WEBVTT 00:00:00.000 --> 00:00:02.240 You look professional, Vincent. I'm 00:00:02.240 --> 00:00:04.080 I haven't seen this side to you. 00:00:04.080 --> 00:00:06.040 >> Let's level up this podcast, yeah? Let's 00:00:06.040 --> 00:00:08.040 start to draw diagrams directly on the 00:00:08.040 --> 00:00:10.720 whiteboard. But, was it in in in mirror 00:00:10.720 --> 00:00:12.720 for you, or was it like 00:00:12.720 --> 00:00:14.120 not mirrored? Cuz for me it was 00:00:14.120 --> 00:00:15.200 mirrored. 00:00:15.200 --> 00:00:16.640 >> Sorry, I wasn't looking 00:00:16.640 --> 00:00:18.640 like One topic I want to get to you is 00:00:18.640 --> 00:00:20.320 tech One topic I want to get to you is 00:00:20.320 --> 00:00:22.680 tech debt, right tech. Oh, yeah, it 00:00:22.680 --> 00:00:24.920 looks great. 00:00:24.920 --> 00:00:25.280 Bell. 00:00:25.280 --> 00:00:26.480 >> Does it? 00:00:26.480 --> 00:00:27.000 >> Yeah. 00:00:27.000 --> 00:00:27.480 >> Okay. 00:00:27.480 --> 00:00:29.160 >> Yeah. 00:00:29.160 --> 00:00:29.840 >> Cool. 00:00:29.840 --> 00:00:32.400 >> This is good. 00:00:32.400 --> 00:00:34.600 >> If I can get this room, cuz I just 00:00:34.600 --> 00:00:36.960 dropped into It's a pod. It's a working 00:00:36.960 --> 00:00:37.120 pod. 00:00:37.120 --> 00:00:39.960 >> Okay. So, maybe for the the context of 00:00:39.960 --> 00:00:42.880 our avid fan base 00:00:42.880 --> 00:00:43.885 >> of one. 00:00:43.885 --> 00:00:44.960 >> [laughter] 00:00:44.960 --> 00:00:46.680 >> Hello. Thank you for watching. 00:00:46.680 --> 00:00:48.160 Hello, Fred. 00:00:48.160 --> 00:00:49.960 >> It's been like a month since our last 00:00:49.960 --> 00:00:52.040 one. So, what happened to you? You got a 00:00:52.040 --> 00:00:55.440 you you got a you got a on-site job now. 00:00:55.440 --> 00:00:56.920 Can you Can you describe what happened 00:00:56.920 --> 00:00:59.000 to you, Vincent? Cuz for me nothing much 00:00:59.000 --> 00:01:01.040 has changed, really. 00:01:01.040 --> 00:01:02.040 >> No, I 00:01:02.040 --> 00:01:03.800 I have an office that I can go to, and I 00:01:03.800 --> 00:01:05.640 can meet people now, which I didn't have 00:01:05.640 --> 00:01:07.840 for the last 6 years. So, maybe I'll be 00:01:07.840 --> 00:01:10.400 less of a recluse. I'll I'll learn to 00:01:10.400 --> 00:01:12.480 behave. 00:01:12.480 --> 00:01:13.520 >> No, don't 00:01:13.520 --> 00:01:16.560 Don't say that, cuz my my client has has 00:01:16.560 --> 00:01:18.400 done one of those like return-to-office 00:01:18.400 --> 00:01:20.280 type policies, and like 00:01:20.280 --> 00:01:23.440 >> No, no. I said I didn't say I said I 00:01:23.440 --> 00:01:25.160 have an office I can go to. It doesn't 00:01:25.160 --> 00:01:27.040 mean I have to be here. Like a lot of 00:01:27.040 --> 00:01:28.520 the colleagues right now are not here. 00:01:28.520 --> 00:01:29.520 >> Yeah. But, 00:01:29.520 --> 00:01:31.760 >> It's I'm usually there on Wednesday. 00:01:31.760 --> 00:01:32.880 >> I guess 00:01:32.880 --> 00:01:35.520 I maybe I I jumped the gun there, 00:01:35.520 --> 00:01:36.720 though. 00:01:36.720 --> 00:01:38.480 It's good to be in office to some 00:01:38.480 --> 00:01:39.480 respect, right? 00:01:39.480 --> 00:01:41.000 >> Exactly. That's what I was in alliance 00:01:41.000 --> 00:01:43.120 with people. Yeah. 00:01:43.120 --> 00:01:44.840 >> How's it going? Are you still in 00:01:44.840 --> 00:01:47.760 probation period, and 00:01:47.760 --> 00:01:50.080 >> Yeah. I'm just having fun, right? 00:01:50.080 --> 00:01:51.400 Probation. 00:01:51.400 --> 00:01:53.040 It's It's just the best time to have 00:01:53.040 --> 00:01:53.920 fun, isn't it? 00:01:53.920 --> 00:01:55.480 >> What What What is your What is your 00:01:55.480 --> 00:01:56.880 role? Can you talk about your work, or 00:01:56.880 --> 00:01:58.360 you can talk about other things? 00:01:58.360 --> 00:02:00.280 >> My role is the most ridiculous. I see 00:02:00.280 --> 00:02:02.520 myself Let's see Let's say how I saw 00:02:02.520 --> 00:02:02.800 this 00:02:02.800 --> 00:02:04.440 >> I see myself. 00:02:04.440 --> 00:02:07.160 My title does not see me. 00:02:07.160 --> 00:02:09.960 >> Well, 00:02:09.960 --> 00:02:12.280 I'm not going to say much, but I see my 00:02:12.280 --> 00:02:16.280 my role more as like focused on 00:02:16.280 --> 00:02:19.400 approaching AI adoption from a platform 00:02:19.400 --> 00:02:22.040 perspective. So, the the tricky bits is 00:02:22.040 --> 00:02:23.600 that of course they already have a 00:02:23.600 --> 00:02:25.920 massive platform team, even a dedicated 00:02:25.920 --> 00:02:28.280 AI platform team. So, when I joined I 00:02:28.280 --> 00:02:30.000 was quite 00:02:30.000 --> 00:02:33.240 scared, but I saw gaps and I feel like I 00:02:33.240 --> 00:02:35.480 can 00:02:35.480 --> 00:02:37.920 I can address some of those gaps. 00:02:37.920 --> 00:02:39.680 Let's see how long 00:02:39.680 --> 00:02:41.360 my views survive. 00:02:41.360 --> 00:02:46.640 >> So, which gap are you working on? 00:02:46.640 --> 00:02:47.920 You can't say. 00:02:47.920 --> 00:02:52.080 >> Trying to 00:02:52.080 --> 00:02:54.320 highlight where agents struggle in 00:02:54.320 --> 00:02:57.000 existing pipelines systems. 00:02:57.000 --> 00:02:59.400 Like, you know, you have an established 00:02:59.400 --> 00:03:02.080 It's It's a It's a highly 00:03:02.080 --> 00:03:05.040 sensitive environment. So, 00:03:05.040 --> 00:03:07.519 compliance requirements and so on. So, 00:03:07.519 --> 00:03:09.560 there's a lot of of of blockers there 00:03:09.560 --> 00:03:11.519 that are there for a good reason, right? 00:03:11.519 --> 00:03:13.320 It's not a startup where, you know, you 00:03:13.320 --> 00:03:15.440 want to deploy in S2 to production as 00:03:15.440 --> 00:03:18.800 fast as possible. There's a very strict 00:03:18.800 --> 00:03:21.920 chokehold that the platform team has for 00:03:21.920 --> 00:03:23.840 good reasons. So, that's kind of like 00:03:23.840 --> 00:03:26.239 the probably the most challenging part 00:03:26.239 --> 00:03:29.000 being, you know, I wanted to be focused 00:03:29.000 --> 00:03:32.680 on AI adoption. Even though lately a lot 00:03:32.680 --> 00:03:35.519 of people are like 00:03:35.519 --> 00:03:38.440 really anti-AI, more I feel like. I 00:03:38.440 --> 00:03:40.400 don't know, maybe. Not not in this 00:03:40.400 --> 00:03:42.560 company, but like generally I think the 00:03:42.560 --> 00:03:44.640 people I think 00:03:44.640 --> 00:03:47.000 >> I'm willing to accept that there 00:03:47.000 --> 00:03:49.239 there there's certainly places where you 00:03:49.239 --> 00:03:51.760 want a human in the loop. Though, at the 00:03:51.760 --> 00:03:53.200 same time 00:03:53.200 --> 00:03:57.920 like given an incident or given a 00:03:57.920 --> 00:04:00.360 given a use case, you want you want the 00:04:00.360 --> 00:04:02.440 agent to be unlocked to be able to 00:04:02.440 --> 00:04:05.000 perform it in a reasonable 00:04:05.000 --> 00:04:07.160 manner in a in a sense. So, 00:04:07.160 --> 00:04:09.240 >> It's it's about signal versus noise. I 00:04:09.240 --> 00:04:10.920 don't I never I never said that I want 00:04:10.920 --> 00:04:12.680 to take humans out of the loop, right? 00:04:12.680 --> 00:04:15.680 It's just where does I think Autopilot 00:04:15.680 --> 00:04:17.959 is a great example. Autopilot was able 00:04:17.959 --> 00:04:21.160 to reduce the signal the noise. We don't 00:04:21.160 --> 00:04:23.800 just press enter prompt. It's able to 00:04:23.800 --> 00:04:25.400 surface very clearly where your 00:04:25.400 --> 00:04:27.360 attention is needed. Says like hold on, 00:04:27.360 --> 00:04:29.200 you know, agent is doing something here. 00:04:29.200 --> 00:04:30.800 That probably shouldn't be 00:04:30.800 --> 00:04:32.120 >> Sorry, the signal wasn't that great 00:04:32.120 --> 00:04:33.919 there. 00:04:33.919 --> 00:04:35.120 Yeah. 00:04:35.120 --> 00:04:36.640 Interesting. 00:04:36.640 --> 00:04:39.000 Yeah, there there is pushback from AI 00:04:39.000 --> 00:04:41.440 from from some people that I 00:04:41.440 --> 00:04:43.840 I went to like an onsite the other day 00:04:43.840 --> 00:04:45.360 and met my colleagues. So, it was 00:04:45.360 --> 00:04:47.600 interesting to have beers with them 00:04:47.600 --> 00:04:49.240 because, you know, 00:04:49.240 --> 00:04:50.840 then you hear 00:04:50.840 --> 00:04:53.080 people's hopes behind the scene. 00:04:53.080 --> 00:04:55.400 Hopes and fears around AI, right? 00:04:55.400 --> 00:04:58.640 >> Yeah. 00:04:58.640 --> 00:05:00.280 So, 00:05:00.280 --> 00:05:03.560 to to jump into one topic I was keen to 00:05:03.560 --> 00:05:07.560 hear from you about 00:05:07.560 --> 00:05:11.440 tech debt. 00:05:11.440 --> 00:05:13.760 How do you Do you have any tips like I 00:05:13.760 --> 00:05:15.360 remember using a 00:05:15.360 --> 00:05:17.919 a SaaS service back with the Italian 00:05:17.919 --> 00:05:19.960 client where where all the 00:05:19.960 --> 00:05:21.960 vulnerabilities were like brought in. 00:05:21.960 --> 00:05:24.040 But, I think Datadog can do this. 00:05:24.040 --> 00:05:25.840 But, anyway, all the vulnerabilities 00:05:25.840 --> 00:05:30.080 were all brought into one dashboard. 00:05:30.080 --> 00:05:32.919 Single pane and we're able we were able 00:05:32.919 --> 00:05:35.280 to like sort of accept the risk or you 00:05:35.280 --> 00:05:36.919 know, prioritize which ones we're going 00:05:36.919 --> 00:05:39.440 to do. I just had I just had to ask you 00:05:39.440 --> 00:05:43.280 quickly if you had any experience of 00:05:43.280 --> 00:05:45.480 using Datadog or something else to to a 00:05:45.480 --> 00:05:47.040 whole bunch of vulnerabilities or or 00:05:47.040 --> 00:05:48.280 tech debt. 00:05:48.280 --> 00:05:50.360 But the since I'm in security, the the 00:05:50.360 --> 00:05:52.160 tech debts are seems to be out of date 00:05:52.160 --> 00:05:55.880 dependencies essentially. 00:05:55.880 --> 00:05:58.360 >> I had quite a few discussions lately as 00:05:58.360 --> 00:06:00.520 I'm in, you know, coming into this new 00:06:00.520 --> 00:06:03.120 role and talking to people about the 00:06:03.120 --> 00:06:05.440 business value and like 00:06:05.440 --> 00:06:07.880 I think it's it the way that 00:06:07.880 --> 00:06:10.040 the team the team I work with uh sees 00:06:10.040 --> 00:06:12.800 tech debt is more of like 00:06:12.800 --> 00:06:15.040 a ceiling of capabilities. 00:06:15.040 --> 00:06:17.240 >> So, as you as you deliver work, you're 00:06:17.240 --> 00:06:18.240 you're 00:06:18.240 --> 00:06:19.680 you know, elevating the ceiling of 00:06:19.680 --> 00:06:21.240 capabilities. 00:06:21.240 --> 00:06:23.320 And until you hit the ceiling, you're 00:06:23.320 --> 00:06:25.320 not forced to make changes on those. 00:06:25.320 --> 00:06:26.640 Like the ceiling 00:06:26.640 --> 00:06:28.560 maybe, you know, able to handle a 00:06:28.560 --> 00:06:31.200 certain capacity. Cuz tech debt in 00:06:31.200 --> 00:06:33.080 general is not just about security 00:06:33.080 --> 00:06:35.160 vulnerabilities that are pending. It's 00:06:35.160 --> 00:06:37.440 also about like, you know, where where 00:06:37.440 --> 00:06:39.360 every week we have to reset the database 00:06:39.360 --> 00:06:42.120 server because there's this background 00:06:42.120 --> 00:06:45.440 job or we need to truncate the logs. 00:06:45.440 --> 00:06:46.200 >> Yeah. 00:06:46.200 --> 00:06:47.960 >> And and things like that are, you know, 00:06:47.960 --> 00:06:49.360 generally not good but they are 00:06:49.360 --> 00:06:51.320 >> of the ceiling. 00:06:51.320 --> 00:06:53.600 The ceiling cuz I I'm assuming you can 00:06:53.600 --> 00:06:56.680 use that as a as a metaphor to describe 00:06:56.680 --> 00:06:58.840 that your team has 00:06:58.840 --> 00:07:00.760 essentially too much toil, right? And 00:07:00.760 --> 00:07:02.520 then you need you need to 00:07:02.520 --> 00:07:04.640 you need to be able to lift that ceiling 00:07:04.640 --> 00:07:06.280 by adding, I don't know, a new team 00:07:06.280 --> 00:07:08.120 member or uh 00:07:08.120 --> 00:07:10.040 addressing the tech debt. Otherwise, 00:07:10.040 --> 00:07:12.240 you're you're you're over capacity or 00:07:12.240 --> 00:07:13.960 something. I don't know. 00:07:13.960 --> 00:07:15.600 If you 00:07:15.600 --> 00:07:17.080 If I if I understood your analogy 00:07:17.080 --> 00:07:17.680 correctly. 00:07:17.680 --> 00:07:19.040 >> It's it's more of 00:07:19.040 --> 00:07:20.880 of capabilities of the of the 00:07:20.880 --> 00:07:23.160 architecture and the software, not not 00:07:23.160 --> 00:07:25.360 just the team. So, when you 00:07:25.360 --> 00:07:27.200 I'm I can't claim this the way of 00:07:27.200 --> 00:07:29.000 looking at the at at the 00:07:29.000 --> 00:07:33.000 at at software and software development, 00:07:33.000 --> 00:07:36.160 that view is not mine. But I do like the 00:07:36.160 --> 00:07:37.760 way that it reframes because the way 00:07:37.760 --> 00:07:41.240 that like Google and SRE identify, 00:07:41.240 --> 00:07:43.880 sorry, or did define tech that. It's 00:07:43.880 --> 00:07:45.960 more about, you know, your error budget 00:07:45.960 --> 00:07:48.200 and other things that trigger 00:07:48.200 --> 00:07:50.800 refactoring or trigger different like 00:07:50.800 --> 00:07:53.320 maybe a code freeze. Cuz cuz what is 00:07:53.320 --> 00:07:55.080 tech that, right? I have a lot of 00:07:55.080 --> 00:07:57.120 arguments in previous companies where I 00:07:57.120 --> 00:07:58.800 go like, oh you know, this all of this 00:07:58.800 --> 00:08:01.040 bash this massive bash script that runs 00:08:01.040 --> 00:08:03.560 in our deployment is tech that. Like we 00:08:03.560 --> 00:08:05.240 need to to get rid of this. We need to 00:08:05.240 --> 00:08:07.160 split up the deployment pipeline. We 00:08:07.160 --> 00:08:09.480 need to support parallel because it's a 00:08:09.480 --> 00:08:10.800 microservice architecture we need to 00:08:10.800 --> 00:08:12.600 support parallel deployments and things 00:08:12.600 --> 00:08:14.560 like that. Yeah and and dependency like 00:08:14.560 --> 00:08:16.480 a proper mono repo build tool and and 00:08:16.480 --> 00:08:18.240 then they might might might team lead 00:08:18.240 --> 00:08:19.919 would be like, no. 00:08:19.919 --> 00:08:22.680 It works. It's not tech that, you know, 00:08:22.680 --> 00:08:24.760 it's it's all of that what you mentioned 00:08:24.760 --> 00:08:25.840 is nice to have. 00:08:25.840 --> 00:08:29.360 >> Well well in my context it's it's like 00:08:29.360 --> 00:08:31.520 a large organization, 00:08:31.520 --> 00:08:33.599 lots of teams, 00:08:33.599 --> 00:08:35.159 and what's happening is that we get 00:08:35.159 --> 00:08:37.599 these alerts from different systems that 00:08:37.599 --> 00:08:40.039 but that dependencies are out of date or 00:08:40.039 --> 00:08:42.560 or vulnerable. 00:08:42.560 --> 00:08:45.839 So that's that's that's my tech debt. 00:08:45.839 --> 00:08:48.320 >> And that's just one, yeah. I mean that's 00:08:48.320 --> 00:08:49.080 just 00:08:49.080 --> 00:08:51.280 >> I work kind of roll my eyes because many 00:08:51.280 --> 00:08:53.960 of these internal 00:08:53.960 --> 00:08:56.120 dependencies many of these systems are 00:08:56.120 --> 00:08:58.520 only internal and many of the the the 00:08:58.520 --> 00:09:00.360 dependencies 00:09:00.360 --> 00:09:02.880 are super low risk in my mind. 00:09:02.880 --> 00:09:05.160 But they but they're flagged as high or 00:09:05.160 --> 00:09:07.280 critical and uh 00:09:07.280 --> 00:09:09.040 what I'm struggling to do is just 00:09:09.040 --> 00:09:11.520 essentially I've done this before. 00:09:11.520 --> 00:09:12.839 I feel 00:09:12.839 --> 00:09:14.400 like an idiot for not remembering how I 00:09:14.400 --> 00:09:17.839 did it. Is is just to essentially uh 00:09:17.839 --> 00:09:20.160 manage them and and govern them and and 00:09:20.160 --> 00:09:21.800 be able to sort of flag them as like 00:09:21.800 --> 00:09:22.560 accepted. 00:09:22.560 --> 00:09:23.320 >> Classify. 00:09:23.320 --> 00:09:25.600 >> I'm disregarding this these ones because 00:09:25.600 --> 00:09:27.640 it's you know, it's internal. It's like 00:09:27.640 --> 00:09:30.000 a system or maybe I have to maybe I have 00:09:30.000 --> 00:09:32.160 to be real and say like, hey, this team 00:09:32.160 --> 00:09:34.160 doesn't exist anymore, I'm afraid to 00:09:34.160 --> 00:09:37.480 say. And it's running like it is and we 00:09:37.480 --> 00:09:39.400 just don't have the 00:09:39.400 --> 00:09:40.480 the 00:09:40.480 --> 00:09:43.800 capability to update this one. 00:09:43.800 --> 00:09:45.200 You know what I mean? Like I'm I'm 00:09:45.200 --> 00:09:47.880 looking for that sort of uh 00:09:47.880 --> 00:09:50.280 governance tool more than anything else. 00:09:50.280 --> 00:09:51.840 I mean 00:09:51.840 --> 00:09:53.280 Yeah, I mean 00:09:53.280 --> 00:09:54.520 I know what you're saying. 00:09:54.520 --> 00:09:55.480 I mean 00:09:55.480 --> 00:09:56.760 when we were looking at Twistlock 00:09:56.760 --> 00:09:58.400 backend a long time ago when I was 00:09:58.400 --> 00:10:00.160 setting up for a for the Singapore 00:10:00.160 --> 00:10:02.160 government integration, like we were 00:10:02.160 --> 00:10:03.400 start cuz 00:10:03.400 --> 00:10:05.440 they have like this IM8 requirements. 00:10:05.440 --> 00:10:07.640 What I wanted to get to is they have 00:10:07.640 --> 00:10:10.640 like a IT standard that they're they 00:10:10.640 --> 00:10:13.120 judge architecture upon and you must, 00:10:13.120 --> 00:10:15.920 you know, be compliant. Uh for example, 00:10:15.920 --> 00:10:19.320 have a three-tier architecture, have the 00:10:19.320 --> 00:10:22.600 the limit the DMZ zone, only allow 00:10:22.600 --> 00:10:25.320 ingress on your data layer on a 00:10:25.320 --> 00:10:27.200 different You know, that's IM8 like 00:10:27.200 --> 00:10:28.080 then. 00:10:28.080 --> 00:10:30.160 And one of the requirements was with 00:10:30.160 --> 00:10:32.200 Twistlock and an intrusion detection and 00:10:32.200 --> 00:10:35.960 prevention. So, IDS and IDP. So, you 00:10:35.960 --> 00:10:38.800 need to detect and but also 00:10:38.800 --> 00:10:41.680 constantly scanning for like workloads 00:10:41.680 --> 00:10:44.040 and vulnerabilities in those workloads. 00:10:44.040 --> 00:10:45.480 And one of the most, you know, 00:10:45.480 --> 00:10:46.960 advertised features of some of these 00:10:46.960 --> 00:10:48.360 platforms I was looking at at the time, 00:10:48.360 --> 00:10:50.080 Aqua Security, Twistlock, 00:10:50.080 --> 00:10:52.080 >> I'm familiar with Twistlock, but what is 00:10:52.080 --> 00:10:52.320 that? 00:10:52.320 --> 00:10:55.600 >> Twistlock, they were acquired by 00:10:55.600 --> 00:10:57.720 I forgot. They were acquired by a big 00:10:57.720 --> 00:10:59.400 brand of but 00:10:59.400 --> 00:11:01.400 they were an Israel security startup. 00:11:01.400 --> 00:11:03.640 And there was another one, Falcon from 00:11:03.640 --> 00:11:06.120 Sysdig. Yeah, Sysdig. Uh Falcon, those 00:11:06.120 --> 00:11:07.320 are the ones that I was evaluating 00:11:07.320 --> 00:11:09.520 because we were Kubernetes cloud native 00:11:09.520 --> 00:11:11.640 platform. And so, one of the hardest 00:11:11.640 --> 00:11:13.000 things is like, "Hey, Kubernetes, you 00:11:13.000 --> 00:11:14.560 have this massive cluster that runs all 00:11:14.560 --> 00:11:16.560 the workloads, batch jobs, and all. But 00:11:16.560 --> 00:11:18.400 then IM8 is like very strict. You have 00:11:18.400 --> 00:11:20.080 to have your ingress layer living on 00:11:20.080 --> 00:11:21.920 completely separate nodes with like 00:11:21.920 --> 00:11:23.680 natural access control lists." And if 00:11:23.680 --> 00:11:25.040 you have a Kubernetes cluster, that's 00:11:25.040 --> 00:11:28.080 hard. You know, even if you you you 00:11:28.080 --> 00:11:30.080 taint ingress nodes so no other 00:11:30.080 --> 00:11:32.200 workloads can live on it, how do you 00:11:32.200 --> 00:11:34.360 ensure that nothing tolerates the taint 00:11:34.360 --> 00:11:35.880 and runs on the on the ingress layer 00:11:35.880 --> 00:11:37.680 anyway? So, you need to have some web 00:11:37.680 --> 00:11:39.240 hook and admission hook and and all 00:11:39.240 --> 00:11:42.360 kinds of enforcement. So, with Twistlock 00:11:42.360 --> 00:11:44.560 we were able to to uh 00:11:44.560 --> 00:11:47.040 achieve some of that. 00:11:47.040 --> 00:11:48.400 What I was trying to get to was that the 00:11:48.400 --> 00:11:50.320 security reports, like what you say. 00:11:50.320 --> 00:11:51.800 We're running CoreOS, which was like 00:11:51.800 --> 00:11:53.040 this super 00:11:53.040 --> 00:11:55.680 you know, con- container OS. 00:11:55.680 --> 00:11:57.480 Yeah, so so the kernel like it's a very 00:11:57.480 --> 00:11:59.160 minimal host operating system. 00:11:59.160 --> 00:11:59.680 >> Yeah, yeah. 00:11:59.680 --> 00:12:02.080 >> And and a very up-to-date kernel cuz 00:12:02.080 --> 00:12:04.480 it's like very aggressively upgrading 00:12:04.480 --> 00:12:06.480 updating as well. It's very security 00:12:06.480 --> 00:12:08.080 focused, right? CoreOS, I mean, the 00:12:08.080 --> 00:12:10.200 company before they 00:12:10.200 --> 00:12:11.280 >> Who didn't they join 00:12:11.280 --> 00:12:12.920 >> No, they they were acquired by Red Hat 00:12:12.920 --> 00:12:15.160 and then their Kubernetes uh yeah, 00:12:15.160 --> 00:12:17.440 solution wasn't was basically replacing 00:12:17.440 --> 00:12:20.000 Red Hat OpenShift. They replaced 00:12:20.000 --> 00:12:20.640 Tectonic. 00:12:20.640 --> 00:12:22.400 >> Yeah, I mean I mean I'm assuming when 00:12:22.400 --> 00:12:25.000 you run Kubernetes on the cloud it does 00:12:25.000 --> 00:12:27.680 the same thing as CoreOS, you know. 00:12:27.680 --> 00:12:29.400 But CoreOS was cool cuz you could have 00:12:29.400 --> 00:12:30.880 got got the same feeling on your own 00:12:30.880 --> 00:12:32.080 machine. 00:12:32.080 --> 00:12:34.560 >> CoreOS the the the biggest thing that 00:12:34.560 --> 00:12:36.760 CoreOS did was immutable architecture 00:12:36.760 --> 00:12:38.720 with dual dual boot. I think we talked 00:12:38.720 --> 00:12:40.040 about this in a previous uh 00:12:40.040 --> 00:12:42.080 >> Yeah, yeah, yeah. 00:12:42.080 --> 00:12:43.360 >> But what I was getting to was the 00:12:43.360 --> 00:12:45.720 security scanning and the reports and 00:12:45.720 --> 00:12:47.640 the Twistlock vulnerability reports 00:12:47.640 --> 00:12:49.680 exactly your problem, which is like 00:12:49.680 --> 00:12:51.680 being able to to to flag certain 00:12:51.680 --> 00:12:54.880 critical findings as you know, you have 00:12:54.880 --> 00:12:56.360 to put the reason why it's not yet 00:12:56.360 --> 00:12:58.000 addressed, right? Or you have to put it 00:12:58.000 --> 00:13:00.080 in like a backlog and and a timeline 00:13:00.080 --> 00:13:02.000 against when it's going to be addressed. 00:13:02.000 --> 00:13:04.280 And we we we Twistlock helped you with 00:13:04.280 --> 00:13:05.600 those things. 00:13:05.600 --> 00:13:07.000 The big selling point of those 00:13:07.000 --> 00:13:09.080 frameworks was to like, hey, you know, 00:13:09.080 --> 00:13:10.400 most of these solutions, definitely if 00:13:10.400 --> 00:13:12.080 you go open source, they just it's very 00:13:12.080 --> 00:13:13.760 easy to get like a full scan of all of 00:13:13.760 --> 00:13:15.360 the possible vulnerabilities and drop 00:13:15.360 --> 00:13:17.880 and and and dumped at you. But then what 00:13:17.880 --> 00:13:19.839 the differentiator is for the commercial 00:13:19.839 --> 00:13:21.720 offerings is to basically help you with 00:13:21.720 --> 00:13:23.640 exactly that. Like, that's where they 00:13:23.640 --> 00:13:27.440 are, you know, making you pay. 00:13:27.440 --> 00:13:29.160 >> Yeah. 00:13:29.160 --> 00:13:30.520 Okay, I'll I had I have to do some 00:13:30.520 --> 00:13:32.120 research. I was just I was just hoping 00:13:32.120 --> 00:13:37.800 that you had a quick solution for me. 00:13:37.800 --> 00:13:39.320 But I mean yeah, I'm sure I'm sure I get 00:13:39.320 --> 00:13:42.160 there. Hey, I wanted to show you 00:13:42.160 --> 00:13:44.800 change the topic to something that I 00:13:44.800 --> 00:13:46.800 might have talked to you about 00:13:46.800 --> 00:13:50.440 about decision-making in organization. 00:13:50.440 --> 00:13:52.240 So maybe this is just me having a chip 00:13:52.240 --> 00:13:54.080 on my shoulder as we say in English. I 00:13:54.080 --> 00:13:56.320 don't know what you say in Belgium. 00:13:56.320 --> 00:13:59.240 Though I'm often irritated 00:13:59.240 --> 00:14:01.560 by it's not just my own employer, it's 00:14:01.560 --> 00:14:04.520 like the different clients find it very 00:14:04.520 --> 00:14:08.360 curious how they poorly communicate a 00:14:08.360 --> 00:14:09.560 policy. 00:14:09.560 --> 00:14:11.720 Right? Like you you were talking about 00:14:11.720 --> 00:14:14.160 the public sector earlier where they had 00:14:14.160 --> 00:14:17.280 like I am a I am a they had like a like 00:14:17.280 --> 00:14:20.320 a good spec and I think Singapore yeah, 00:14:20.320 --> 00:14:21.200 like 00:14:21.200 --> 00:14:23.120 for the most part there was good 00:14:23.120 --> 00:14:25.560 leadership and good leadership for me is 00:14:25.560 --> 00:14:26.920 essentially 00:14:26.920 --> 00:14:28.440 very boring and this 00:14:28.440 --> 00:14:30.120 it's basically writing things down and 00:14:30.120 --> 00:14:31.720 sharing what that that writing is and 00:14:31.720 --> 00:14:33.280 making sure everyone's aligned on that 00:14:33.280 --> 00:14:35.000 that piece of writing. 00:14:35.000 --> 00:14:37.240 So the thing that I've noticed in many 00:14:37.240 --> 00:14:39.320 in many of my gigs is that you might 00:14:39.320 --> 00:14:42.400 have a town hall or you might have like 00:14:42.400 --> 00:14:45.040 a social media post nowadays like like a 00:14:45.040 --> 00:14:47.120 leader saying like, "Oh, we are changing 00:14:47.120 --> 00:14:49.520 the policy on this. I hope you feel 00:14:49.520 --> 00:14:50.800 comfortable this. If you have any 00:14:50.800 --> 00:14:52.920 feedback but like you might not like you 00:14:52.920 --> 00:14:54.960 know, since these different mediums like 00:14:54.960 --> 00:14:57.760 you know, like a YouTube or a like a 00:14:57.760 --> 00:15:01.400 Confluence and or Workvivo or they're 00:15:01.400 --> 00:15:03.240 all they're all like kind of different 00:15:03.240 --> 00:15:07.040 mediums difficult to to track a but but 00:15:07.040 --> 00:15:08.880 they usually don't have a single source 00:15:08.880 --> 00:15:12.440 of truth. They don't have like a 00:15:12.440 --> 00:15:15.120 a URL like you know, company decisions. 00:15:15.120 --> 00:15:16.320 dot 00:15:16.320 --> 00:15:18.400 whatever.com/the 00:15:18.400 --> 00:15:20.480 decision. 00:15:20.480 --> 00:15:24.040 So this inspired me to to vibe code. 00:15:24.040 --> 00:15:26.200 >> You You shared the link. 00:15:26.200 --> 00:15:26.520 >> Yeah. 00:15:26.520 --> 00:15:28.680 >> Right. I was just going to open it. 00:15:28.680 --> 00:15:30.760 >> Yeah. So, this basically 00:15:30.760 --> 00:15:33.760 inspired me to to vibe code something. 00:15:33.760 --> 00:15:34.720 And 00:15:34.720 --> 00:15:36.320 yeah, like I don't know. Do you think 00:15:36.320 --> 00:15:37.520 I'm just making a mountain of a 00:15:37.520 --> 00:15:40.240 molehill? Like I It's It's the same 00:15:40.240 --> 00:15:41.720 thing like I'm 00:15:41.720 --> 00:15:43.400 I'm sure you know what architecture 00:15:43.400 --> 00:15:45.560 architecture decisions 00:15:45.560 --> 00:15:47.400 But it's the same concept. Like 00:15:47.400 --> 00:15:49.680 essentially, you make a decision. 00:15:49.680 --> 00:15:52.000 You have a URL. 00:15:52.000 --> 00:15:54.760 You You copy it around. And And of 00:15:54.760 --> 00:15:57.280 course, in my mind, good communication 00:15:57.280 --> 00:16:00.040 is repeating the message. Saying like 00:16:00.040 --> 00:16:00.920 you 00:16:00.920 --> 00:16:02.720 the same message all the time. But like, 00:16:02.720 --> 00:16:04.400 "Hey, if you want to know more, check 00:16:04.400 --> 00:16:06.560 out decisions. whatever." 00:16:06.560 --> 00:16:09.120 And you will or follow this QR code and 00:16:09.120 --> 00:16:11.480 you will get um 00:16:11.480 --> 00:16:13.840 you will uh 00:16:13.840 --> 00:16:16.280 see see exactly all all about it. So, I 00:16:16.280 --> 00:16:18.640 I did get a little bit crazy. I did even 00:16:18.640 --> 00:16:21.080 define a schema. 00:16:21.080 --> 00:16:23.320 And make sure that And when I build it, 00:16:23.320 --> 00:16:25.320 it sort of like validates it. You know, 00:16:25.320 --> 00:16:28.520 you you can have like a like an RFC or a 00:16:28.520 --> 00:16:30.839 W3C spec, you can have like 00:16:30.839 --> 00:16:34.600 >> Does it Does it have an MCP? 00:16:34.600 --> 00:16:36.600 >> No, it doesn't. But it It has an 00:16:36.600 --> 00:16:39.440 endpoint and an LLM's text. 00:16:39.440 --> 00:16:42.360 It generates LLM's text, dude. 00:16:42.360 --> 00:16:45.360 >> Does it then have like an open API or 00:16:45.360 --> 00:16:46.839 some other schema 00:16:46.839 --> 00:16:49.640 to validate it? 00:16:49.640 --> 00:16:53.240 >> Uh yeah, it's Well, it's It's just got a 00:16:53.240 --> 00:16:55.839 YAML schema. So, when you when you 00:16:55.839 --> 00:16:58.480 actually write a decision, you you mark 00:16:58.480 --> 00:17:01.200 it up and it validates that 00:17:01.200 --> 00:17:03.839 these fields are are are set. And that's 00:17:03.839 --> 00:17:07.079 about it. 00:17:07.079 --> 00:17:08.920 >> It's like ADR, but 00:17:08.920 --> 00:17:09.720 uh 00:17:09.720 --> 00:17:11.839 but then I guess 00:17:11.839 --> 00:17:13.400 >> But this is what it's supposed to be 00:17:13.400 --> 00:17:15.079 self-hosted. 00:17:15.079 --> 00:17:17.400 Yeah, it's self-hosted. You can All need 00:17:17.400 --> 00:17:19.360 to do is 00:17:19.360 --> 00:17:22.199 well, ask Claude to build it or or run 00:17:22.199 --> 00:17:27.760 this build.py or this makefile. 00:17:27.760 --> 00:17:29.480 I I I mean 00:17:29.480 --> 00:17:32.520 it's never mind the implementation, but 00:17:32.520 --> 00:17:35.200 but do you often feel that communication 00:17:35.200 --> 00:17:37.320 in a company is could be better or I 00:17:37.320 --> 00:17:39.600 mean, have you seen better communication 00:17:39.600 --> 00:17:47.200 than what I'm envisioning here? 00:17:47.200 --> 00:17:49.680 >> I mean 00:17:49.680 --> 00:17:52.960 I'm in an environment now where 00:17:52.960 --> 00:17:54.400 I talk to many people. Like one of the 00:17:54.400 --> 00:17:57.160 biggest issue is is is not just, you 00:17:57.160 --> 00:17:59.080 know, tracking decision records, but 00:17:59.080 --> 00:18:01.040 basically feeding context to agents 00:18:01.040 --> 00:18:02.920 because ultimately the decision records 00:18:02.920 --> 00:18:03.760 are 00:18:03.760 --> 00:18:05.640 possibly some of the stuff you want to 00:18:05.640 --> 00:18:07.680 to have agents access to, right? It's 00:18:07.680 --> 00:18:08.760 It's disparate. It's completely 00:18:08.760 --> 00:18:11.120 separate, you know, Confluence, 00:18:11.120 --> 00:18:14.440 Bitbucket, uh sorry, GitHub code, like 00:18:14.440 --> 00:18:15.400 uh it's all over the different 00:18:15.400 --> 00:18:18.000 repositories. Maybe there's some in 00:18:18.000 --> 00:18:22.240 Slack. So, what we're focused on is I 00:18:22.240 --> 00:18:25.080 think one of the biggest issues is also 00:18:25.080 --> 00:18:26.920 the same data is being represented in 00:18:26.920 --> 00:18:30.360 different ways. And and and you know, 00:18:30.360 --> 00:18:32.240 documents run out of out of sync. Like 00:18:32.240 --> 00:18:34.440 the problem you had in the previous gig 00:18:34.440 --> 00:18:36.120 where you say like we have a whole bunch 00:18:36.120 --> 00:18:38.680 of Confluence articles and we want to 00:18:38.680 --> 00:18:41.400 ingest them into like a Markdown 00:18:41.400 --> 00:18:43.400 database managed by agents and agents 00:18:43.400 --> 00:18:45.240 can generate multiple views 00:18:45.240 --> 00:18:46.480 >> but but that same Markdown 00:18:46.480 --> 00:18:48.360 >> And just for the context, we actually 00:18:48.360 --> 00:18:50.960 unleashed unleashed did that migration, 00:18:50.960 --> 00:18:52.520 but the trouble is that the the AI 00:18:52.520 --> 00:18:55.280 agents sort of made the the content in 00:18:55.280 --> 00:18:57.960 Markdown somehow repetitive 00:18:57.960 --> 00:18:59.760 >> like the broken mirror of different 00:18:59.760 --> 00:19:00.280 views. 00:19:00.280 --> 00:19:02.920 >> Yeah, and it created more noise than 00:19:02.920 --> 00:19:04.400 than what we almost had originally. I 00:19:04.400 --> 00:19:05.600 mean 00:19:05.600 --> 00:19:07.880 >> And that's a common problem, right? 00:19:07.880 --> 00:19:09.400 It's a common problem. Even if you use 00:19:09.400 --> 00:19:11.160 back-driven development, one of the 00:19:11.160 --> 00:19:12.520 first issues you're going to face is 00:19:12.520 --> 00:19:14.280 that you you write something down in 00:19:14.280 --> 00:19:16.760 your specify like in your original user 00:19:16.760 --> 00:19:18.440 stories given when then acceptance 00:19:18.440 --> 00:19:20.360 criteria. And then when it comes down 00:19:20.360 --> 00:19:22.160 to, you know, implementation, the 00:19:22.160 --> 00:19:24.480 library, the technical, you know, the 00:19:24.480 --> 00:19:26.160 the API routes, the data model. 00:19:26.160 --> 00:19:27.000 >> Yeah. 00:19:27.000 --> 00:19:29.080 >> And then when you go down all the way to 00:19:29.080 --> 00:19:31.920 the implementation plan, the phases, 00:19:31.920 --> 00:19:33.800 that's the same information. 00:19:33.800 --> 00:19:34.240 >> Yeah. 00:19:34.240 --> 00:19:36.720 >> That points back to like this task needs 00:19:36.720 --> 00:19:39.080 to be implemented because of this this 00:19:39.080 --> 00:19:41.400 task is this because it this acceptance 00:19:41.400 --> 00:19:43.040 criteria is fulfilled. 00:19:43.040 --> 00:19:44.880 >> to that source of truth. Otherwise, 00:19:44.880 --> 00:19:46.760 there's there's an actual link. 00:19:46.760 --> 00:19:48.880 >> So what happens is you get like 20 00:19:48.880 --> 00:19:51.840 markdown files on disk and they end up 00:19:51.840 --> 00:19:54.400 disagreeing. There will be inaccuracies 00:19:54.400 --> 00:19:56.760 because of different terminology used 00:19:56.760 --> 00:20:00.040 across the documents. And so so that's 00:20:00.040 --> 00:20:02.000 where I I then was like, okay, Matt 00:20:02.000 --> 00:20:04.960 Pocock, he he solves it using his 00:20:04.960 --> 00:20:07.320 drill with docs with domain language. So 00:20:07.320 --> 00:20:09.800 you establish a terminology so that the 00:20:09.800 --> 00:20:11.400 agents can only write documents with 00:20:11.400 --> 00:20:12.440 that terminology. 00:20:12.440 --> 00:20:14.840 >> that video that Yeah, establishing a 00:20:14.840 --> 00:20:16.880 terminology is so key, isn't it? Yeah. 00:20:16.880 --> 00:20:18.560 >> Yeah, so so then once you have a 00:20:18.560 --> 00:20:20.240 established a terminology, it also needs 00:20:20.240 --> 00:20:22.080 less tokens. I mean, I like what he's 00:20:22.080 --> 00:20:24.040 saying there cuz everyone wants less 00:20:24.040 --> 00:20:26.200 tokens. And hopefully as a human you 00:20:26.200 --> 00:20:28.280 also can read and understand it faster 00:20:28.280 --> 00:20:30.120 because you have a shared language. That 00:20:30.120 --> 00:20:31.640 should reduce the number of 00:20:31.640 --> 00:20:34.360 inconsistencies across disk and across 00:20:34.360 --> 00:20:37.160 conference and across sources. But then 00:20:37.160 --> 00:20:38.640 you still have the problem that, you 00:20:38.640 --> 00:20:40.880 know, a data model is a data model. It's 00:20:40.880 --> 00:20:43.360 it's it's one set of of it's one set of 00:20:43.360 --> 00:20:46.680 information. And then re you know, 00:20:46.680 --> 00:20:49.760 visualizing the data model as tables or 00:20:49.760 --> 00:20:51.680 or visualizing it in a in a sequence 00:20:51.680 --> 00:20:54.280 diagram explaining the interaction 00:20:54.280 --> 00:20:56.440 between systems and and the impact the 00:20:56.440 --> 00:20:57.760 mutations that happens on the data 00:20:57.760 --> 00:21:00.000 model, it's the same data model. And and 00:21:00.000 --> 00:21:03.080 and those run out of sync. So I don't 00:21:03.080 --> 00:21:05.400 like I talked to a friend and their 00:21:05.400 --> 00:21:08.280 solution to to solve the context problem 00:21:08.280 --> 00:21:10.720 for agents and also Google's OKR which 00:21:10.720 --> 00:21:13.520 is like a a markdown on disk. I think 00:21:13.520 --> 00:21:15.640 that's not the right way. I think you 00:21:15.640 --> 00:21:17.200 just end up with the same problem that 00:21:17.200 --> 00:21:18.920 you had with your confluence uh you 00:21:18.920 --> 00:21:20.160 know, you have a thousand different 00:21:20.160 --> 00:21:21.560 views of the same. Even if you have a 00:21:21.560 --> 00:21:22.920 shared language, the LLM is 00:21:22.920 --> 00:21:24.520 non-probabilistic. Uh sorry, it's 00:21:24.520 --> 00:21:26.480 probabilistic non-deterministic and it 00:21:26.480 --> 00:21:28.240 will just create different you know, 00:21:28.240 --> 00:21:29.760 slightly different 00:21:29.760 --> 00:21:33.320 texts and the data is the data. 00:21:33.320 --> 00:21:34.440 >> Yeah. 00:21:34.440 --> 00:21:36.440 I get that. So, don't don't you think 00:21:36.440 --> 00:21:38.000 that that thing I just showed you, the 00:21:38.000 --> 00:21:39.400 decisions 00:21:39.400 --> 00:21:42.560 which is consumable by I AI agents cuz 00:21:42.560 --> 00:21:45.280 it has LLM's to dot text. I purposely 00:21:45.280 --> 00:21:47.080 >> one view of the data. 00:21:47.080 --> 00:21:49.480 It's just one view of the data. It's 00:21:49.480 --> 00:21:51.640 just an ADR view of the data. There's 00:21:51.640 --> 00:21:53.920 actual data behind it and that's the the 00:21:53.920 --> 00:21:57.200 thing we need to to to to capture. 00:21:57.200 --> 00:21:59.040 >> What do you mean? Like the the data 00:21:59.040 --> 00:22:01.640 structure and like with the detail some 00:22:01.640 --> 00:22:03.200 further details you mean? 00:22:03.200 --> 00:22:05.880 >> you may you may have an uh a decision 00:22:05.880 --> 00:22:07.920 record and 00:22:07.920 --> 00:22:10.840 I don't use them that much but 00:22:10.840 --> 00:22:12.760 it may 00:22:12.760 --> 00:22:17.480 modify a data model. So, the problem 00:22:17.480 --> 00:22:19.000 that that I think a lot of people now 00:22:19.000 --> 00:22:20.280 are trying to solve with what they call 00:22:20.280 --> 00:22:22.840 ontologies is to to try and model this. 00:22:22.840 --> 00:22:25.200 And there's a very good talk from I 00:22:25.200 --> 00:22:27.280 think AI engineer conference from a 00:22:27.280 --> 00:22:29.440 professor who talks about expert systems 00:22:29.440 --> 00:22:31.680 back in the '90s and the concept of 00:22:31.680 --> 00:22:33.840 ontologies and how that applies to 00:22:33.840 --> 00:22:37.000 context management for agents. And and 00:22:37.000 --> 00:22:38.480 they basically when they talk about 00:22:38.480 --> 00:22:40.400 ontologies, they talk about a not a 00:22:40.400 --> 00:22:42.520 relational database or a or a view on 00:22:42.520 --> 00:22:44.160 markdown but actually a graph with 00:22:44.160 --> 00:22:45.720 relationships because a 00:22:45.720 --> 00:22:46.840 >> Oh, yeah. 00:22:46.840 --> 00:22:48.880 I noticed that in my feed maybe 00:22:48.880 --> 00:22:51.360 serendipitously that there's a lot of 00:22:51.360 --> 00:22:54.560 talk about marrying a graph database 00:22:54.560 --> 00:22:56.800 with your harness. 00:22:56.800 --> 00:22:58.360 >> And that's exactly what Beats is, right? 00:22:58.360 --> 00:23:00.640 I mean, if you think about Beats, it's a 00:23:00.640 --> 00:23:02.720 it's a graph. It's a node of graphs. I'm 00:23:02.720 --> 00:23:04.160 sorry, it's a graph of nodes. It's a 00:23:04.160 --> 00:23:06.560 graph of nodes of type issue, of type 00:23:06.560 --> 00:23:08.960 feature, of type epic. And they have 00:23:08.960 --> 00:23:11.920 relationship, uh blocked by um 00:23:11.920 --> 00:23:12.280 >> Sure. 00:23:12.280 --> 00:23:14.920 >> And and and and and basically that's 00:23:14.920 --> 00:23:18.080 what makes Beads so interesting because 00:23:18.080 --> 00:23:20.680 you create a graph of tasks and uh you 00:23:20.680 --> 00:23:22.720 leave comments. That's also another node 00:23:22.720 --> 00:23:23.440 type and 00:23:23.440 --> 00:23:23.960 >> Yeah. 00:23:23.960 --> 00:23:25.200 >> that attaches to it. 00:23:25.200 --> 00:23:27.400 >> You're right. I never actually made that 00:23:27.400 --> 00:23:28.680 connection. 00:23:28.680 --> 00:23:30.480 Well, I I feel like I just want to go go 00:23:30.480 --> 00:23:33.640 back to basics. Like I want I wanted to 00:23:33.640 --> 00:23:36.800 because because I think 00:23:36.800 --> 00:23:39.320 leadership in typical companies won't 00:23:39.320 --> 00:23:40.880 won't uh 00:23:40.880 --> 00:23:43.760 even probably even grok Beads, but they 00:23:43.760 --> 00:23:46.280 might grok a webpage. 00:23:46.280 --> 00:23:48.480 You know, and then and then from from a 00:23:48.480 --> 00:23:51.000 decision maybe maybe you can have a like 00:23:51.000 --> 00:23:53.000 a Bead link, which would have more 00:23:53.000 --> 00:23:55.200 details and you gradually Yeah, 00:23:55.200 --> 00:23:57.440 progressive disclosure but essentially. 00:23:57.440 --> 00:24:00.000 >> And you can see with Beads like the 00:24:00.000 --> 00:24:01.400 Agent is really good at creating the 00:24:01.400 --> 00:24:03.040 task and managing dependencies between 00:24:03.040 --> 00:24:05.200 them, seeing which and doing a query on 00:24:05.200 --> 00:24:07.080 the graph to find out which tasks are 00:24:07.080 --> 00:24:07.560 >> Yeah. 00:24:07.560 --> 00:24:09.440 >> you know, not blocked. But what you can 00:24:09.440 --> 00:24:11.640 also do with Beads is you can generate a 00:24:11.640 --> 00:24:13.600 Kanban board from it. You can show a 00:24:13.600 --> 00:24:16.120 user which status all of the Beads are 00:24:16.120 --> 00:24:16.280 in. 00:24:16.280 --> 00:24:17.840 >> yeah, you can you can, sorry. 00:24:17.840 --> 00:24:19.400 >> Another one that you can do is you can 00:24:19.400 --> 00:24:22.680 show a like if you if you know the Kiro 00:24:22.680 --> 00:24:24.400 and you see the the phases and the 00:24:24.400 --> 00:24:26.400 tasks, you you have a a markdown 00:24:26.400 --> 00:24:28.000 document with little checkboxes when a 00:24:28.000 --> 00:24:30.560 task is done. And and you can That's 00:24:30.560 --> 00:24:32.160 like a tree view because you have like 00:24:32.160 --> 00:24:33.560 phase and then you have all the tasks 00:24:33.560 --> 00:24:35.240 and you have sub children. You can do 00:24:35.240 --> 00:24:36.800 that with Beads as well, right? That's 00:24:36.800 --> 00:24:38.240 basically what I was doing with Pearl, 00:24:38.240 --> 00:24:39.760 the the twee twee. 00:24:39.760 --> 00:24:40.240 >> Yeah, yeah. 00:24:40.240 --> 00:24:42.480 >> I was just I was showing like here's my 00:24:42.480 --> 00:24:44.520 task graph, here's the tree, here's the 00:24:44.520 --> 00:24:46.120 phases, here's what's completed. I can 00:24:46.120 --> 00:24:47.760 see it in both views. And because it's 00:24:47.760 --> 00:24:49.280 one set of data, you can get like 00:24:49.280 --> 00:24:51.240 different views from from from Beads. 00:24:51.240 --> 00:24:53.200 >> Yeah. I guess I I've been using Beads 00:24:53.200 --> 00:24:55.760 myself just internal to my own little 00:24:55.760 --> 00:24:58.560 projects and I love it. And I find that 00:24:58.560 --> 00:25:00.320 useful. I haven't I haven't managed to 00:25:00.320 --> 00:25:02.960 get my my my my team or my colleagues 00:25:02.960 --> 00:25:04.480 adopting that sort of flow. Yeah, I 00:25:04.480 --> 00:25:06.520 mean, have you I mean, you must have 00:25:06.520 --> 00:25:08.720 come into this new this is this your 00:25:08.720 --> 00:25:10.680 this um this company that you now work 00:25:10.680 --> 00:25:11.640 for 00:25:11.640 --> 00:25:14.320 with a ton of ideas. 00:25:14.320 --> 00:25:15.480 Has any 00:25:15.480 --> 00:25:17.280 of any of your ideas been adopted, would 00:25:17.280 --> 00:25:19.400 you say? Like your your adversarial 00:25:19.400 --> 00:25:22.480 reviews, your usage of the adversarial 00:25:22.480 --> 00:25:23.080 reviews. 00:25:23.080 --> 00:25:25.400 >> arrogant to assume that they are not 00:25:25.400 --> 00:25:26.960 doing that already. They were already 00:25:26.960 --> 00:25:28.960 doing that, yeah. 00:25:28.960 --> 00:25:30.640 >> I often find that 00:25:30.640 --> 00:25:32.360 s- like 00:25:32.360 --> 00:25:33.080 the policy they have 00:25:33.080 --> 00:25:34.640 >> have joined a company if they wouldn't 00:25:34.640 --> 00:25:37.240 be if I wouldn't be the dumbest person 00:25:37.240 --> 00:25:40.640 in the room. 00:25:40.640 --> 00:25:43.280 >> Yeah, well, I often find 00:25:43.280 --> 00:25:45.560 that this I've there's things that I 00:25:45.560 --> 00:25:47.400 like to do, but my team doesn't know 00:25:47.400 --> 00:25:49.520 about and it's very and that, you know, 00:25:49.520 --> 00:25:51.960 they're they use Atlassian and for or 00:25:51.960 --> 00:25:53.880 Jira and then you're just like, okay, 00:25:53.880 --> 00:25:54.960 now we 00:25:54.960 --> 00:25:57.240 now we're stuck or 00:25:57.240 --> 00:25:59.080 I want to share my screen with you many 00:25:59.080 --> 00:26:01.600 times before. 00:26:01.600 --> 00:26:02.920 >> Oh, what was that, sorry? 00:26:02.920 --> 00:26:05.920 >> I want to share my screen, but Chrome 00:26:05.920 --> 00:26:08.360 doesn't have the ability to 00:26:08.360 --> 00:26:10.600 to share it. 00:26:10.600 --> 00:26:12.440 Maybe you can share it. I sent you this 00:26:12.440 --> 00:26:14.760 the link. It's a 00:26:14.760 --> 00:26:15.960 >> I have Hold on. 00:26:15.960 --> 00:26:17.760 Are you sharing it with me? WhatsApp, 00:26:17.760 --> 00:26:18.800 hopefully. 00:26:18.800 --> 00:26:21.080 >> No, no, I'll share I share it on this 00:26:21.080 --> 00:26:23.960 chat here and Riverside. 00:26:23.960 --> 00:26:25.000 >> Oh my god, I don't even know where the 00:26:25.000 --> 00:26:26.720 chat window is. 00:26:26.720 --> 00:26:28.360 >> I don't have WhatsApp on this work 00:26:28.360 --> 00:26:29.440 laptop. 00:26:29.440 --> 00:26:32.720 >> Okay, okay, okay. 00:26:32.720 --> 00:26:36.240 I'm going to share the screen. 00:26:36.240 --> 00:26:39.120 Piaz, oh my god. They all look the same, 00:26:39.120 --> 00:26:41.400 these sites. All right. 00:26:41.400 --> 00:26:43.120 Where people in Asia share one 00:26:43.120 --> 00:26:45.320 understanding, yeah, right. 00:26:45.320 --> 00:26:46.760 >> It's what we were discussing, basically, 00:26:46.760 --> 00:26:47.320 right? 00:26:47.320 --> 00:26:49.280 >> Yeah. I mean, this this is this is 00:26:49.280 --> 00:26:50.840 several implementations, like the 00:26:50.840 --> 00:26:55.320 Mozilla CQ CQ is one one that I've CQ 00:26:55.320 --> 00:26:57.000 is one that I would talked about in a 00:26:57.000 --> 00:26:58.520 previous pod. 00:26:58.520 --> 00:27:01.000 Where I was really impressed by 00:27:01.000 --> 00:27:03.920 by Mozilla's approach because they had 00:27:03.920 --> 00:27:07.120 these tool calls like validating the the 00:27:07.120 --> 00:27:08.240 knowledge. 00:27:08.240 --> 00:27:09.560 Uh 00:27:09.560 --> 00:27:11.640 the knowledge. 00:27:11.640 --> 00:27:14.280 I'm not sure what they called them now. 00:27:14.280 --> 00:27:15.480 Uh and I thought that was quite 00:27:15.480 --> 00:27:16.640 powerful. 00:27:16.640 --> 00:27:17.600 But 00:27:17.600 --> 00:27:19.280 what what I'm trying to say here Vincent 00:27:19.280 --> 00:27:20.440 is that 00:27:20.440 --> 00:27:22.760 there are lots of cool projects, 00:27:22.760 --> 00:27:25.040 agent-friendly projects. 00:27:25.040 --> 00:27:27.480 Though I'm looking for a way where we 00:27:27.480 --> 00:27:28.600 start 00:27:28.600 --> 00:27:30.720 essentially with an idea like that an 00:27:30.720 --> 00:27:33.240 old idea like the ADR and then 00:27:33.240 --> 00:27:35.080 progressively build on that using I 00:27:35.080 --> 00:27:36.640 don't know something like this or beads 00:27:36.640 --> 00:27:40.320 or whatever a graph database. But it 00:27:40.320 --> 00:27:42.240 But you must understand like I'm super 00:27:42.240 --> 00:27:45.200 frustrated when when clients or or 00:27:45.200 --> 00:27:47.120 different companies they just don't have 00:27:47.120 --> 00:27:48.800 a central point. This I'm just looking 00:27:48.800 --> 00:27:51.360 for a source of truth at this point. 00:27:51.360 --> 00:27:53.680 And that and that's that's step one in 00:27:53.680 --> 00:27:55.720 my opinion get a source of truth and 00:27:55.720 --> 00:28:00.880 then the rest of it will follow. 00:28:00.880 --> 00:28:04.120 >> I'm trying to see if CQ is using graph 00:28:04.120 --> 00:28:06.680 relationship. Well, you know beats is 00:28:06.680 --> 00:28:08.320 also in 00:28:08.320 --> 00:28:10.200 built using SQL light, right? It 00:28:10.200 --> 00:28:12.720 represents nodes within the graph using 00:28:12.720 --> 00:28:15.000 tables and the relationship tables and 00:28:15.000 --> 00:28:16.280 so on. So 00:28:16.280 --> 00:28:18.840 you don't need a graph database but 00:28:18.840 --> 00:28:20.840 I never use graph database before. But 00:28:20.840 --> 00:28:22.440 in terms of like what you just asked me 00:28:22.440 --> 00:28:26.200 of of like being frustrated. I mean I 00:28:26.200 --> 00:28:28.600 think beats 00:28:28.600 --> 00:28:30.880 it's I I got rid of it, right? I I 00:28:30.880 --> 00:28:34.080 replaced it with my own implementation. 00:28:34.080 --> 00:28:35.160 But 00:28:35.160 --> 00:28:37.600 I mean the idea I think under the hood 00:28:37.600 --> 00:28:40.120 the idea of graphs and nodes 00:28:40.120 --> 00:28:41.440 the problem is they're very very 00:28:41.440 --> 00:28:43.800 specific like Jira has it, right? I mean 00:28:43.800 --> 00:28:46.080 beats all look like me like Jira, right? 00:28:46.080 --> 00:28:47.760 You have issues, you have child issues, 00:28:47.760 --> 00:28:49.000 you have epics, you have features. 00:28:49.000 --> 00:28:50.720 There's all different nodes of different 00:28:50.720 --> 00:28:52.440 of of different types and relationships 00:28:52.440 --> 00:28:53.920 that can they they can have with each 00:28:53.920 --> 00:28:56.840 other. So, I don't think you need to 00:28:56.840 --> 00:28:58.960 move out of those existing platforms. 00:28:58.960 --> 00:29:00.960 Like when I I built Spec Ledger, I was 00:29:00.960 --> 00:29:02.720 like, I hate Jira, I hate Confluence. I 00:29:02.720 --> 00:29:04.200 think those systems are horrible. All 00:29:04.200 --> 00:29:06.520 you need is markdown on disk. And then I 00:29:06.520 --> 00:29:08.720 learned like the broken mirror issue of 00:29:08.720 --> 00:29:11.320 everything is like running out of out of 00:29:11.320 --> 00:29:13.960 sync and it's not easy to manage. I 00:29:13.960 --> 00:29:16.040 think this I this approach of building 00:29:16.040 --> 00:29:19.040 ontologies for companies to build graph 00:29:19.040 --> 00:29:22.160 data like graph relationships of how 00:29:22.160 --> 00:29:24.080 things link together and then building 00:29:24.080 --> 00:29:26.640 on top of that, that's very interesting, 00:29:26.640 --> 00:29:28.760 but it still requires you to move out of 00:29:28.760 --> 00:29:31.000 those existing systems. And I think the 00:29:31.000 --> 00:29:33.320 focus of a lot of the like even Cloud 00:29:33.320 --> 00:29:35.160 Tag and and you know, the way that 00:29:35.160 --> 00:29:37.720 people are building is to to land those 00:29:37.720 --> 00:29:39.960 AI systems in the tools that humans used 00:29:39.960 --> 00:29:41.520 used to use today. 00:29:41.520 --> 00:29:42.720 >> Yeah. 00:29:42.720 --> 00:29:44.480 >> And I think that's that's the realistic 00:29:44.480 --> 00:29:45.880 approach 00:29:45.880 --> 00:29:47.960 is to get AI to live where the humans 00:29:47.960 --> 00:29:48.480 live. 00:29:48.480 --> 00:29:49.000 >> Yeah. 00:29:49.000 --> 00:29:53.760 >> Yeah, I mean that's assuming like 00:29:53.760 --> 00:30:00.240 Well, like 00:30:00.240 --> 00:30:03.120 I I've I personally noticed that in some 00:30:03.120 --> 00:30:05.560 clients that even decisions are not 00:30:05.560 --> 00:30:07.840 recorded properly in Slack cuz Slack is 00:30:07.840 --> 00:30:11.080 more of like a informal space or they 00:30:11.080 --> 00:30:13.040 they're rather jump on a Zoom call or 00:30:13.040 --> 00:30:15.320 whatever you to explain 00:30:15.320 --> 00:30:16.520 what the 00:30:16.520 --> 00:30:18.840 what what the decision is. 00:30:18.840 --> 00:30:21.280 I mean, I guess I guess it's maybe it's 00:30:21.280 --> 00:30:24.120 maybe it's just a symptom 00:30:24.120 --> 00:30:25.440 that many many of the clients that I 00:30:25.440 --> 00:30:28.680 work for are not like, you know, 00:30:28.680 --> 00:30:31.560 very mature software company. So, 00:30:31.560 --> 00:30:34.440 so leadership in a sense is not used to 00:30:34.440 --> 00:30:36.400 cuz like like what do you what do we do 00:30:36.400 --> 00:30:38.000 as a software engineers? We we 00:30:38.000 --> 00:30:40.320 essentially codify 00:30:40.320 --> 00:30:42.400 the business requirements. 00:30:42.400 --> 00:30:43.920 But sometimes 00:30:43.920 --> 00:30:46.080 and and that's the job. But like and 00:30:46.080 --> 00:30:47.920 maybe we should take responsibility for 00:30:47.920 --> 00:30:50.240 just writing down these decisions. 00:30:50.240 --> 00:30:51.320 Though what I'm trying to say is that 00:30:51.320 --> 00:30:53.440 many many times that like 00:30:53.440 --> 00:30:55.360 even Slack doesn't have the context 00:30:55.360 --> 00:30:57.960 because it's in it's in it's in some 00:30:57.960 --> 00:31:00.520 executive's head or something like that. 00:31:00.520 --> 00:31:03.120 Like just just like I like what 00:31:03.120 --> 00:31:04.560 just the fact 00:31:04.560 --> 00:31:06.720 that some 00:31:06.720 --> 00:31:08.440 I'm trying to think of the name of the 00:31:08.440 --> 00:31:12.040 guy there was a a a some executive a guy 00:31:12.040 --> 00:31:13.720 with a ponytail. I can't remember his 00:31:13.720 --> 00:31:16.640 name now. He used to just write 00:31:16.640 --> 00:31:20.680 blogs and and I think of course 00:31:20.680 --> 00:31:22.200 >> Jeff No. 00:31:22.200 --> 00:31:24.760 The and of course of course like uh 00:31:24.760 --> 00:31:27.840 >> I think Jeff Bezos is is also good a 00:31:27.840 --> 00:31:30.360 good example. Of course his blogs are 00:31:30.360 --> 00:31:32.960 more internal, but he just releases 00:31:32.960 --> 00:31:35.400 these memos and then it's clear to the 00:31:35.400 --> 00:31:36.880 rest of organization what what we're 00:31:36.880 --> 00:31:38.480 trying to achieve here. 00:31:38.480 --> 00:31:40.960 The memo is the memorandum. 00:31:40.960 --> 00:31:44.000 And it's clear. I just That's what I'm 00:31:44.000 --> 00:31:46.960 I I miss a lot of the time. Just a blog. 00:31:46.960 --> 00:31:48.760 You know, we talk about Jira, we talk 00:31:48.760 --> 00:31:51.480 about all these graph databases, but 00:31:51.480 --> 00:31:55.080 like just a blog would be so refreshing 00:31:55.080 --> 00:31:57.840 in this darn age. 00:31:57.840 --> 00:31:59.320 And that's where I'm at. That's that's 00:31:59.320 --> 00:32:02.320 my frustration. 00:32:02.320 --> 00:32:05.000 >> So you want your CTO blog. 00:32:05.000 --> 00:32:06.280 >> Essentially. 00:32:06.280 --> 00:32:07.280 That would be so 00:32:07.280 --> 00:32:08.520 That would be so awesome. 00:32:08.520 --> 00:32:12.160 >> I did have one CTO at some point 00:32:12.160 --> 00:32:14.600 that had a blog. Yeah, at at GoodNotes 00:32:14.600 --> 00:32:17.800 the the the the the founder he wrote 00:32:17.800 --> 00:32:20.600 blogs about like what he learned, how he 00:32:20.600 --> 00:32:22.680 saw the company evolve, how he saw the 00:32:22.680 --> 00:32:25.280 ecosystem evolve, what he he believes 00:32:25.280 --> 00:32:28.280 was like coming next, and he he he 00:32:28.280 --> 00:32:30.720 shared them, and yeah. 00:32:30.720 --> 00:32:32.600 They were insightful. He's a smart He's 00:32:32.600 --> 00:32:33.480 a smart person. 00:32:33.480 --> 00:32:35.720 >> that's the stuff I really miss. Cuz like 00:32:35.720 --> 00:32:37.920 when people do their little demos on 00:32:37.920 --> 00:32:39.840 Zoom and there's Zoom recordings and no 00:32:39.840 --> 00:32:42.000 one can ever find the Zoom recording cuz 00:32:42.000 --> 00:32:44.800 for some bizarre reason Zoom makes it 00:32:44.800 --> 00:32:47.320 impossible to find Zoom recordings. This 00:32:47.320 --> 00:32:50.120 is There's so much context that's lost. 00:32:50.120 --> 00:32:52.400 And I hope And if we can just start with 00:32:52.400 --> 00:32:53.920 a single point of truth, a blog, and 00:32:53.920 --> 00:32:55.840 then progressively just goes, I'll be so 00:32:55.840 --> 00:32:58.600 much more happier and less confused all 00:32:58.600 --> 00:33:00.360 the time. And and And it's not so much 00:33:00.360 --> 00:33:02.640 me being confused. 00:33:02.640 --> 00:33:04.080 It's like when you work with AI, when 00:33:04.080 --> 00:33:05.760 you work with agents, and you notice 00:33:05.760 --> 00:33:07.200 that it doesn't have the context, you're 00:33:07.200 --> 00:33:09.440 like you're frustrated. And like, how do 00:33:09.440 --> 00:33:10.960 I get the context here? I have to copy 00:33:10.960 --> 00:33:13.040 it here. I have to I have to summarize 00:33:13.040 --> 00:33:14.680 this video 00:33:14.680 --> 00:33:17.040 for it to know what we're talking about 00:33:17.040 --> 00:33:18.720 for this particular 00:33:18.720 --> 00:33:21.280 activity or job that we're trying to do. 00:33:21.280 --> 00:33:22.920 And 00:33:22.920 --> 00:33:24.800 you know, context engineering from for 00:33:24.800 --> 00:33:27.360 me is almost 00:33:27.360 --> 00:33:28.600 like 00:33:28.600 --> 00:33:30.000 the frustrating experience of just 00:33:30.000 --> 00:33:31.720 gathering bits and pieces from 00:33:31.720 --> 00:33:34.680 colleagues and what have you to try work 00:33:34.680 --> 00:33:38.400 out what the decision was. 00:33:38.400 --> 00:33:39.720 And then And then when I've got it all 00:33:39.720 --> 00:33:41.680 there, I'm not even sure it's right 00:33:41.680 --> 00:33:43.320 because people have said different 00:33:43.320 --> 00:33:45.680 things and what have you and things have 00:33:45.680 --> 00:33:47.800 changed since then. 00:33:47.800 --> 00:33:49.880 Yeah, anyway, that's just That's my 00:33:49.880 --> 00:33:51.880 rant. Yeah. 00:33:51.880 --> 00:33:52.320 Awesome. 00:33:52.320 --> 00:33:55.600 >> Yeah, I mean, I think with agents and 00:33:55.600 --> 00:33:57.240 the amount of, you know, these large 00:33:57.240 --> 00:33:59.200 language models, the amount of text that 00:33:59.200 --> 00:34:00.840 they produce and the inconsistency 00:34:00.840 --> 00:34:03.840 between the text and and every human now 00:34:03.840 --> 00:34:05.880 being hooked up to one of those one or 00:34:05.880 --> 00:34:07.840 more of those generating more and more 00:34:07.840 --> 00:34:10.399 text, I don't think the the solution is 00:34:10.399 --> 00:34:15.480 going to be solved easily by by large 00:34:15.480 --> 00:34:17.159 text document. Apparently, when they 00:34:17.159 --> 00:34:20.360 look at usage of these LLM, a large part 00:34:20.360 --> 00:34:23.840 of it is right like a very short status 00:34:23.840 --> 00:34:25.919 update that gets converted into like a 00:34:25.919 --> 00:34:28.200 massive long blob from like, "Hey, 00:34:28.200 --> 00:34:30.399 Cloud, can we do a quick status update?" 00:34:30.399 --> 00:34:32.639 And and creates like paragraphs of text 00:34:32.639 --> 00:34:34.800 and then somebody else goes simplify 00:34:34.800 --> 00:34:37.520 this. Hey Cloud, simplify this. Like the 00:34:37.520 --> 00:34:39.520 LLMs are being used to to generate 00:34:39.520 --> 00:34:41.159 massive amount of text and then they are 00:34:41.159 --> 00:34:43.320 used to to summarize this amount of text 00:34:43.320 --> 00:34:46.200 again. Where like if you want to There 00:34:46.200 --> 00:34:48.399 was one of the insights of of of a 00:34:48.399 --> 00:34:50.480 senior like leader in the company that 00:34:50.480 --> 00:34:52.679 I'm at. He's like I looked at the 00:34:52.679 --> 00:34:55.480 statistics of token usage and if we want 00:34:55.480 --> 00:34:57.120 to save cost because it's way through 00:34:57.120 --> 00:35:00.120 the roof, let's start by, you know, stop 00:35:00.120 --> 00:35:02.600 using LLMs to generate 00:35:02.600 --> 00:35:05.400 status updates and and and because 00:35:05.400 --> 00:35:06.880 nobody understands them. 00:35:06.880 --> 00:35:08.200 >> Yeah. 00:35:08.200 --> 00:35:10.040 Like one thing that I'm thinking of 00:35:10.040 --> 00:35:12.680 mandating in my I mean I mean I have 00:35:12.680 --> 00:35:14.920 sort of mandated it in my team. 00:35:14.920 --> 00:35:17.120 We don't allow people to write with into 00:35:17.120 --> 00:35:19.760 confluence documents with a with agents. 00:35:19.760 --> 00:35:22.000 If you're going to write a document, 00:35:22.000 --> 00:35:24.400 write it by hand. 00:35:24.400 --> 00:35:26.200 Because because often 00:35:26.200 --> 00:35:27.320 >> How can you enforce that? 00:35:27.320 --> 00:35:28.680 >> How do you enforce that? Uh well, you 00:35:28.680 --> 00:35:29.800 can 00:35:29.800 --> 00:35:32.320 Well, you could theoretically disable 00:35:32.320 --> 00:35:34.160 the right tool. 00:35:34.160 --> 00:35:36.520 Um but unfortunately, the whole the rest 00:35:36.520 --> 00:35:38.560 of the organization doesn't share my my 00:35:38.560 --> 00:35:40.800 team's view or the 00:35:40.800 --> 00:35:43.040 least the views I project on my team. 00:35:43.040 --> 00:35:44.480 >> Oh my god, I'm late for a meeting. I'm 00:35:44.480 --> 00:35:45.680 sorry. 00:35:45.680 --> 00:35:47.640 >> You you can just you can have some text 00:35:47.640 --> 00:35:49.680 in the in the conference document like 00:35:49.680 --> 00:35:51.160 agents do not write into this document. 00:35:51.160 --> 00:35:54.200 That that also works. Sorry. Oh man, 00:35:54.200 --> 00:35:56.560 there's some 00:35:56.560 --> 00:35:58.040 >> Oh gosh. Do you have any time more 00:35:58.040 --> 00:35:59.480 today? There's other things I wanted to 00:35:59.480 --> 00:36:01.000 pick up on. Otherwise, I guess this is 00:36:01.000 --> 00:36:02.360 the pod. 00:36:02.360 --> 00:36:05.720 Episode What was the last one? 00:36:05.720 --> 00:36:07.800 >> I'm sorry. I 00:36:07.800 --> 00:36:09.280 I didn't give too much updates on but 00:36:09.280 --> 00:36:11.040 but I have another call and a meeting. 00:36:11.040 --> 00:36:13.360 >> Okay. No worries. That that was episode 00:36:13.360 --> 00:36:17.440 39. At least we we we kind of got back 00:36:17.440 --> 00:36:19.200 together again. 00:36:19.200 --> 00:36:20.320 >> Okay. [laughter] 00:36:20.320 --> 00:36:21.280 >> Okay. See you, man. 00:36:21.280 --> 00:36:23.680 >> start on the hour, then it can be it can 00:36:23.680 --> 00:36:25.000 be like 50 full minutes. 00:36:25.000 --> 00:36:27.360 >> Yeah, yeah. Sorry. I was but I'm on 00:36:27.360 --> 00:36:29.000 holiday, so I was a bit slow out of bed. 00:36:29.000 --> 00:36:30.120 Okay, see you. 00:36:30.120 --> 00:36:30.760 >> Okay, 00:36:30.760 --> 00:36:33.120 see you.