AI Pair Programming (NAATW 2025)

Transcript

This transcript has been lightly edited for readability from the original caption file. Speaker names are standardized using first name and last initial where available. Timestamps are adjusted so this presentation begins at 0:00 within the edited audio.

Presentation: AI Pair Programming. Presenters: Tim R., David V., and Matt B. From the NAATW 2025 Workshop, Cincinnati.

[0:00] Presenter Introduction

[0:04] Host: meeting data worldwide to my friend Tim R., joined by David V. and my friend Matt B., who is joining us all the way from New Zealand online. to present AI pair programming, Empowering A.A. Software. Together, they'll share how A.A. volunteers are using artificial intelligence to support open source development, shortening our build times, expanding. Participation and improving quality while staying grounded in service and humility. Please join me in welcoming all three.

[0:47] Opening and Context

[0:47] Tim R.: We're between you and dinner. Good afternoon, my name is Tim, and I am an alcoholic.

[0:55] David V.: Hello, my name's David. I'm also an alcoholic.

[0:58] Matt B.: Kia ora, everyone. My name’s Matt, I'm an alcoholic.

[1:02] Tim R.: Thank goodness we got through the sound check. All right, the three of us represent a, uh… subteam, if you will, of developers inside of Code for Recovery, and. For some time now, we have been working on. what is, fingers crossed, going to be the replacement, the next generation of meeting finder for the Online Intergroup, and so it involves a whole bunch of coding and a little bit of the different architecture, and… and all sorts of good stuff, and what we want to talk about today is. our experience, because each of us. Individually, and kind of a little bit at different times, started using AI to help us in our coding contributions as we were, you know, working on this project together. So we're going to share some of those experiences, and we're going to do this in a couple of ways.

[1:53] Tim R.: the first thing is, I’m going to pose some questions for… these two guys to answer, and I'm gonna, you know, provide. my answers, too, if I don’t break anything up here. And then we're gonna do a little bit of a demo. Okay. We're gonna do a little bit of demo, and the purpose of the demo, don't worry, there's not going to be a quiz, right? You don't have to try to read everything that's going to be on the screen. The purpose of the demo is just to introduce the ideas that we have been working with to try to make this work effectively. And, you know, this is a learning process. We are still learning how to use this tool to help us with coding. And there's definitely a lot more to learn. I do expect that we are going to use up our full allotment of time.

[2:40] Tim R.: If there's time at the end, we'll happily answer questions. If not, happy to talk with people over dinner. or later this evening. All right, so, just out of curiosity, and I think I already know that. The answer to this first part, but if you’ve used AI, you know, raise your hand. Now keep your hand up if you’ve used it to help with coding. Oh, sure, yes, thank you, Josh, yeah, okay. Alright, so quite a few of you, good, alright.

[3:12] Initial Resistance to AI

[3:12] Tim R.: So let's get started. I'm gonna… excuse me, jump into some questions first, and I'm going to start with my friend David over here. David, did you resist. Using AI at the start, and if so, why?

[3:27] David V.: I embraced AI, or ChatGPT, you know, the chatbots, pretty early on, I got so excited when I started playing around with it, I would, like, throw in everything that was in my refrigerator and see what recipe it could come up with, and I'm… Improve communication, but I was really nervous, actually, about using it for coding. And it's funny, it kind of ran the gamut. On one end, I thought, you know, it's gonna. come out with some junk. I don't wanna, you know, I didn't think it would be very good. And on the other hand, I thought, well, if it's really good, people are gonna… think that I'm cheating. I think I would… people would think less of me me if the… if the AI was actually doing all the hard work. So, a lot of fear. A lot of fear.

[4:07] Tim R.: Matt, how about you?

[4:08] Matt B.: Thanks, Tim. Yeah, for me, very much the same. Fear of the unknown. I… I resisted it a lot. And it sat there for a long time. And, um, I didn't… I didn't start off with coding either, um, but I went to, uh… I went to a seminar, and they talked about ChatGPT, and I… content creation, and that seed was planted, and I was talking to my brother about it, who works in software development as well. And he said to me, you know, he said, look, he goes, if this thing's gonna take our jobs from us, don't you think we'd better abuse it while we can? And that's how I got over my, um, that's how I got over the resistance towards it.

[4:50] Tim R.: Yeah, awesome, thanks. You know, for me personally, fear. You know, I know that's true, right? Definitely fear, and particularly, you know, not wanting to look like I didn't know what I was doing. Right? And that's easy. And, you know, I also might possibly be a bit of a control freak. So, you know, there's this fear of loss of control, right? And what's going on in the computer and all that good stuff.

[5:13] How the Presenters Use AI in Coding

[5:13] Tim R.: All right, back over to David. How are you using AI today in your coding?

[5:18] David V.: I use it for just about everything. one of the most exciting things is just kind of like, uh. having it help me kind of just explain code. You know, I can… I can point it to a repository, or I can just kind of select some code that I don't quite get to understand what's going on, and ask it, explain this to me, and it does such a great job of doing that, you know, much better than, you know. spending hours and hours just kind of testing things, or looking at Stack Overflow or whatnot, so it's a great. assistant, a great teacher in that way. And also, just to write new stuff. Sometimes, you know, if we… if a new feature is requested, you know, I used to be reluctant to start off with AI because it felt like cheating.

[5:57] David V.: now I'm very comfortable, it's just like, okay. Give me your solution, and then I… that's a starting point, and from there, I know I can improve on that, or… Hope I can, anyway. Thanks.

[6:10] Tim R.: Matt?

[6:11] Matt B.: Yep, I use it very extensively. I use it every day. I do software development from home, I use it in my business a lot, I use it for planning, I use it for strategy, I use it for, uh, refactoring, um, I use it for content creation, I do email campaigns for clients, and I use it for that. I work in industries that I know nothing about, so therefore I can use it for product descriptions and stuff like that for the website. So, there's many, many different ways I use it. I even use it as a guidance counselor in the marriage. you know, my wife always used to say to me, you know, how do you spell remembrance and stuff like this? And, uh, you know, it was kind of like… it would always bug me. Now ChatGPT's here, all I had to do was give her the login, and now she goes and asks,

[6:57] Matt B.: The, um… she goes and asks ChatGPT how to spell things now, so it keeps the peace in the household as well.

[7:07] Tim R.: Thanks, Matt. You know, I use it in a lot of ways as well, and I… I have found that for me, using it incrementally is probably the safest, right? So I really do try to keep it as an incremental use. I might ask, say, for example, Copilot. Which is what I use inside of the Visual Studio Code, VS Code. IDE… which is just a development environment. And I'll ask it. you know, to try to generate a function that I'm trying to get done. And, you know, it'll generate that function and then I'll go in and I'll say, okay. Now try helping me refactor that function. One of the things is that. I like functional programming, and I'm not great at it, and really, for those people who know functional programming, it gets really advanced and ugly pretty quick.

[7:58] Tim R.: And so, a lot of it is, you know, I have to work on to understand, and so I will use Copilot to help me get something refactored into a functional approach. you know, I also use Copilot to ask helping for, um, excuse me, to help identify specific violations. And let me explain that a little bit more. You know, there's a couple of principles that are really important in coding. One of them is, you know, don't repeat yourself dry. So, you know, we want to have one section of code. That does a thing. And it's not elsewhere found in the code, right? It gets reused. as it's needed in other parts of the code. And so I found that Copilot is an excellent tool for helping me find a violation like that. Same thing if there's, you know, if a function is doing more than one thing, and it shouldn't be.

[8:50] Tim R.: I can use it to help find those kinds of violations. And I'm also starting to use it more for code reviews. So, what we mean by code review is that somebody makes a contribution that they want to have as part of the code base. They're suggesting it could be in the app. And so, when we… when they make that commitment and they open that pull request in GitHub. Copilot now is a feature in there, you can click a button, it'll run through, it'll summarize the code. It'll even come up with some suggestions based on some guidance that you've given it. So it's really pretty powerful. and I'm using it to help, you know, we've talked earlier about. generative content, and using it to create. comments, issues, pull requests, and the documentation that goes with that.

[9:38] Successes and Confidence Building

[9:38] Tim R.: Okay, next question. David, back over to you. Greatest success in using AI so far?

[9:46] David V.: In your coding. Yeah, thanks, Tim. Two things really kind of come to mind. You know, when I joined the OIAA Direct, you know, subgroup there, sub-project. I, uh, I didn't have any experience with, uh, TypeScript. I knew a little bit of JavaScript, hadn't worked on TypeScript, hadn't done anything on React, really, so it felt very intimidating, so… AI is just so incredibly useful in sort of, like, helping you kind of learn and really kind of jump… jumpstart. into these new languages, new frameworks, and whatnot, so it's been really useful in that way. The other thing is just… just little pet projects. I actually, for my intergroup, we started using, uh.

[10:23] David V.: We had an ad campaign that we're going to use some QRs, and we needed a way to track the scans of those QRs, and I knew I could go out and probably get some third-party piece of software that could do this, but it… It would have been overkill, and instead I asked, you know. Copilot, or actually ChatGPT, to come up with something, and, you know, within seconds, it had exactly what I needed. And within 2 or 3 hours. you know, I kind of worked out the kinks, so it's just, like, it's just incredible how quickly I could come up with a solution, something that would have taken me probably a couple of days, if not more, I could do in two to three hours.

[11:00] Tim R.: David, you said earlier, you know, you commented about how. You had fear that concerns that this would. tool would make you look or feel dumber.

[11:10] David V.: Has that been true? at this moment, it feels that way. No, um… no, actually, I was afraid that it would make me feel a little dumber, it made me look a little dumber, but that hasn't been the case. If anything, it's really kind of like, uh, increased my productivity and made me so much more curious about how. Curious about things, and I'm just a lot more, less intimidated about approaching huge new, uh, territories. That's awesome. Thank you. Matt, how about you?

[11:38] Matt B.: Yeah, thanks, Tim. Yeah, so obviously, um, I've had a lot of success in my business with this, but this is a recovery, um, a recovery workshop, so I want to kind of keep it on that. So, you know, it's taken my business to a whole new level, but in terms of recovery, how… the difference that AI has made to me is… I got involved with Online Intergroup, and I was told that the Tech Committee needed some help. with the website. I like playing with websites, I'll go and I'll go along and see what I can do. Okay. And I went to… I went to the tech meeting, and I got involved, and they wanted to make a little change in the quality of data, and they wanted to change They wanted to put a form on the meeting guide, and they say, why don't we get people to repositoryrt a problem? You know? And, um, and this thing came up.

[12:26] Matt B.: Now, in order to get involved in the… in the meeting, in the online integrated website, I had no idea that the meeting guide was a single-page application that was running inside of WordPress. I thought it was just all on WordPress, and I would just have to go and find this thing. I had no idea, and over a six-week period, I had to figure out what was going on. in this whole website thing. And the guys at Code for Recovery were great, I got a lot of support. And with that, it helped me when I found out that that's where it was, because it took me 2 weeks to find out that these were the people that I had to contact.

[13:02] Matt B.: And… and then I had to figure it all out by myself, you know? And then the difference that AI makes is we're going to show you in a minute, exactly how I go about getting involved with the project, and we've been working on this thing, like Tim said, for about a year. And I had a full understanding of this project within 10 minutes through talking to AI. And that's the kind of difference that it does for me. It gives me an overview of what's going on, it teaches me, it allows me to ask questions that are specific. The advantage that I get over Slack

[13:30] Matt B.: of a stack repositoryrt, is that, you know, I've got to adapt the variables and things like that when I go and get an answer off Stack Overflow, you know? But when it comes to AI, I get my exact variable names, and I can actually use it specifically, and I can talk about this entire… this exact repository, and get an answer to my questions. So it's a very powerful tool for me, yeah. I like it a lot.

[13:53] Tim R.: Awesome, yeah, thanks. So, my greatest success with AI and my coding. You know, persistence is one of my character traits, I know that. really to the point of being a defect. You know, if I had something I knew I kind of wanted to do. I would start researching it. I mean, we're talking about the kind of stuff that the coders, the developers, like Doug and Josh could do in 15 minutes. For me, it was days. And so, I would dive into medium articles, research stack overflow, keep trying to come up with these questions that I could find. And I would spend days on this stuff. And what I found is when I started using Copilot. Minutes. Right? I mean, minutes. It just… that's how much it accelerated my development. … huge confidence boost. Right? When you can actually start. taking something that you didn't understand, you can understand it.

[14:51] Tim R.: And then you can get it into a code, you can get it into a commit. Now you're part of. Right? That's what I experienced.

[15:00] Demo Setup: Training the AI on the Project

[15:00] Tim R.: All right. Matt, would you go ahead and start sharing your screen, and let's start getting this demo going, because. It takes a little while, but we're not gonna actually just… stand here and stare at it. While he's running with this, there's 3 things we're going to try to cover in this demo. The first one is the idea. is to… Start training the AI tool about what is in your project. And so, Matt, go ahead. Oh, actually, I need to back up. so we made an assumption here that the person who is going to try to do this has done what's called a clone of this website. They brought it down to, you know, their local computer, and now they're going to want to try to start learning what is in this project and getting their AI.

[15:48] Tim R.: agent, local, set up and ready to work with the project, so… And I have to do a shout-out to Matt on this, because he, of the three of us, he has done the most work. At trying to figure out how to train that agent and make it, you know, interactive with him, give him the results that he's kind of expecting. So, uh, Matt, go ahead and click on, uh, generate agent instructions. And so, while that's running, I'm going to jump into our fourth question, but before I do that, is there anything else you want to say on the demo? No, okay. So our fourth question, while this is running.

[16:25] Lessons Learned and Gotchas

[16:25] Tim R.: Over to you, David. greatest gotchas. What is, like, the lesson?

[16:30] David V.: that you've learned that's been most impactful. Yeah, two things come to mind. It has happened a little bit, but early on, I did find myself in this situation where I was… I was trying to solve a problem with AI, and I kept giving him. some information, it kept giving me the next step, and I was just, like, giving it more information. And, uh, before I knew it, I ended up in the loop. I didn't realize I was in the loop. It's just like, it was feeding me the same solution over and over again. And it's just like, you know, go figure, you get, you know, you get the same information to a computer, it's gonna spit out the same result. So, and it's a little bit more, you know, it's a little bit more complicated than that in that it.

[17:07] David V.: It really is sometimes, um, difficult to realize that you're actually caught in this loop, so you have to, like, know. Realize that, step back, and actually feed some new information. And the other thing that just comes to mind is just, it really is quite powerful. It's powerful enough where it can really breed complacency. So it's so easy to just kind of give it a problem. It… you think it's got a great solution, throw that solution up, but you really, you know, for me, it's just like, I need to understand the solution, and I'm lazy, so sometimes I actually skip that step, and it's just like, oh, it's good enough. So, complacency is definitely something that I'm gonna have to be.

[17:49] David V.: on the lookout at all levels when using AI. I was thinking about this, and I'm sorry to digress, but it's just like, you know, I'm really worried about how AI is going to affect our ability to sort of, like. problem solve, because it's so good at doing that for us.

[18:05] Generated Project Instructions

[18:05] Tim R.: Okay, so Matt's computer has gotten to the point where it's generated a document, and we're not going to go through the document because it's just… it's too much for us right now, but we're happy to go through it later offline. But the idea of the document is twofold. It has gone through it, it has analyzed this project, and it's… it's provided some observations about architecture and what libraries and frameworks are required, and things like this. So for a new developer. who is wanting to get onboarded, right, to a project, this could be a really good document to start to read, along with the regular README. And also, as Matt has discovered, is this becomes a key document.

[18:45] Tim R.: to help guide Copilot in future interactions. Okay. So the next thing, though, typically, because the process, kind of, when somebody's going to start working with a project, is to get their IDE set up and ready to actually. work on making coding changes and have them run. So, Matt, would you go ahead and put a prompt in that goes and starts to. check and set up… the development environment.

[19:17] Development Environment Setup

[19:17] Tim R.: Okay, so, um, I'm gonna read it. I am in a… well, no, I'm not. Basically, it says I'm an aspiring developer, would you please set up my development environment? Okay, so this is an important step of the process, and it's one that normally we work together on, and so… And that's great, we're happy to continue to work with people to do this, but this can definitely accelerate this process significantly. while that's running, let me tell you about my lessons learned. So, I love the one about. I love the one about loops. The other one I've come across is reinserting bugs that I found. You know, I would find a bug. And I would spend some time figuring out how to fix it.

[20:03] Tim R.: And then I would run some code refactoring. And I'm like, what the… I thought… And sure enough, Copilot put the bug back in. So there are definitely some shortcomings, and what I've learned is that we. This does not take away. Our obligation as developers. To know and understand every single line of code that we are putting into this. Right? We still have to understand that, and we still have to be able to analyze it and know when it's right and when it's wrong. I've also run into an issue of getting. some bad guidance because Copilot didn't understand what version of any particular library I might be using in the code. And I think what Matt has taught us with this generate Copilot instructions is probably going to help with that. It'll be interesting to see how that succeeds.

[20:57] Tim R.: Okay, so we're actually paused. So, Matt, we had talked two different possibilities here. You would, one, tell your own.

[21:03] Matt B. / Tim R.: Mobile-first story, why don't we do that? Why don't you tell us your mobile-first story?

[21:08] Mobile-First Refactor Story

[21:08] Matt B. / Tim R.: So this was the greatest gotcha, wasn't it? in terms of… the meeting, so we'd had a meeting with the Online Intergroup, And Tim and David have been along, and we'd done a presentation as to what we wanted to do. And in typical fashion, you know, it's like, always… they're always looking for the things that are wrong, and you know, one of the big things that was pointed out was this needs to be mobile-first. So, the next week, we had a meeting. And it got to the morning of the meeting, and I hadn't done anything. And I realized that, oh, we need this thing, and I've got nothing to repositoryrt for the meeting. I know what I'll do, I'll just go and do this. mobile-first refactor. So I went into Copilot, I gave it the instructions. This was before I knew about Copilot instructions, and I said, I want to make this repositorysitory, um…

[21:58] Matt B. / Tim R.: Mobile first. I'm like, change the things. We talked about having separate things and extracting components, and keeping everything separate, and making the code base organized. So I had Copilot do that for me, and it did it in, like, 45 minutes before this meeting. And it produced a PR. I pushed this PR through to the repository, and… It produced over a thousand lines of code. I started crying. Well, yeah, he panicked, he swore at me, there was all kinds of stuff flying around this virtual room while we were, uh, in this thing. Because, like David said, it gets carried away, and away it goes. You know, and this is for Jim's benefit, you know? I'm the rip-shift-and-bust guy. You know, I will go and give it there. I like to see things happen fast. I like to see it create. And that's what… and that's what AI'll do.

[22:54] Matt B. / Tim R.: But the thing is, is what it does when it's moving that fast is it puts everything there, all on one document. So we end up with 2,000 line files. And then I'm going through, and I'm telling Tim that I want a code review done on these 2,000 lines, and he's like, well, how the hell am I gonna do this? And that's why he started using Copilot for reviews. I'll tell you the truth. You know, so… it's like, you know, that was the biggest gotcha, you know, is I… and that's where the Copilot instructions come into it, and what happens nowadays, I run these instructions, and I have individual files, and I… and the rules are there. So every time I ask something for Copilot, it's running off this basic set of rules.

[23:33] Matt B. / Tim R.: You know, and I've learned how to… I've learned how to take documentation to a whole new level in my business. This is an absolute bare minimum that I would do right now. You know, this Copilot instructions here gives me an understanding of the codebase, it allows me to be able to ask questions as to how can I start. But it's only the bare beginning, you know? You've got to be crystal clear with AI and your instructions, otherwise it will run amok. So, I think that'll do it, yeah?

[24:00] Live Demo: Adding a Button with Copilot

[24:00] Tim R.: Awesome, thank you. So, uh, Matt, we're now at the stage, we've got the development environment set up, and I understand we have a little meeting finder app here, right? And so what we would like to do is let's go ahead… there it is. Right? Shows 3 meetings, again, not the real key here. So, if you could go back into Copilot for us, and give it the instruction, let's, uh, let's add a button that pops up a window that says, you know, hello NAATW. So what's happening here, actually, while this isn't working. Yeah, while this is working, David, let me ask you this. If somebody in the audience had some basic coding skills.

[24:46] David V.: And they would like to join C4R, or how would they do that? I guess the very first thing to do is go to our website. We got a website. It's Code for Recovery, code4recovery.org, and there's not much on the website, but what is there is a form.

[25:02] How to Join Code for Recovery

[25:02] David V.: that you can fill out to join Code for Recovery, and we'll send you some useful information. The most important bit of which is our Slack channel, so come to Slack, and then come and join us at our biweekly meeting, and uh… That's… that's about it. Yeah, and I'll talk in a minute about… we do actually have a couple membership requirements, but he talked about the very bare website, and it is… it's very minimalist, I think is the polite way to… minimalist website, but we also do have a contribute button on there, and because we are a 501c3, and we… minimal expenses, but we cover them through your contribution, so…

[25:38] Matt B.: Okay, Matt, so it looks like, uh, it's still working, or it's almost… should be almost done, huh? Yeah, it's just about… it's just about done now, it's just doing the CSS now for the button. now it wants to run the lint. Do we want to go on and look at this now, do we?

[25:52] Tim R.: Yeah. Yeah, so what we wanted you to kind of see with this part of the demo is that the way Copilot works.

[26:02] Reviewing AI-Generated Code

[26:02] Tim R.: is it will put. suggested code up in your development environment, and it gives us an opportunity. You know, I love this characterization that somebody said in our meeting. Copilot is the Copilot. You know, we are still the pilot. Right, so we have to go in and look at this code and make sure that we understand. what it's doing, right? Now let's assume this is good, so you can go ahead and click keep on that block of CSS. That's the Copilot instructions. Do you want me to go to the website? … Should we… should we see if it's worked? Okay. So now we have our button here. Yeah, go to the website, please. No. And, uh, so now, yep, now he runs it. And it's got the new button that says Hello NAATW. So, yeah. Obviously a very… very simplistic example, but it kind of shows the process that can be potentially used to help.

[27:04] Closing Reflections

[27:04] Tim R.: make coding contributions. Okay, the last thing I want to just comment, because I said I was going to, unless, David, do you have anything else you want to add?

[27:12] David V.: I just wanted to stress again that, you know, it's just like, I remember years and years ago when the first kind of, like, code scaffolding kind of, like, things happened, where it was just, like, you press a button, and basically this program would automatically appear, just a… a skeleton of a program that then you could add features. It was great. It was a great time saver. But this is interactive. This is something that you can continue having a conversation with. And, uh, it's just a game changer. Thank you. Matt, any closing comments, final words?

[27:41] Matt B.: Yeah, just to build on what David said there, you know, this is not a… it's not at the stage where I can just say, go build me this app. You know, at the end of the day, I have to sit and nurture, I have to sit with… the AI, I sit and I do sit and have conversations, and we do… and I do plan and strategize things out with it. I do ask questions about the code that it's just created. How's that working? Why have you done that like that? Because if I just leave it to itself, and its own devices, then I'm not going to be happy with the results. But if I… what if I'm using it as a genuine, uh, a collaborative partner, as opposed to a replacement, then it does really good things for me.

[28:24] Tim R.: Wise words. my final thoughts. One. Code for Recovery is… I was gonna say this earlier, but it… we're not an A.A. entity. Right? And so, we… welcome coders who are from other recovery programs. We support other recovery programs, so our membership requirements. Number one is, we want the individual to be, you know, in good standing with a recovery program, loosely 12-step based. You know, Al-Anon, A.A. C.A., N.A., that kind of thing. The second one is that we want developers. And, you know, for us, the webmaster is kind of the client, right? The customer. They're the people we support.

[29:10] Tim R.: What we're looking for are the people who are willing to roll up their sleeves with PHP, JavaScript, TypeScript, Python, that kind of level of coding, okay? And we don't expect you to be an expert. We… We accept people who want to learn, as long as it's… goes back to have an honest desire, right? If you have an honest desire to come in and work to contribute code. then we would welcome you into our, our, uh… Membership. you know, the one. thing that I would also ask is that if you choose to do this. and you meet those requirements. Then, before you start, like, just generating a bunch of stuff. Communicate with us, right? Come in, have a conversation about what it is you want to do with the code, or what feature you would like to add, or ask us, hey.

[30:00] Tim R.: Which issue do you want me to start with? And we'll coach you through that issue and help you try to succeed with that process, right? Because that is good for us. And it's good for you. Overall, this has been an excellent learning process for me, because. I just am finding that I can start to contribute more and more, and we can get code, I think, turned out that's just as good, if not better, and we can do so faster.

[30:28] Q&A / Discussion

[30:28] Tim R.: So I think we're ready for any questions, if we still have a few minutes?

[30:45] Mike: get one online. Is that Mike with TIA form? There he is. Hey, thanks for stepping up to do this, fellas. Thank you, Tim. Thank you, Matt. Thank you, Dave. My question, Tim. is, you know, you guys spoke about. … to review. And, you know, double check and, you know, you're the pilot and Copilot is the Copilot, all that sort of thing. To what extent do you think this will be advantageous. in helping people who may not be particularly skilled. But without a particular background in web development. to maintain A.A. websites. As you know, there's a crying need for that. To what extent do you think this would be applicable in helping people get on board to do that kind of work? Thanks again, guys.

[31:54] Tim R. / Matt B.: I'm not… I have to be honest, I'm not entirely sure I'm understanding. Do you want me to get it? the question, Mike. So, to what extent… Yes, if you understand my question, otherwise I can repeat it, Matt.

[32:06] Matt B.: Yeah, sure. So, to a huge extent. You know, this is exactly what we've just shown you. is the ability that, you know, I mean, it was, okay, it was a very simple thing that we did, yeah? We didn't want to get too complex, because we just want it to be really simple. But ultimately, as a semi-skilled person, you know, if you can clone a Git repository, then you can generate a set of instructions on Copilot. So what it's… what this thing is doing here is it's giving you the ability… you don't have to know React, you don't have to know PHP, you don't have to know JavaScript, you have to know how to be able to ask the right questions of the AI, and the AI will do the work for you.

[32:46] Matt B.: So it opens up all of those doors. You know, if I'd known back in the day, if AI had been there back in the day, I would have given it that, uh, the Online Intergroup repository, and it would have told me that the meeting guide is a single-page application that runs inside the… it tells me exactly what architecture I'm working with inside. It tells me the rules, and it allows me to be able to ask questions about this specific repository. And the thing is, because it's AI, there's no dumb questions, so I don't feel stupid asking the AI a question that I might feel silly asking in a Code for Recovery meeting, because I don't want to look dumb. You know? So… so it's a huge help, Mike.

[33:24] Mike: Thank you, fellas.

[33:25] Tim R.: Thanks, Matt. Any other… All right. I think we're done. I think… oh, she gets to release you for dinner, not me.