Ruby Security Breaches, AI Agents, and Coverage Standards の全文書き起こし

right. 00:02:00.000 --> 00:02:07.920 So as a reminder, this show we're gonna cover the main events as determined by our AIs. 00:02:08.479 --> 00:02:14.240 And so I think the the thing that we get out of this is we've got fresh eyes. 00:02:14.319 --> 00:02:21.039 We'll have fresh takes on all of the stuff that comes up here because we have done virtually no preparation for the show. 00:02:21.280 --> 00:02:31.520 And on the plus side, then you're going to get my honest take, you're gonna get Valentino's honest take, you're not gonna get our take plus AI, you know, and whatever we've done to research it. 00:02:31.759 --> 00:02:32.639 So it's been exciting. 00:02:32.800 --> 00:02:37.120 Last time we did it was really fun, and I think this one is gonna be even better. 00:02:37.520 --> 00:02:40.960 There's a lot of great ones in here, and a lot is happening as usual. 00:02:41.120 --> 00:02:42.479 Yeah, I'm excited about it. 00:02:42.719 --> 00:02:44.800 You know, the fun never stops here. 00:02:45.840 --> 00:02:52.560 We also just found out before the show that neither one of us has any sense of the time in between bells. 00:02:52.800 --> 00:02:56.000 So as soon as the bell goes off, we're supposed to move to the next topic. 00:02:56.159 --> 00:02:57.759 It'll just be a shock to all of us. 00:02:57.840 --> 00:02:58.159 Yeah. 00:02:58.319 --> 00:03:02.159 And maybe we'll work that out, or maybe it's just a fun little part of the show. 00:03:02.400 --> 00:03:05.759 You know, I'm gonna have to get a little like one of those desktop gongs. 00:03:05.919 --> 00:03:08.240 We'll get a whole big ruckus of noise. 00:03:08.400 --> 00:03:08.960 There we go. 00:03:09.439 --> 00:03:09.840 There it is. 00:03:10.000 --> 00:03:11.120 I knew we were getting close. 00:03:11.360 --> 00:03:12.400 Right on cue. 00:03:13.039 --> 00:03:15.120 All right, let's dig into the first one. 00:03:15.439 --> 00:03:17.919 So, what a time to be alive. 00:03:18.159 --> 00:03:22.159 The title of Aaron Patterson's post on this topic. 00:03:22.479 --> 00:03:23.919 And true, what a time. 00:03:24.319 --> 00:03:35.120 As announced in many channels, but in the latest Ruby AI newsletter, OpenAI's own AI agents secretly attacked RubyGems back in May. 00:03:35.360 --> 00:03:41.520 And for those that don't know, OpenAI is a sponsor of the Ruby security team's efforts. 00:03:41.680 --> 00:03:42.319 Yep. 00:03:42.800 --> 00:03:50.719 And apparently Aaron Patterson couldn't have put it better in this write-up, which is actually very technical, but also has some amusing anecdotes. 00:03:50.879 --> 00:03:51.120 Yeah. 00:03:51.280 --> 00:04:00.000 The first one being that they knew about some of the vulnerabilities ahead of time, and they exploited them anyway, or their agents did. 00:04:00.319 --> 00:04:00.560 Right. 00:04:01.039 --> 00:04:05.360 I think because of the fact that maybe because of them, and you know, never stopped them. 00:04:05.520 --> 00:04:05.840 I don't know. 00:04:06.000 --> 00:04:07.280 What's your take on this, Joe? 00:04:07.680 --> 00:04:10.400 So this is the one of the stories we're gonna cover today. 00:04:10.479 --> 00:04:13.759 This is the one that I had some advanced notice on just because this was everywhere, right? 00:04:13.840 --> 00:04:16.639 And it was all over Deaf Method, people were talking about it, and I agree. 00:04:16.800 --> 00:04:24.639 Aaron Patterson's write-up, it's both technical and concise, which you don't always get, and it's got a couple of good anecdotes, so I would recommend checking that out. 00:04:25.040 --> 00:04:27.040 I like Simon Willison's take here. 00:04:27.199 --> 00:04:36.959 Either OpenAI reviewed its logs after the hugging face incident because the exact same security breach is what happened here, or the exact same mechanism, I should say. 00:04:37.120 --> 00:04:46.480 So either they'd reviewed the logs and still could not connect themselves to the RubyGems attack, or OpenAI made the connection and just decided not to tell the RubyGems team. 00:04:46.560 --> 00:04:48.240 So both of those are bad. 00:04:48.480 --> 00:04:50.160 We don't even know which one. 00:04:50.560 --> 00:04:57.759 My take on it, you know, I saw this question: should AI labs have to disclose it when their own agents cause a security incident? 00:04:58.000 --> 00:05:00.480 My immediate thought was there's no regulation. 00:05:00.879 --> 00:05:02.079 This is the Wild West. 00:05:02.560 --> 00:05:04.560 I don't know if they're supposed to do it. 00:05:04.639 --> 00:05:12.240 I mean, I know they're not going to do it unless they get pressure from the community or there's some kind of regulation. 00:05:12.399 --> 00:05:15.199 You know, I mean, that's eventually what happened with the hugging phase. 00:05:15.279 --> 00:05:21.199 You know, hugging phase came out with its entire write-up of the security incident before OpenAI said anything. 00:05:21.439 --> 00:05:23.920 And it seems like the same kind of thing is happening here. 00:05:24.319 --> 00:05:37.519 I don't know that I agree with you on the are they responsible or not, or are there regulations like security companies have had automated means of exploring exploitations for a long time. 00:05:37.759 --> 00:05:40.720 They are held to a legal standard, right? 00:05:40.959 --> 00:05:50.560 Of well, if you're running penetration tests in a white hat capacity, you are responsible for those exploits and being in control of them. 00:05:50.800 --> 00:05:59.360 And from a compliance standard, like if you're doing this for like HIPAA compliance, trying to disclose something for a compliance reason, you have to disclose it as part of your agreement. 00:05:59.600 --> 00:06:05.279 OpenAI has these HIPAA agreements and BAAs across many companies. 00:06:05.759 --> 00:06:06.160 That's true. 00:06:06.240 --> 00:06:06.560 Yeah, yeah. 00:06:07.519 --> 00:06:10.160 Are they exploiting those companies? 00:06:10.399 --> 00:06:11.279 They must know. 00:06:11.439 -->

00:06:15.199 And if they're not disclosing it, that is a breach of their contract. 00:06:16.079 --> 00:06:16.959 I agree with you there. 00:06:17.199 --> 00:06:22.240 I don't really understand what the question is around or why there's even a need for regulation. 00:06:22.319 --> 00:06:29.360 It's like they're an organization that has tooling that is making mistakes and they're not disclosing it. 00:06:29.439 --> 00:06:30.480 That's kind of illegal. 00:06:30.800 --> 00:06:31.199 Yeah. 00:06:31.439 --> 00:06:41.600 Well, I disagree with you on that, you know, from the perspective of Ruby Central, which is managing and trying to secure RubyGems. 00:06:41.680 --> 00:06:46.560 I mean, hilariously, with OpenAI's help, you know, what is their recourse? 00:06:46.800 --> 00:06:48.079 Will they seek to use it? 00:06:48.240 --> 00:06:52.319 I should be clear here, it's not like OpenAI is the only company that's helping RubyGems. 00:06:52.399 --> 00:06:55.519 You know, they get help from Anthropic as well, they get help from some of the other major players. 00:06:55.920 --> 00:06:58.879 And when I say help, I mean mostly money and free tokens. 00:06:59.199 --> 00:07:02.319 But I don't really know what recourse RubyGems has. 00:07:02.399 --> 00:07:09.600 I mean, this was a huge deal when it happened back in May, and we're just now figuring out that oh, it was coming from inside the house. 00:07:09.920 --> 00:07:13.920 It's hard to say what the right reaction is to have. 00:07:14.319 --> 00:07:18.160 Is the money that we gotta switch? 00:07:18.879 --> 00:07:22.000 Actually, I think was that me or is that it could be either. 00:07:22.240 --> 00:07:23.199 I think that was mine. 00:07:23.279 --> 00:07:23.519 Yeah. 00:07:23.759 --> 00:07:34.240 But if like the money that OpenAI is contributing to you know the Ruby Gems team, is that sufficient to cover whatever costs it took to like remediate this? 00:07:34.480 --> 00:07:36.560 That I wonder if that's the case. 00:07:36.720 --> 00:07:39.519 Does it matter if all the precautions were taken? 00:07:39.680 --> 00:07:43.120 I guess it matters a little because they didn't notify anybody that they were running it, right? 00:07:43.279 --> 00:07:46.160 So it matters in that way. 00:07:46.480 --> 00:07:50.480 It's hard because like this affects more than just RubyGems, right? 00:07:50.560 --> 00:07:52.319 It affects everybody using RubyGems. 00:07:52.480 --> 00:07:54.000 Yeah, absolutely, yeah, it does. 00:07:54.240 --> 00:08:00.240 And so if Marty Hott is at Rails World next week, I'm gonna pull him aside. 00:08:00.319 --> 00:08:01.600 I'm gonna get a quote for the show. 00:08:01.680 --> 00:08:03.120 I want to know what he has to say. 00:08:03.360 --> 00:08:04.800 Because he's not gonna be subtle about it. 00:08:05.439 --> 00:08:06.399 He can't be happy about it either. 00:08:06.639 --> 00:08:07.680 He's definitely not happy. 00:08:08.160 --> 00:08:11.519 I want to find out the level of his discontent right now. 00:08:11.759 --> 00:08:12.079 Right. 00:08:12.240 --> 00:08:14.879 Yeah, I mean, I imagine it's still ongoing. 00:08:15.519 --> 00:08:17.439 The remediation, I'm sure, yeah. 00:08:17.759 --> 00:08:20.959 Majej, I'm sorry, I'm mispronouncing his name, Mensfeld. 00:08:21.120 --> 00:08:25.040 On May 12th, we're dealing with a major malicious attack on RubyGems right now. 00:08:25.199 --> 00:08:26.399 Sign-ups or pause. 00:08:26.560 --> 00:08:32.000 These guys were in the trenches trying to fight OpenAI's own agents. 00:08:32.879 --> 00:08:37.679 I hope that the outcome is that OpenAI gives some free credits to RubyGems at least. 00:08:37.840 --> 00:08:38.960 And maybe some automated credits. 00:08:45.679 --> 00:08:48.320 Let's just battle the hackers with anti-hackers. 00:08:48.639 --> 00:08:49.519 Yeah, right. 00:08:50.080 --> 00:08:51.440 Okay, next question up. 00:08:51.519 --> 00:08:52.879 Uh, why don't you take it, Joe? 00:08:53.279 --> 00:08:53.519 Sure. 00:08:53.600 --> 00:08:59.519 So GPT-6 Astra tops the quote agents on Rails stage two. 00:09:00.320 --> 00:09:07.039 So there's a benchmark team that's been evaluating all the new models as they come out. 00:09:07.360 --> 00:09:21.679 Most recently, they threw 37 Signals fizzy Kanban tool to uh 10 different models and asked it to run its agents against you know real features, quote unquote real features. 00:09:22.000 --> 00:09:25.919 And so GPT Astra won, 35% solved in nine minutes. 00:09:26.080 --> 00:09:27.440 I'm not exactly sure what that means. 00:09:27.519 --> 00:09:36.000 Uh Claude Fable 5.1, play second, 30% solved, but that costs four times as much as GPT-6 Astra. 00:09:36.480 --> 00:09:38.480 One test model scored zero of 60. 00:09:38.559 --> 00:09:40.320 I'd be curious to know what that is. 00:09:40.639 --> 00:09:50.960 So I guess the question is, does 35% efficacy on real rails work mean that these agents are close or nowhere near ready? 00:09:51.120 --> 00:09:55.759 I assume nowhere near ready to take on end-to-end development. 00:09:56.399 --> 00:10:03.919 V, you I think already are in the business of having agents do end-to-end development, and so are many other businesses. 00:10:04.240 --> 00:10:06.320 I don't know if it's most, but many other businesses. 00:10:06.399 --> 00:10:10.159 So what do you think about agents' readiness to take this on? 00:10:10.559 --> 00:10:12.799 I guess it depends what you mean by agent. 00:10:13.120 --> 00:10:19.360 You know, there's so many harnesses out there, and spinning up OpenAI and running Ostra. 00:10:19.519 --> 00:10:25.039 If you're opening up a fresh Clawed instance and then running Ostra on it, is that ready? 00:10:25.279 --> 00:10:25.600 No. 00:10:26.000 --> 00:10:28.399 And I say that because it depends, right? 00:10:28.480 --> 00:10:38.879 Like everybody has a different setup, everybody has a different taste and preferences to the choices they've made architecturally throughout their Rails application or Ruby application.

00:10:39.039 --> 00:10:43.440 Everyone's unique, and these tests aren't covering for all those cases. 00:10:43.600 --> 00:10:50.480 They're testing this one specific app and that setup, which is, I would say, a little bit unique, to be honest. 00:10:50.960 --> 00:10:52.399 The Kanban tool, yeah. 00:10:52.639 --> 00:10:52.960 Yeah. 00:10:53.200 --> 00:11:03.840 And so 37 Signals, they have a great workflow for how they structure their Rails apps, but it's by no means like the norm that I've seen across the industry. 00:11:04.159 --> 00:11:06.639 And so I worry about that as a benchmark. 00:11:06.960 --> 00:11:11.840 But in addition to that, there's so many checks you need in place to validate. 00:11:12.000 --> 00:11:20.559 If you have no verifiers or verification in place in all of those checks and balances, you're not really going to get the end-to-end performance that's reported. 00:11:20.960 --> 00:11:28.080 And I imagine that hopefully, if you go look at the Ruby AI newsletter, there are a ton of tools you can use today. 00:11:28.639 --> 00:11:35.039 Stanlow, the IRB maintainer, has a ton of great ones for Ruby that you should take a look at. 00:11:35.360 --> 00:11:42.080 Just skills that you can add to improve your workflow and get better benchmarks out of it natively. 00:11:42.399 --> 00:11:47.279 But yeah, I think we're missing a lot of variation to claim any victory here. 00:11:47.759 --> 00:11:49.120 Yeah, I agree with you. 00:11:49.200 --> 00:11:59.360 I mean, I so I think that this benchmarking is worthwhile if only to see the numbers go up over time, which I assume is what's happening. 00:11:59.759 --> 00:12:09.440 We're taking this one slice at a time, and you know, at some point several months down the road, there'll be a GPT-7, and maybe that'll get 42%, and I guess that means it's better. 00:12:09.679 --> 00:12:19.279 But I totally agree with you that at least today, nobody's relying on a model alone, and everybody seems pretty okay with that. 00:12:19.519 --> 00:12:24.159 We all know that the harness and all the context that we have to give it, that's part of it. 00:12:24.240 --> 00:12:29.840 And this is maybe a digression, but that's the engineering part that that's left for us. 00:12:30.159 --> 00:12:32.879 So for me, I always feel like, well, that's great. 00:12:32.960 --> 00:12:39.519 That's where I can be creative and discerning and look at you know how things are actually working and make some improvements and iterate. 00:12:39.840 --> 00:12:42.559 So I don't even know if I want it to get that much better. 00:12:42.879 --> 00:12:44.960 It doesn't really matter what I want. 00:12:45.200 --> 00:12:54.320 But this is just an aside that I like the fact that, hey, to get anything done today, you really need that combination. 00:12:54.399 --> 00:13:01.919 You need the tools, you need the people that are creating tools, harnesses, and and skills to make these agents really effective. 00:13:02.240 --> 00:13:02.879 Absolutely. 00:13:03.039 --> 00:13:08.399 And I mean, really great work for that Evil Martians is doing with all of these benchmarks. 00:13:08.559 --> 00:13:12.320 Keep it up, we we need more, and it's really the only way to know for sure. 00:13:12.559 --> 00:13:14.559 Yeah, you know, if it's doing what we expect. 00:13:14.720 --> 00:13:17.200 And like you mentioned, you know, this is the engineering part. 00:13:17.279 --> 00:13:18.399 So yeah, it is. 00:13:18.559 --> 00:13:22.720 And interesting that you say that because Evil Martians is on the next story. 00:13:23.039 --> 00:13:24.639 They're doing exactly that, right? 00:13:24.720 --> 00:13:26.000 They're shipping with agents. 00:13:26.080 --> 00:13:27.200 What do you tell us about that? 00:13:27.440 --> 00:13:35.039 Yeah, so Evil Martians has shipped Rails hyperdrive, so agents stop guessing at your app. 00:13:35.279 --> 00:13:37.360 So they released it just yesterday. 00:13:37.600 --> 00:13:46.240 It's an open source Rails engine that gives AI coding agents live introspection into a running Rails app instead of just letting them improvise. 00:13:46.799 --> 00:13:59.120 And they say uh that they used it on their own data, leading models that reach for the Rails API the task turns on in only 41% of the runs, and the rest of the time it writes its own version. 00:13:59.600 --> 00:14:02.240 So that's kind of an interesting uh anecdote from that. 00:14:02.639 --> 00:14:11.919 It ships with eight MCP tools for live routes, schemas, logs, read-only SQL that it can run on the booted app. 00:14:12.399 --> 00:14:17.200 And you know, their overall bet with this is does it ship agent skills? 00:14:18.080 --> 00:14:22.399 Does it have docs on the checklist every Rails developer runs before adding a gem? 00:14:23.200 --> 00:14:24.399 What's your take on this? 00:14:24.879 --> 00:14:32.559 Yeah, so I we get a little distracted by okay, it's gonna ship with agent skills, and is that going to be the norm? 00:14:32.720 --> 00:14:34.960 And I, you know, I think the answer in the short term is yes. 00:14:35.440 --> 00:14:39.360 I don't know what the long-term life is for agent skills in general. 00:14:39.519 --> 00:14:40.480 I have no idea. 00:14:40.720 --> 00:14:45.440 Um I think just zooming out for a second, I really love what Evil Martians is doing. 00:14:45.519 --> 00:14:50.080 And you mentioned this on the you know evaluating agents, which I totally agree with. 00:14:50.240 --> 00:15:07.360 Um I really feel like uh ThoughtWorks used to be the poster child for this of a company that comes in and they're a consultancy, they've got great engineering, uh, they are a booster to the Ruby community, and they go and they take on open source. 00:15:07.600 --> 00:15:19.120 And the old days, what Thoughtbot used to do is go in and take over aging but very useful and highly used open source projects, take them over and maintain them. 00:15:19.200 --> 00:15:21.840 And it it was a real boon to the Ruby community. 00:15:22.000 -->

00:15:28.960 And I see Evil Martians doing that and also being innovative and releasing everything that they do. 00:15:29.200 --> 00:15:30.480 It's really helpful to the community. 00:15:30.559 --> 00:15:34.000 It's helpful to companies like mine, it's helpful to our engineers. 00:15:34.240 --> 00:15:36.639 I haven't played with Rails Hyperdrive yet. 00:15:36.720 --> 00:15:39.600 It was just released yesterday, but I planned to when I read about this. 00:15:39.679 --> 00:15:40.480 I was like, this is great. 00:15:40.559 --> 00:15:42.559 I've got a Rails project I'm about to start. 00:15:42.960 --> 00:15:45.840 I would incorporate this right away and see how well it works. 00:15:46.159 --> 00:15:47.039 Yeah, absolutely. 00:15:47.120 --> 00:16:00.159 And I, you know, reading through it here, one of the biggest things that stands out that maybe I didn't glean from the original title here is that it allows any Ruby gem to bundle their own knowledge. 00:16:00.639 --> 00:16:09.519 It uses uh you know a standardized protocol if you're using Rails HyperDrive to pull all that knowledge from any any nested gem that gets bundled. 00:16:09.840 --> 00:16:13.679 And it serves it via MCP over a cloud command and agents. 00:16:14.080 --> 00:16:14.639 Really cool. 00:16:14.960 --> 00:16:22.320 Okay, so we're not we don't have to go out to the open web every time we want to figure out what's happening with our gems or how it's impacting our project. 00:16:22.799 --> 00:16:23.200 Exactly. 00:16:23.360 --> 00:16:25.279 Yeah, and they'll just happen a lot. 00:16:25.759 --> 00:16:32.559 They have like a default Rails hyperdrive layered Rails skills from Vladmir. 00:16:33.039 --> 00:16:45.840 And it has basically some defaults that you can get out of all different like planning and review and spec tests and things like that that you can layer into uh the knowledge that gets served to your agents, which is pretty cool. 00:16:46.720 --> 00:16:52.799 I'm gonna have to go through my gems now and see what I can add from a knowledge perspective, because that that definitely seems like the future. 00:16:53.039 --> 00:16:57.679 Package up what knowledge is available to your agents or anybody that uses it, yeah. 00:16:58.000 --> 00:16:58.399 Right. 00:16:58.720 --> 00:17:04.880 I'm curious, are you using anything at def method that runs and loads with your AM for agencies? 00:17:05.359 --> 00:17:06.960 Across the team, like across teams? 00:17:07.119 --> 00:17:11.039 No, and it's I think it's mainly just because so much of it is client-based work. 00:17:11.279 --> 00:17:22.640 So if I have something, I might use it across a couple of projects that are just internal to def method, and my team may use them, but you know, when they go out to one client and then another client and another client, they've just been too different. 00:17:23.119 --> 00:17:26.559 You should package this up and have a client installer knowledge. 00:17:26.799 --> 00:17:27.519 Yeah, there you go. 00:17:27.680 --> 00:17:32.079 And have your clients configure their knowledge base and share it with your teams that make it easier. 00:17:32.319 --> 00:17:32.640 You're right. 00:17:32.799 --> 00:17:34.079 I feel like there's something there, you know. 00:17:34.319 --> 00:17:35.119 You're right. 00:17:37.039 --> 00:17:38.000 We have timing. 00:17:38.319 --> 00:17:39.039 Yeah, I know. 00:17:39.599 --> 00:17:43.119 All right, so this next one I'm uh I'm I'm uh really interested in. 00:17:43.279 --> 00:17:48.079 So simple cove 1.2 was released on September 4th. 00:17:48.240 --> 00:17:50.319 It adds production coverage. 00:17:50.640 --> 00:17:51.680 So it adds a few things. 00:17:51.759 --> 00:17:58.880 So production coverage, meaning it's showing what is actually running in production and whether your tests are testing that, which is interesting. 00:17:58.960 --> 00:18:07.279 It also has a simple cove dead code command, surfacing code that a task technically covers, but it never runs for a real user. 00:18:07.599 --> 00:18:13.279 It also has simple cove effected, which tells you which tests you need for a given diff. 00:18:14.079 --> 00:18:21.920 And finally, it has simple cov Ratchet, which lets a team set a coverage floor that can only go up over time. 00:18:22.799 --> 00:18:32.319 So I'm curious, uh, assuming that you use SimpleCove or something like it for most of your projects, how meaningful is this announcement to you? 00:18:32.720 --> 00:18:35.119 Yeah, this is pretty interesting. 00:18:35.680 --> 00:18:43.279 I've been leaning heavy into cover band, which I know I think simple cov is used in some capacity underhood of that. 00:18:44.000 --> 00:18:51.839 I always ask at Czar, we have some pre-commit checks that are like, hey, run the full suite before you even push and open up PR, right? 00:18:52.160 --> 00:18:53.359 And it can be painful, right? 00:18:53.440 --> 00:18:56.960 Like some of the tests, like full test suite, I could take like 25 minutes. 00:18:57.200 --> 00:18:57.519 Oh really? 00:18:57.759 --> 00:19:00.799 That's what you know, sometimes, depending on the change. 00:19:01.119 --> 00:19:03.519 So there's some caching in place that helps bring that down. 00:19:03.680 --> 00:19:06.240 But you know, so I don't want to wait that long. 00:19:06.640 --> 00:19:10.799 I have some local overrides that are like just focus on the changes under test. 00:19:11.200 --> 00:19:19.519 And I have another like spin-off subagent that then goes and gathers what other tests may be related that just aren't directly related to the files that are changed, right? 00:19:19.599 --> 00:19:21.279 From maybe from a behavior standpoint. 00:19:21.440 --> 00:19:22.319 It works pretty well. 00:19:22.480 --> 00:19:29.920 Maybe every so often I'll miss one test that should have run, but like overall, like I'm not asking people to review something that test's gonna break. 00:19:30.240 --> 00:19:34.480 And so I'm hoping that maybe this can help solve a lot of that.

00:19:34.799 --> 00:19:40.400 I like simple cove affected because I think that solves the kind of issue that you're talking about. 00:19:40.720 --> 00:19:41.680 I'd be interested in this. 00:19:41.759 --> 00:19:54.960 I mean, one challenge is that it's a dynamic code base, and so it's probably difficult to see you know if it's loaded code versus it's code that is actually touched by running through the actual workflow that's under test or that's being changed. 00:19:55.119 --> 00:20:00.799 Uh simple cove dead code, I think, is probably gonna end up being less useful than it seems. 00:20:00.880 --> 00:20:05.039 Like I, you know, I first read it, I was like, oh, that's cool, maybe you'll start deleting some code, right? 00:20:05.119 --> 00:20:06.079 But maybe not. 00:20:06.319 --> 00:20:07.599 You never really know, right? 00:20:07.680 --> 00:20:08.960 This thing is under production. 00:20:09.200 --> 00:20:13.839 Somebody could use it, you know, somebody might get to that code that you think is dead. 00:20:14.079 --> 00:20:18.960 Again, dynamic language, it's difficult to tell unless you're actually analyzing it yourself. 00:20:19.279 --> 00:20:24.480 I'm actually most excited by SimpleCove Ratchet, which was buried in this release. 00:20:24.720 --> 00:20:31.920 I'm looking at the announcement, you know, you have to scroll for like you know half of the page or more before Ratchet comes up. 00:20:32.000 --> 00:20:33.440 But let me tell you about Ratchet. 00:20:33.599 --> 00:20:38.000 So we've been at Deaf Method, we've been building our own custom ratchet for a long time. 00:20:38.160 --> 00:20:38.319 Right? 00:20:38.480 --> 00:20:44.240 The idea is that you know you set a floor, the coverage floor, whatever you start at, which better be above 90%. 00:20:44.720 --> 00:20:47.519 And then, okay, so then the code coverage can never go down. 00:20:47.839 --> 00:20:53.920 You know, you'd think it's easy to do, but sometimes it actually is challenging as a percentage to just never go down. 00:20:54.000 --> 00:20:55.920 And so, but we like to maintain that. 00:20:56.079 --> 00:21:05.920 The challenge is before SimpleCove Ratchet, we used SimpleCove, we had to kind of put a Kluge in ourselves, which was written in code or using some kind of GitHub actions. 00:21:06.079 --> 00:21:19.039 And as soon as I let my agents add it, if they put up a PR where the coverage went down and they couldn't immediately find a reason or a way to get the coverage up, they would then adjust the floor down. 00:21:19.839 --> 00:21:22.880 Like, no, no, no, that's you're not allowed to do that. 00:21:22.960 --> 00:21:25.039 But I have to catch them in the act. 00:21:25.440 --> 00:21:27.119 It's like having a puppy. 00:21:27.359 --> 00:21:29.680 I'd be like, no, no, no, no, no, you don't change that. 00:21:29.839 --> 00:21:34.400 So I'm very excited about having Ratchet where it's there and the agents can't touch it. 00:21:34.559 --> 00:21:35.200 Hopefully. 00:21:35.440 --> 00:21:38.559 You know, that just reminds me the Onion movie. 00:21:38.880 --> 00:21:43.519 If you're not familiar with the Onion newspaper, it's like kind of like a party newspaper. 00:21:43.599 --> 00:21:44.240 I've never seen a movie. 00:21:44.480 --> 00:21:44.640 Yeah. 00:21:44.720 --> 00:21:58.960 And so there's a series of movies, but the first movie they had like a little bit where they were announcing a news of the federal government just recently announced that they are raising the limit of what it means to be obese. 00:22:00.160 --> 00:22:10.640 And they're like in a they're like interviewing people on the streets, and it's like previously obese man, and he's like, I'm just so happy that the government has finally done something about my obesity. 00:22:11.920 --> 00:22:12.799 That's good. 00:22:14.160 --> 00:22:15.119 It was really funny. 00:22:15.279 --> 00:22:17.279 But yeah, we're we're facing the same thing here. 00:22:17.440 --> 00:22:23.920 Yeah, agents are trying to change files all the time to just push themselves forward rather than fixing it. 00:22:24.079 --> 00:22:27.599 So hopefully this can help reduce that. 00:22:27.839 --> 00:22:28.319 I don't know. 00:22:28.400 --> 00:22:29.759 Maybe we'll just have the same problem. 00:22:29.920 --> 00:22:31.759 They use the command line instead. 00:22:32.000 --> 00:22:32.640 I know, I know. 00:22:33.039 --> 00:22:35.519 As I was saying that, you know what? 00:22:35.839 --> 00:22:36.960 Nothing is safe. 00:22:37.200 --> 00:22:37.839 Nothing is safe. 00:22:37.920 --> 00:22:38.720 Nothing is safe. 00:22:38.880 --> 00:22:42.160 OpenAI's own agents are infiltrating Ruby gems. 00:22:42.240 --> 00:22:43.279 Nothing is safe. 00:22:43.599 --> 00:22:48.480 Yeah, we use uh cover band in production, you know, for the dead code aspect. 00:22:48.720 --> 00:22:50.240 So do you find it effective? 00:22:50.400 --> 00:22:51.200 Oh, yeah, very effective. 00:22:51.279 --> 00:22:53.200 We've cut a ton of dead code. 00:22:53.519 --> 00:22:53.680 Okay. 00:22:54.319 --> 00:23:00.880 Because you know, we can be pretty confident that after a certain time has elapsed, right, that code is no longer used. 00:23:01.200 --> 00:23:02.240 And we just burn it. 00:23:04.319 --> 00:23:04.720 Not yet. 00:23:04.960 --> 00:23:05.200 Cool. 00:23:05.279 --> 00:23:06.240 Yeah, not yet. 00:23:08.319 --> 00:23:11.039 Alright, let's move on to the next item here. 00:23:11.599 --> 00:23:16.960 Google has shipped Gemini 3.8 Live, their live model. 00:23:17.440 --> 00:23:20.240 Claims to be the top voice AI benchmark. 00:23:20.720 --> 00:23:25.680 So two new real-time voice models Google says can talk and reason at the same time. 00:23:26.400 --> 00:23:35.519 Gemini 3.8 live extended thinking hit number one on artificial analysis' speech-to-speech quality index, which is pretty impressive. 00:23:35.759 --> 00:23:40.720 And 97.7% on the big bench audio, which is also impressive. 00:23:41.519 --> 00:23:52.079 So it auto-detects and

switches between 97 languages mid-conversation, the extended thinking variant, reasons, and speaks at once, dropping in a natural pillar. 00:23:52.160 --> 00:23:54.480 Like, let me check that while it works. 00:23:55.279 --> 00:24:04.880 I haven't had a chance to play with this yet, but uh do these benchmarks wins mean a uh better assistant, or just a better test taker? 00:24:05.039 --> 00:24:06.000 That's a good question. 00:24:06.720 --> 00:24:09.599 Yeah, I guess we won't know until we try it. 00:24:10.240 --> 00:24:14.960 I'm more excited about this than I have been about recent voice announcements. 00:24:15.519 --> 00:24:18.400 That's probably just because I've dug into it a little bit more lately. 00:24:18.480 --> 00:24:34.559 I've been listening to the Shell Game podcast, which is uh Evan Ratliff's podcast, and the first season he takes a deep dive into the practical uses and sometimes impractical uses of his voice uh agents, right? 00:24:34.720 --> 00:24:40.640 And he creates a you know an agent with his own voice and he's uh setting it loose on all kinds of things. 00:24:40.880 --> 00:24:56.720 But the thing that impresses me about this is that uh we are all going to, if we're not already, uh be speaking more and more to uh AI and not to each other, and we're not even gonna necessarily know the difference. 00:24:56.880 --> 00:25:07.440 Now, the big reason that we can tell the difference now, one, you could get these voices, they still sound a little wooden, they still don't have you know as much personality as a real person, but that's kind of fading. 00:25:07.599 --> 00:25:17.200 The big thing, I think, is that uh it takes a long time to reason and then spit out its answer, and that makes it feel like you're not having a human conversation. 00:25:17.440 --> 00:25:23.839 And one thing that came up in the show is that humans have about 300 milliseconds between question and answer. 00:25:24.000 --> 00:25:27.359 I ask you a question, you answer in about 300 milliseconds, right? 00:25:27.599 --> 00:25:32.319 Unless you're taking your time to think it, because I'm asking you such thought-provoking questions like I do all the time, Vee. 00:25:32.640 --> 00:25:37.599 But otherwise, you're just spitting out an answer, and you know, an AI can't do that, and for a while it was like two seconds. 00:25:37.680 --> 00:25:40.480 And in two plus seconds, you're just like, uh, this is a robot. 00:25:40.640 --> 00:25:40.799 Right. 00:25:41.039 --> 00:25:59.839 So, what I'm learning here from this announcement is putting in the natural filler, that's one thing, but then actually giving an answer and starting to feel like, okay, uh, I've got an answer pretty quickly, all of a sudden it lowers the barrier to how these things can be used and how often they can be used. 00:26:00.240 --> 00:26:03.839 That was one thing about uh what was OpenAI's tool use? 00:26:04.319 --> 00:26:15.039 They had the preambles where you could basically give like generic text output that would be used before the tool was even called, uh, that it would just instantly show. 00:26:15.440 --> 00:26:17.200 That was great advancement, right? 00:26:17.279 --> 00:26:28.799 Because like then anytime a tool was called ever throughout the process during real-time streaming, it would just like instantly start showing you feedback and like make it look like it's doing more work than it is, right? 00:26:29.440 --> 00:26:31.119 And I feel like that's what's needed, right? 00:26:31.200 --> 00:26:32.319 It really does need more time. 00:26:32.640 --> 00:26:35.680 And uh the more that it can interject those. 00:26:35.920 --> 00:26:37.759 I mean, we do that ourselves, right? 00:26:37.920 --> 00:26:40.160 Where we're just talking about something together. 00:26:40.720 --> 00:26:41.119 We make ourselves. 00:26:44.880 --> 00:26:46.720 We do it on the show here too, you know? 00:26:48.000 --> 00:26:50.480 It's happening right now, it's happening. 00:26:52.400 --> 00:26:55.759 And it's exciting to see these live models, to be honest. 00:26:55.920 --> 00:27:06.720 I haven't had a chance to play around with it since uh GPT Live originally came out, but I'm gonna have to revisit the podcast buddy and have him join our next show. 00:27:06.960 --> 00:27:07.359 Oh, yeah. 00:27:07.680 --> 00:27:10.400 Uh I mean using these models, it should be pretty easy. 00:27:10.640 --> 00:27:10.960 Yeah. 00:27:11.200 --> 00:27:23.279 You know what's interesting to me too is the things that we thought were going to terrify us or make us feel like we're in some dystopian future, sometimes they drop in and you're just like, oh, okay. 00:27:23.440 --> 00:27:31.279 I had this conversation with a friend of mine the other day where she was asking me, When I email you, is that you that responds, or is it your assistant, or is it AI? 00:27:31.519 --> 00:27:33.279 And she actually did not care about the answer. 00:27:33.440 --> 00:27:37.839 She's like, you know, if we if I want to talk to you, I'll pick up the phone and talk to you because we're old enough where we actually do that. 00:27:38.000 --> 00:27:38.160 Right. 00:27:38.319 --> 00:27:47.680 But she's just like, yeah, I just assume most of the time, you know, the higher up I go from talking to a director or above, I'm not getting an answer from a human. 00:27:48.400 --> 00:27:49.759 Well, that's interesting. 00:27:50.000 --> 00:27:53.279 And so I can see that going the same way with this. 00:27:53.519 --> 00:27:55.279 You know, it remains to be seen, I guess. 00:27:55.519 --> 00:27:56.559 Yeah, we'll see. 00:27:56.799 --> 00:28:00.400 I know a few people working in this real-time voice space. 00:28:00.480 --> 00:28:04.720 Maybe we can have them on and see where they're at with the great. 00:28:05.200 --> 00:28:05.839 All right. 00:28:06.079 --> 00:28:07.680 The gong has been hit. 00:28:07.920 --> 00:28:08.559 Yeah. 00:28:09.839 --> 00:28:11.359 You want to announce our next one? 00:28:11.599 --> 00:28:13.519 Yeah,

so this is an interesting one. 00:28:13.680 --> 00:28:19.039 OpenAI opens Codex's infrastructure as a public quote-unquote agent's API. 00:28:19.359 --> 00:28:22.400 So this one came out through OpenAI.com. 00:28:22.480 --> 00:28:23.519 They made an announcement. 00:28:23.759 --> 00:28:30.880 So one API call can spin up a production-ready cloud agent with long-running sessions and multi-agent orchestration. 00:28:30.960 --> 00:28:42.000 Now, the example they gave was: well, let's say you've got a production app where you want an agent that is always looking at the logs and making an assessment based on something that may be coming through the logs. 00:28:42.240 --> 00:28:45.680 Whether it's an error or it's an info or it's something that you need to interact with. 00:28:45.920 --> 00:28:52.640 What OpenAI is claiming is that, hey, you don't have to build your agent's harness and infrastructure from the ground up. 00:28:52.880 --> 00:29:07.359 You can borrow ours, and you can either use it in a sandbox or you can use it in your own infra, and it will just come sort of ready to go with a couple of your own custom tool calls, your own custom runbook, and off it goes. 00:29:07.519 --> 00:29:12.640 And they've got some early adopters that have you know said some nice things about it, of course. 00:29:12.880 --> 00:29:15.359 It's in public beta, there's no platform fee. 00:29:15.440 --> 00:29:18.960 So you're paying for tokens, of course, you're paying for the tools that are used. 00:29:19.440 --> 00:29:28.480 So I guess the question here is is renting out your own agent infrastructure a smart strategy, or should you really be focusing on building out your own ecosystem? 00:29:28.640 --> 00:29:29.759 I think that's a good question. 00:29:30.079 --> 00:29:31.119 I'm torn here. 00:29:31.279 --> 00:29:40.000 Like it comes back to like, you know, our previous one, right? 00:29:41.119 --> 00:29:45.039 R harness is the equivalent of model advancements. 00:29:45.200 --> 00:29:50.480 And what I mean by that is the bitter lesson was originally written with models in mind, right? 00:29:50.720 --> 00:30:00.960 Where as the models improve, the things that you add and customize to make it work for you can be removed, you know, over time, and then eventually you're just using the model for everything. 00:30:01.440 --> 00:30:05.039 And can the same be applied to like harnesses? 00:30:06.000 --> 00:30:21.759 I take this back to like maybe the original do you serve your own MCP server, or do you just like give access to the MCP server to open AI and like let it run the hosted MCP for you, right? 00:30:22.000 --> 00:30:22.319 Yeah. 00:30:22.880 --> 00:30:27.119 Really, if you're trying to dive into all this stuff, just use the harnesses that are there, right? 00:30:27.200 --> 00:30:27.920 Like, this is great. 00:30:28.079 --> 00:30:32.079 Build cloud agents, easily hook up one of GitHub Action or something like that. 00:30:32.319 --> 00:30:33.920 You wouldn't even need

a GitHub action, right? 00:30:34.000 --> 00:30:38.559 It could just automatically be built into your code using one of these sessions. 00:30:39.279 --> 00:30:42.240 And you know, everything is moving into the cloud. 00:30:42.559 --> 00:30:48.319 If you're trying to build your own harness locally on your machine, you're doing it wrong. 00:30:48.480 --> 00:30:48.880 Yeah. 00:30:49.119 --> 00:30:49.599 Right? 00:30:49.920 --> 00:31:04.880 You should be packaging up and trying to spin up some claud-related thing or codex thing in a virtual environment that's got its own space to work in a safe capacity that you control. 00:31:05.200 --> 00:31:08.559 Yeah, I don't know how many people's machines have gotten ruined at this point. 00:31:08.880 --> 00:31:10.880 Running things on their own computer. 00:31:11.279 --> 00:31:23.279 And you know, having it in the cloud also forces you to think about the capabilities and like access and all these things that you're just like, oh, you're on your machine and you just let it run. 00:31:23.440 --> 00:31:25.519 Like it's an easy mentality to get into, right? 00:31:25.599 --> 00:31:28.160 When you just like, well, I just spin up plot and it does my work. 00:31:28.319 --> 00:31:28.559 Right. 00:31:28.799 --> 00:31:32.319 And you still you don't think about all the things that your machine can do, you know? 00:31:32.559 --> 00:31:34.799 And I feel like this makes you think about it. 00:31:34.880 --> 00:31:36.559 You're like, oh, it's a machine, right? 00:31:36.720 --> 00:31:39.440 Yeah, it's absolutely right about that. 00:31:39.759 --> 00:31:43.519 So that makes me both agree and disagree with you at the same time, huh? 00:31:43.759 --> 00:31:44.160 Somehow. 00:31:44.319 --> 00:31:46.000 Where I think that yes, you're right. 00:31:46.160 --> 00:31:48.880 If you're building a harness on your own machine, you're doing it wrong. 00:31:49.039 --> 00:31:52.319 You definitely want to get into the cloud, you definitely want to do all those things. 00:31:52.640 --> 00:31:55.279 This is sort of replacing that also. 00:31:55.759 --> 00:32:09.599 And so I'm kind of like, okay, I understand that probably everything is moving this way, but I became however efficient or effective I am at running AI agents by doing the work. 00:32:09.759 --> 00:32:15.119 And then, you know, yes, you swallow the bitter pill, you learn the bitter lesson, you strip that away, but the learning doesn't get stripped away. 00:32:15.359 --> 00:32:16.319 And I think so. 00:32:16.400 --> 00:32:20.160 I'm not a hundred percent sure at an enterprise level, it's probably better to use this. 00:32:20.480 --> 00:32:29.200 At a small business level where you're trying to get your people better and you're trying to learn as you're growing, I would be careful with it. 00:32:29.440 --> 00:32:30.319 Yeah, for sure. 00:32:30.480 --> 00:32:36.000 You know, this is a great way to get the interfaces right that you need in order to switch to your own thing too, right? 00:32:36.160 --> 00:32:40.720 So, like if you don't want to have to think about what it all takes yet, right? 00:32:40.799 --> 00:32:44.000 But you do want to make use of it, I feel like it's a great step. 00:32:44.160 --> 00:32:46.960 And you know, Anthropic has their own version of this as well. 00:32:47.920 --> 00:32:51.920 And like the open code also has a similar version of this. 00:32:57.359 --> 00:33:07.039 But the main takeaway here is try and get your specific tasks in a cloud agent capacity where you're running it off your machine on a different one. 00:33:07.440 --> 00:33:08.720 Yeah, that's a good lesson to learn. 00:33:08.799 --> 00:33:12.400 I mean, it's a good thing for our listeners to take away. 00:33:12.720 --> 00:33:15.920 Alright, our last one here is supposed to be the fun one. 00:33:16.079 --> 00:33:17.440 I have no idea what to make of this. 00:33:17.599 --> 00:33:22.799 An e-ink picture frame that turns backyard bird song into Victorian illustrations. 00:33:22.960 --> 00:33:26.319 So there's a Raspberry Pi project called Foglarom. 00:33:26.480 --> 00:33:28.240 I have no idea if I'm pronouncing that right. 00:33:28.400 --> 00:33:40.799 It listens to your yard, it uses an AI audio classifier called BirdNet-Go to identify the bird and draws it as a 19th-century natural history illustration on an e-ink screen. 00:33:40.880 --> 00:33:42.319 So why am I talking about this? 00:33:42.559 --> 00:33:54.880 Well, because it hit number one on Hacker News just yesterday with 1,770 plus points, proving once again that we never have any idea what people are interested in on Hacker News. 00:33:55.519 --> 00:34:07.440 And so my question to you, V is identifying birds in your backyard the most delightful use of AI all week, or proof that AI's best wins have nothing to do with chatbots? 00:34:07.599 --> 00:34:09.599 Or is it something else entirely? 00:34:10.079 --> 00:34:20.079 I think of anything that's successful on hacker news related to something obscure like this, it's kind of a good social signal for what people are talking about. 00:34:20.320 --> 00:34:29.599 Because I have had three different people in my family now talk about a bird listening and identification app that they used to be. 00:34:29.760 --> 00:34:30.159 Really? 00:34:30.559 --> 00:34:31.119 Yes. 00:34:31.440 --> 00:34:33.119 In the past month, right? 00:34:34.239 --> 00:34:36.079 You're gonna have to have me over for Thanksgiving. 00:34:36.480 --> 00:34:38.239 I'm gonna I'm gonna meet these people. 00:34:38.800 --> 00:34:45.039 Yeah, my uh my brother-in-law and my sister, they were both, oh yeah, you don't have the app. 00:34:45.280 --> 00:34:46.320 I think it's bird net. 00:34:46.559 --> 00:34:47.440 I have to have to tell them. 00:34:47.920 --> 00:34:48.400 Okay, yeah. 00:34:49.199 --> 00:34:49.920 I don't know. 00:34:50.000 --> 00:34:52.000 I think it's an app from Stanford or something like that. 00:34:52.239 --> 00:34:52.559 Okay. 00:34:52.800 --> 00:34:56.239

They're all like just bird identification, and they like they're so excited. 00:34:56.400 --> 00:34:57.440 Oh, what is that noise? 00:34:57.519 --> 00:34:59.440 And then they like start bird hunting. 00:34:59.599 --> 00:34:59.920 Yeah. 00:35:00.079 --> 00:35:03.599 And somebody else, I don't even know who it was now. 00:35:03.840 --> 00:35:06.000 Somebody else like also was very interested. 00:35:06.159 --> 00:35:10.320 Then I'm sitting on my porch the other day, and I'm like, man, I wish I knew like what kind of bird. 00:35:10.559 --> 00:35:10.719 Right. 00:35:11.440 --> 00:35:12.079 Now you're missing out. 00:35:13.199 --> 00:35:17.840 Now I don't know if it's like they got in my head and now I care, right? 00:35:18.800 --> 00:35:21.599 Or if it's like a genuine interest of mine, you know. 00:35:21.840 --> 00:35:23.119 Yeah, yeah. 00:35:25.280 --> 00:35:25.920 All right. 00:35:26.159 --> 00:35:28.239 That does it for the news this week. 00:35:28.400 --> 00:35:41.360 I want to just say that I'm delighted that none of the news items that our AI brought up were about an AI researcher publicly quitting and screaming about how AI is gonna kill us all. 00:35:41.599 --> 00:35:46.320 I talked to my one you know normie non-tech friend right before the show, and he's like, Are you gonna talk about that? 00:35:46.400 --> 00:35:48.880 It's all anybody was talking about on the news all weekend. 00:35:49.039 --> 00:35:51.360 I saw the headline and never even read the article. 00:35:51.440 --> 00:35:53.280 I just do not care. 00:35:54.079 --> 00:35:57.519 So if you're in that boat, then I'm glad you listened. 00:35:57.599 --> 00:36:01.119 And if you were hoping we would talk about it, but I'm sure you could find other outlets. 00:36:01.519 --> 00:36:13.360 You know, I bet people were all in all the rage back in the day when the automobile came out and they were just like screaming from their porches, how could you bring this noise? 00:36:13.599 --> 00:36:15.760 You know, it's bad enough with the horse feet. 00:36:16.000 --> 00:36:18.800 At least the automobile was actually killing people in the street. 00:36:18.960 --> 00:36:21.440 You could at least point to that and be like, hey, look, this is bad. 00:36:21.760 --> 00:36:23.760 We didn't kill nearly as many people with horses. 00:36:23.920 --> 00:36:24.559 We haven't, yeah. 00:36:24.639 --> 00:36:26.320 We didn't kill nearly as many people with horses. 00:36:26.719 --> 00:36:27.199 Maybe we did. 00:36:27.280 --> 00:36:27.599 I don't know. 00:36:27.760 --> 00:36:28.719 Maybe we did, right? 00:36:29.280 --> 00:36:34.000 Rails World coming up, September 23rd to 24th in Austin, Texas. 00:36:34.159 --> 00:36:40.960 I'm giving a little lightning talk about the guide principles of next generation software development that I've been writing a lot about. 00:36:41.039 --> 00:36:44.239 And we've been experimenting a lot with the deaf method. 00:36:44.320 --> 00:36:46.639 So if you're there, I don't actually think you'll have a choice. 00:36:46.719 --> 00:36:48.239 It's just gonna be in between one of these talks. 00:36:48.320 --> 00:36:50.800 So just don't run out of the room if you see me coming. 00:36:51.360 --> 00:36:54.960 It'll be good, it'll be worth your seven minutes, I promise. 00:36:55.280 --> 00:37:01.519 We'll have DHH, of course, we'll have Matt's, we'll have Aaron Patterson, they're all confirmed speakers, and yeah, we'll have some fun. 00:37:01.599 --> 00:37:03.519 And I'll be recording from there. 00:37:03.760 --> 00:37:05.280 I'll let you know what it's like. 00:37:05.599 --> 00:37:09.920 Yeah, I'm definitely having FOMO already, and I it hasn't even started. 00:37:10.159 --> 00:37:17.280 Well, if you get out to ExoRuby in October, I'll be the one with FOMO because I'll be in Chicago that weekend, and I think that's gonna be a really good one. 00:37:17.519 --> 00:37:18.400 Yeah, for sure. 00:37:18.719 --> 00:37:19.679 This has been great. 00:37:19.840 --> 00:37:21.039 I love this new format. 00:37:21.199 --> 00:37:25.599 I love that we get to chat through all of the news that I I miss out on, some of it. 00:37:25.920 --> 00:37:30.079 And you know, until next time, folks, keep it coming. 00:37:30.239 --> 00:37:31.440 Yeah, we'll see you next week. 00:37:31.519 --> 00:37:40.639 Yeah, and let us know on Substack or any other means you choose what you want to hear about, because that's one way you know you can influence the show, and we'll get something in there. 00:37:40.880 --> 00:37:44.079 Other than that, we have no idea what's coming, which has been really fun. 00:37:44.239 --> 00:37:44.800 Yeah. 00:37:45.119 --> 00:37:48.559 Surprise! Surprise, you're talking about bird night. 00:37:50.079 --> 00:37:51.599 I'm so gonna try this. 00:37:51.840 --> 00:37:52.400 Yeah, yeah. 00:37:52.480 --> 00:37:53.039 I'm gonna want to do that. 00:37:53.199 --> 00:37:57.840 I have an ink display and I haven't had a use for it yet, so uh this is perfect. 00:37:58.079 --> 00:37:59.280 The birders are nothing to mess with. 00:37:59.360 --> 00:38:01.280 I I have nothing bad to say about the birders. 00:38:01.360 --> 00:38:06.400 That's a whole army of people that I do not understand, but I know that they love their birding. 00:38:06.559 --> 00:38:08.800 It's huge in New York City, especially in Central Park. 00:38:08.960 --> 00:38:10.159 So this is pretty cool. 00:38:10.320 --> 00:38:10.719 Yeah. 00:38:10.960 --> 00:38:11.920 All right, everyone. 00:38:12.239 --> 00:38:12.559 Later. 00:38:12.880 --> 00:38:13.840 We'll see you next week.