think I think eight, lah.
00:40:37.840 --> 00:40:39.039
Oh yeah, that's so nice, right?
00:40:40.000 --> 00:40:41.920
Uh uncle more demanding.
00:40:42.159 --> 00:40:42.559
Yeah.
00:40:42.719 --> 00:40:45.039
But but thanks so much for joining us, Brian.
00:40:45.519 --> 00:40:46.559
Oh, thank you for having me.
00:40:46.719 --> 00:40:47.360
Thank you for helping.
00:40:48.400 --> 00:40:48.719
Yeah.
00:40:49.199 --> 00:40:49.760
All great.
00:40:49.840 --> 00:40:51.280
And yeah, have a good evening.
00:40:51.599 --> 00:40:52.320
Hey, you too.
00:40:52.559 --> 00:40:52.880
Alright.
00:40:53.039 --> 00:40:53.280
Alright.
00:40:53.519 --> 00:40:53.840
See ya.
00:40:54.719 --> 00:40:55.039
See you.
00:40:55.920 --> 00:40:58.960
Thanks so much for joining us for this episode.
00:40:59.119 --> 00:41:06.320
If you found this helpful, feel free to leave us a rating on your favorite podcasting apps or like, share, subscribe on YouTube.
00:41:06.800 --> 00:41:11.039
And if you want to get into the next episode, feel free to fill up the form as well.
00:41:11.280 --> 00:41:13.360
So, yeah, thanks so much for joining us.
00:41:13.519 --> 00:41:16.480
Uh, really appreciate that you even made it to the end.
00:41:16.800 --> 00:41:19.360
And see you on our next episode.
00:41:20.000 --> 00:41:20.480
Bye.