Transcription provided by Huntsville AI Transcribe
So what we’re going to do, first off, welcome to Huntsville AI. We’ll do it quick. We got. two folks that haven’t been here before so we’ll do a quick uh go around you know what your name is kind of what you we used to call it twitter status but twitter lets you put a lot of characters in now so short who you are what you do with kind of like ai stuff and then pass it on gary uh my name is gary and i work at the sandy island south of the town uh curly bump and bang my way through devops and uh we’re Implementing some of the stuff here and there, Gemini, Q, and I’ve got my own little hold-lapse set up for robotics. Still, I found the 3-ohms resistance on the ground water for my race car, you see. So I can finally get that fixed and get it running, along with the robots for raspberry pies or tweens.
Just rabbit hobbyist, really. Hi, Alex. Hi, I’m Alex. I work for Northrop. I currently work on Albert McVillan. Cool. Swing. Yeah, I’m Lewis. I’m new at this community.
And I’ve got a couple things. I do work at Lytos.
We do a lot of software development.
They’re starting to make big AI there. pioneering a little bit there. I also have an open source project called Red Horse VR that I kind of maintain and some other field side things including robotics that Tinker and mostly it’s just those two right now. John. Oh, we got two challenge.
This is awesome.
Go ahead. Go ahead. My name is John William Taylor. I’m new to this group.
I mainly deal with like IT, cyber security, open with like AI as well as I’ve been learning that, and I’ve also learned that some of my SISC internships.
I’m currently looking for new opportunities right now since I was laid off six months ago. That sucks. All right, good to know though. I’ll continue on the back row of Hugh. Yes, I’m Hugh. I work for SISC on base software development using a… exploring AI pretty much at home.
All right.
Thank you. I’m Jack Francis. I’m the AI solution instructor at Davidson Technologies. So my role is a lot of strategy around AI, specifically around Department of Defense contracting. I’ve done a lot lately with generative AI, large language models to chatbots, to rag. I’ve been an AI nut for a long time. I do it in my free time.
It’s just what I love doing. So yeah, a little bit about me. Cheng?
My name is Cheng Chen, a faculty member from ISEEM department at UEH.
So most of my research has been trying to focus on building different model architecture for requirements and management. And also in my manufacturing side, research I do about a different type of model and computer vision and everything. Oh, Kai.
You have to talk. I’m going to say my name is Da. Kai.
OK. Staff at UEH. What page? Cool. John number two.
Network systems engineer.
I just like AI, so I just jumped into anything I could learn. Started coming here. Cool. In the same boat as you are. It’s not a bad place to network and find who works where and what they might have open, so feel free to do that. Let’s see, online.
How about Lauren?
Hey Jay, can you hear me?
Yes. Sweet.
Yeah, I’m Lauren Bales.
I work at CULSA as a senior data scientist. There I’ve worked on cyber and natural language processing, and now I’ve been transitioned into the drone department. So we’re looking to do some more AI work in there. That’s pretty cool. On the side, I like to work on my heart rate monitor. I build with like arduinos, things like that. And then I’m also into astrophotography, where I use some AI models to clean up images. That’s it. How about Chris?
You got a mic?
If not, it’s okay. They’re looking for the mute button. Yeah, they’re looking for the unmute button.
Okay.
There we are.
All right.
There you go. It was on another monitor. I couldn’t figure out what it was. I’m Chris Anders. I work at Replicated. I’m actually remote, so I’ve been in Huntsville for 20 years. I just don’t work in Huntsville. I do cloud operations and infrastructure. We do a lot of guys support, like how do you do customer success and support through AI, which is what’s got me really interested here. For hobbies, I work a lot on infrastructure, Kubernetes, DevOps, that sort of thing. So if you’re into how do you deploy and launch non-native things, I’m probably interested in talking to you. So go for it. Cool. We’ve got Phil boarding his owner. At least it showed up here. So, Phil, if you’ve got a mic, you can introduce yourself.
If you don’t, I’ll introduce you for you. Okay. Can you hear me?
Yes. Go for it, Phil. Okay. Jay, I’m Phil Boarding.
I worked in an AI group for a major oil company in 1980. No joke. At any rate, I’m living in Oklahoma. And so I’m remote now. Welcome to our remote folks.
Tom, you just walked in and missed most of it.
Say hi.
Hey everybody.
All right. He’s got me on the list for number seven there.
Yep, you’re on here.
So thanks for that. Going through, I’m going to hit this real quick and then move it over to folks that y’all would probably rather hear from. This week at the AI Huntsville Task Force, we’re on the Community Development, sorry, Workforce Development Committee trying to just do more about education for AI, what you can do with it, that kind of thing. Open to the community. This was at the Chamber of Commerce yesterday. So I had about 60 people in the room and then went through the whole catchphrase here was beyond writing emails. This is beyond just summarizing things, just your normal chat GPD, just to throw things out and get an idea of, hey, what’s possible? I’ll skip through most of this. This was a pretty interesting. I’m interested to hear some of your thoughts on this later. Maybe I tried to break this out into four buckets. One of them is people that use AI and have no idea they’re using AI. The second is who I was trying to talk to yesterday, trying to learn more about it, trying to use it more and more. Number three and four are pretty much what we do in this group.
You know, the people trying to build AI and then the one at the bottom is more of our paper review series for how at the very low level does this work.
That kind of thing.
If you have different thoughts on how to break this up into segments, because normally what I’m thinking about is the AI task force is usually hitting the top two levels and then we’re down at the bottom, you know, builders kind of thing. Oh, I think that’s a good way of framing it. One thing, and you may have seen this open AI release like a how people are using chat GPT two days ago, and they break down to like how people are primarily using and how trends have changed over time over the past two years and stuff.
So maybe a lot of like info from that that you’d be able to add into like kind of further dive down into like that group too. Right. Cause there’s a lot of surprisingly, like one that was surprising to me is that out of the.
over a million chats they sampled, only 4% of them are actually used for code development. So like even though we use it primarily for that, it’s actually a pretty small use case from a chat stream team perspective overall.
Cool.
Yeah, one of the things with, I can’t remember who it was that mentioned it, at the paper review we did last week on Microsoft’s kind of overview of Copilot and all, there was a giant assumption that when people were using Copilot, they were doing it for work.
when I’m at work and I have co-pilot, I’m not always using it for work.
I’m using it because I have it at work. You know, it’s just me, I saw like a little screenshot thing. They do a full paper on that, okay? Yeah, it’s pretty, I think it’s almost 50 pages. It’s pretty extensive, but yeah, saw the screenshot as well, but there is a link to, there’s a paper behind that. In general, y’all know most of this. The one I had to come back to over and over again is number one.
You’re responsible for the stuff that you push out there. Oh, more pages. What I went through was kind of a, I had to frame it somehow. So I decided I was going to open a, I had a jewelry boutique and I’m going to do a marketing campaign. So deep research to figure out market research and things like that in this generation, video generation, audio generation, shoving it all together. Deep research was actually pretty interesting. Came up with basically my strategy for all of this and then came up with the piece over here. It broke out all of our local jewelers and what their key items are, what their media campaigns are for and how to… I was pretty interested. One of the things it found was some company called Shop Boss that I’d never heard of, and I’m like, I don’t think this is real, and my wife’s like, oh no, I know them, and I’m like, oh, shoot. You thought it was an hallucination? I thought it was. That was a pure virtual store that is now moving into a physical location, things like that. Anyway, pretty interesting.
Also used it to plan a trip to Napa. Just as a side, we had some pretty good discussion in the room about where else can you use these things? And that was kind of the whole, this is just for that, but how could you use this? Where would you see this? And had some pretty good discussion.
I won’t do too much here. Image generation, I went from just the branded stock photo into an image of an actual jewelry store. Actually, this was… No, come on. Oh, forget it. I’ll just go right to Jim and I, since I’m on my own laptop.
No thanks. So this was… So I went from this image down to an actual jewelry store, because the one I was talking about doesn’t exist anymore. and there aren’t any images of it. So I’m just, well, let’s make up our own. So we did that, and they’re like, okay, that’s cool. And I was like, well, okay, what about if we make the floor a lighter color? And it just does that, okay. What would this look like at night? Okay, there you go.
And just trying to get this into the, sometimes folks when they see it in use, it’s like, oh, it’s not magic.
You can actually do these things.
That was pretty cool jumping over that We’ve done a lot with video generation here. So I’m not gonna get too far into that Audio generation with 11 labs.
We’ve done some of that here. So I won’t get too far into that Putting it all together. I Don’t know if the final result I’m gonna have to just go ahead and jump into video see if it’ll let me get out of the way this window Okay This is always fun when I’m doing it live.
I think I’m on this email account Which is nice I Don’t know if it’ll play the audio Or if I have to it may not Oh, you put it with AI music? I put it with a voiceover. It was with a voiceover. Let me see if my sharing is… important cloud, whatever.
Yeah, you might have muted it to keep it from repeating that. Yeah, sometimes if you share, it gives you options. Nope, that’s not what I was looking for. See a little speaker at the bottom is XTown.
Where do you see it?
It’s the battery. Oh yeah, that’s my, that’s my speaker here. Yeah, but usually if that’s XTown, it might not transmit it, zoom. Maybe.
There we go.
Okay, share screen, options, share salve.
Make sure you optimize for me, do I?
Okay, that’s good. I don’t want more.
I actually wanted to go back because this one doesn’t have audio. This was actually the video. I started off with the sandwich and I said zoom in to the bracelets in the case. It got kind of close. Interesting thing is the reflections that you actually see in the glass of the lighting above.
You know, things like that.
I mean, this was like super, super good.
I don’t want to jump to that one yet.
So you’re telling it to make the people move?
No.
Let’s see.
Here’s the audio maybe.
Starting next Friday, Jay’s jewelry boutique is offering 10% off selected earrings and bracelets. Shop early. This sale runs only while supplies last. So that was fine. And then of course, yeah. So the fun part about it was I’ve been sitting here trying to say the words Jays jewelry boutique and I can’t even get it out of all but you know I mean so it kind of shows how you could just patch all this stuff together to do a lot of interesting things. I think that’s the last thing I’m going to show you though the best part about it that’s not even a real baby. I generated the image of the baby first because I didn’t want to use somebody’s kids. Yeah. But I didn’t have whatever. That’s the first thing I thought I was like, is that even relative? No.
Yeah, I tried. So the tool I was using for that is Hedra AI using their free stuff. I tried a whole bunch of stuff. I’m not going to do one of me. is pretty horrible in the dogs doing it. The one I thought about doing was the one of me on the right, but getting 11 laps to do it in Spanish. I don’t speak Spanish, so you could have me speaking Spanish about, you know, my jewel. The other fun part, I didn’t tell it Jay’s jewelry boutique. I put the prompt in the chat GPT. I said, hey, I’m open to the jewelry store. I need a whatever. And it knows that my name is Jay.
It decided my store was called Jay’s Jewelry Boutique in blah, blah, blah. Like, let’s roll with it. Let’s go.
This one wasn’t starting next Friday. Jay’s Jewelry Boutique is offering 10% off selected earrings and bracelets. Shop early. This sale runs on. I was hoping to get a dog to do it for me, but it’s just not quite that good yet. Yeah. That was kind of fun. So with that, and that was basically it, I think. Yeah, list of tools, all that kind of stuff. Questions. Some of the best questions I got were related to copyright. And if I were to make an ad or whatever and push it out there, do I have the rights to these images? Do I have the rights to these things? you know, that kind of thing. So I said, I don’t know. I mean, if they want to, it’s possible to get someone to talk about that. Yeah. I spent like two hours talking about the AI symposium last year. Yeah. Right. Yeah. Well, that’s.
I did law and AI and everyone gave me copyright questions. Right. I’ll leave it there. Yes. Oh, okay.
When this was actually made do it again. We did a law and AI session with Andrew Tuggle and I try to remember what one of his co-workers came over. We’re in the executive conference room.
That was more wound up being more about copyright as well. and what can be copyrighted, what can be patented, what can be, you know, all of this kind of stuff. It was pretty interesting to show. If you want me to, I could add 30 seconds onto my five minute thing.
All right. Let’s get me off the stage for the not that we have a stage in here.
I did have you up next if you want to go. All right. All right. Or if you want to stall, I can get this. Well, no, I’ll go ahead and set mine up. I’ll send my laptop up while I answer that other question. OK. If you played with voice generation and stuff like that on your laptop at home and your pets start stalking you, like, I’m over here. What are you doing? Yeah.
Yeah. Like, Cat does that.
He’ll come up and get between me and the screen, like, meow.
Slick, are you OK?
Or if you want to try that way It’s not like great We can probably go a different if you want to try to use one of these things It works.
We did it last time that way. Okay. Yeah, it was. Jay, it might be more interesting if you didn’t rush an answer. What? You’re just right.
Yeah. You’re talking rushing on the job you do. The problem is, I think it would get out somewhere. And my facility security officer would call you in. Of course, we’ve met.
So she would not surprise her. But yeah, the last thing I ran in with her, it was, we were going on a cruise back in I think May. She didn’t like that. No, I forgot that going on a cruise is actually leading the country, even though I’m not like really going to any other countries. Yeah, I need a pass. Well, that’s the first thing I pulled out my passport to pack.
And I’m like, no, there is no need this.
I’m like, oh man. So next up I’m going to actually put a post-it note on my passport that says call Johnny. I know. This was like a day before.
We did the debrief and all the stuff and of course not going to like any other, you know, I didn’t have to go through customs anywhere. I was going to say in some places they would let you on with your driver’s license onto the ship, but that was two or three years ago. Well, even if you go to different countries, so I was looking for a last increase. If you leave out of Seattle, even if you’re stopping Canada along the way and you come back to Seattle, you don’t have to have your passmorph in for that. Yeah, because you’re technically always, I guess, I don’t know. I didn’t get it either. Yeah. Oh, so you didn’t have to report that trip? I did. No.
but it was still kind of weird. Where is the zoom? Is it in the light?
Is it in the calendar right?
Uh, scroll up a little bit. It should be. And I didn’t go into Chinese for the exact reason. I can’t imagine filling out that big part. Try the email. I didn’t have a clearance at the time, but I didn’t know if I was going to get one or need one later. So I said, no, I’m not touching that. My office mate went to China earlier this year, and to make that a peak on, not to a sharing-style level. Because of it? Yeah. Even though he reported it? Yeah, he reported it. Wow. I need to turn. OK. I would turn. I don’t know. And when you join, I would just… That’s really important, because I feel like I want to go to China, like it would be such a cool place to go and visit, but yeah, there is the worry of… Some videos, things, it’s like a really cool place too.
Yeah, I mean, I’m sure Beijing is awesome in Shanghai, like there’s so many cool cities that would be fun to explore and go eat all the good food there. There’s quite a lot of you, but yeah, I think it’s cool. There’s so many jobs. I’m not sure if China does. So before I applied for a clearance, I had like two or three people that had talked to me. Now I work with like, let me try to go for a clearance now and you know, list all the foreigners that you talk to. It’s going to be a total book.
All right, so I think I will first before I start this I’m gonna piggyback off of Jay for their comment about AI and law and stuff and copyright questions and ask are there any questions about that topic that anyone has and then that’ll give me five minutes to think about my answers while I’m doing this you know because I can multitask just like we all multitask right we we use AI to multitask now we can have the AI working visually and stuff while we’re doing stuff so that’s my question does anybody have any questions related to AI and law and in particular copyright When the question that came up, yes, I didn’t have an answer was like the baby. I generated a photo of a baby. I didn’t start with the photo of some other, but I just said, I need a baby with a big tail. And I’m using that in an ad campaign. Do I have to, you know what I mean?
It’s not like I need a model release.
It’s not a real person.
I don’t think. What do I care about?
Okay. So that’s his question.
Anybody else have a question?
Oh, go ahead.
I made a generated image of parody of cracker burgl. I don’t know if I can put it on a t-shirt. And you want to put on a t-shirt? Sell the t-shirt? Yes. Now, not make it for yourself personally, but sell it to others. Right. OK. Third question. So with Anthropic, the lawsuit they had been closed up about a month ago, where they were essentially All of the books that they used, as long as they manually went through them, I think is my understanding, they’re now, they were able to use them as part of like AI training. So how does that now impact from a, like for using anything from online, what are like the copyright restrictions around that given like what anthropic is now able to do?
All right. So let’s, we gotta think about three.
Actually, I’m going to give you a short answer on those and then I’ll come back at the end and give you an answer. had answers to some of this stuff. I am an attorney. I am an IP attorney.
I have represented Oracle, Xerox PARC, and a bunch of other companies actually doing AI stuff. I currently am not representing them legally. I used to, I work for Oracle on the business side now.
Everything I’m talking about is nothing new for Oracle today though. But I will give you some legal stuff. And what I’m giving you for legal stuff is not legal advice, because you’re not paying me. We have no relationships, so you can’t rely on anything. Treat this as sort of general academic stuff that you might read about, but not actual legal advice. So first quick answer, babies. Let’s imagine that you use that ad campaign and no one ever bothers you. OK, life is good. Let’s imagine scenario number two.
You get a nasty letter in the mail saying you used to follow my baby Yeah, we’re gonna sue you unless you pay us Okay So the question then becomes is you’ve been contacted by someone who claims that is a photo of their baby And they even send you a photo of their baby and you can’t tell the difference right? What is your legal? Problem that is a questionable answer in five minutes. All right He has the little Gemini story. Yeah. Parody of Cracker Barrel. So parodies in copyright law fall under a category known as fair use. You know, you should be able to comment on things. You should be able to create parodies and so forth. However, there’s also this idea that, you know, You shouldn’t make money off of cracker barrels, imagery, unless you have an agreement with them. And there’s this question that you might be hurting their brand by making money off of them without having an agreement. And so it gets into that little daisy era where you’ve got these two legal doctrines that are in conflict. And so that’s what we’ll get back to in five minutes.
And then the third question had to do with the Anthropic, one of the Anthropic cases, cause there’s actually multiple cases going on. And actually all of the big AI vendors who are running models, you know, Anthropic, Google, Microsoft, they’re everyone who’s open AI and you name it, that they’re all in various lawsuits. That’s a more complicated one because I haven’t read that particular sediment’s details, but what I will say is that when you’re dealing with individual cases, Remember how our federal system works, right?
We’ve got a federal government, we’ve got state governments. At the federal government, you know, you’ve got districts where reps are, you’ve got states where senators are, and so on, all the way up to the president. Well, the court system’s the same way. So just because you have a court that reaches a certain decision at a lower level, doesn’t mean the whole U.S. is bound by it.
So, you know, and particularly when you talk about settlements, settlements don’t even bind anybody except the parties to the settlement. And so, for example, if one book author makes an agreement with Anthropic that they can use their books in certain approaches, other book authors haven’t made that agreement.
They can sue Anthropic all day long.
You have to look very carefully at a settlement to understand what the rules are as it applies to others.
And generally it has no impact on others.
It only affects the current parties. So that’s something to keep in mind. Now that being said, even if it just affects the current parties, when there’s new controversies, they may look at that pattern and adopt that pattern as well. That’s same reasonable for them. Maybe it’s reasonable for us. So that’s some of the things you need to think about.
All right.
So we’ll get back to that in a minute. All right. So one of the questions I hear about a lot is how do I get into an AI career?
I really think this is hot.
And then I go and I look and help one of dads and everything says 15 years plus of experience.
But these technologies haven’t even been around for 15 years.
How do I get into it? So I put together this little slide deck.
about talking about how you can get into AI careers without having 15 years of experience doing AI. And spoiler, one of the things I’m going to recommend, say a lot. We’ll try to do this quickly, because I know we have some other stuff coming up. All right, so there’s a lot of alternatives to meet AI experience of X plus years. Now, some ads say 10, some ads say 3, 5, whatever.
But however many years of experience that you want to say you have when you really don’t have it. So we need substitutes. All right, so some of the major substitutes you can approach is to do presentations at this event.
Jay is looking for other people to stand up here and talk about AI. So everybody in the room, you now have a challenge, stand up here and talk about AI. But you know, do presentations on AI at conferences, YouTube. I’m giving a presentation at the National Cyber Summit in Huntsville next week on agentic AI and cybersecurity. Why? because I said I’d do it and no one else did.
So now I’m doing it. So volunteer to talk about AI. It’s a way to substitute for experience. And usually I try and put everything I do on YouTube also. So write articles, blog posts, books.
I’ve done all the above. Volunteer for open source projects.
Open source projects never turn anyone down. It’s easy to volunteer. Just do a little bit of help. And even if you don’t know the code base they’re working with, they always need help with documentation because open source developers are horrible at writing documentation. So technical certifications are a great way to show that you know the information because there’s a lot of people who’ve never bothered to do them. And if the person doesn’t know you, this is a good way to say, yeah, you’ve actually done some stuff. And nowadays, there’s a lot of brand new AI technical certifications being offered by all the usual players, NVIDIA, AWS, Microsoft, Oracle, and so on. I did a presentation about a couple months ago about how to do the NVIDIA certification. So I’ve done, I don’t work for NVIDIA, but I’ve done for their certs. It’s not that hard. It gives you a little practice at looking at their technology in depth and understanding what’s going on there. Now we get to the important part of tonight’s topic. Participated in hackathons.
I talked about certs, substitute for job experience, volunteering for nonprofits, we talked about that. All right, so hackathons. They’re a great way to demonstrate two things.
Your interest in AI and your experience with AI from an entrepreneurial and coding perspective. So what do AI startups look for? They look for someone who’s interested, someone who’s entrepreneurial, and someone who has coding experience.
So a hackathon does all three for you.
And a lot of startups get started out of hackathons.
And you can get paid money for participating in a hackathon.
I’ve gotten thousands of dollars by spending just a couple hours in a hackathon.
So usually most hackathons are over a weekend.
Nowadays with some of these virtual hackathons, they can be a little bit longer.
So it’s up to you to decide and weigh what you want to invest your time in. But usually a pretty short commitment. And the other thing is, is when you work on a hackathon, you produce a software project that’s typically public.
You put it on GitHub, you’re putting it in the contest, people are taking a look at it.
So you can always use that as a project in the future. You know, by none hackathon projects. I can’t show any of those to anybody because they’re all locked away by work and you know I can say I worked on it but I can never really show the code hackathons I can show every single code base from every hackathon I participated now I’m probably not going to do that because some of those hackathons we only spent 20 minutes on and but you know I mean I can if I want to share every single hackathon I worked on um you meet people This is a really important topic because where do you get careers and jobs from? For people. Meeting more people is a good thing.
And not only that, but you get to spend time with them. If I come over here to this event and I spend an hour here, okay, I spend an hour with you this week. But if I’m working with you on a hackathon project, I’m spending maybe 30 hours with you over a weekend. I really get to know you better and you build a… an intense relationship in a two or three-day period. If you’re doing a virtual hackathon, there’s really no cost other than your time. If you’re traveling to an in-person hackathon, then there is an in-person cost if you’re actually traveling. I highly recommend you save all your project materials in GitHub and YouTube so you can use them as public references in the future. Even during the hackathon, you might be keeping it somewhat private, but in the future, if the hackathon’s over, Go ahead and make them public.
So how many people here have done a hackathon?
All right, so for those of you who’ve never done a hackathon, do you have any questions about what a hackathon is and what it’s like?
How rigorous is it with math?
It’s entirely up to you. So here’s the thing. So there’s two different types or models of hackathons, okay, in general.
In one case, you have what’s called a bounty approach, where some person who’s given an award has put out a bounty and says, we’ll give you X dollars if you meet this very detailed goal. All right. So Kaggle competitions are like that and some other things are like that.
On the other hand, You have hackathons where they say, we’re going to give X dollars to the best project.
Well, anything could be a best project.
So you don’t need to know math or anything else.
You just do what you think would be a best project. And if people like it, you win. All right. And so I generally stay away from the bounty of ones. where it’s very defined as to what you have to do, under the theory that if you need to do something very defined, you can hire me for a job and I’ll do it. But if I tend to focus on the ones where it’s the best project, where I can decide exactly what I want to do, because I can’t remember, I want to use this as a reference in the future. And it also has to be something I’m really interested in. You know, something that’s really defined, I’m probably not interested in, unless they define my perfect problem. In which case, I’ll do it.
If you wanted to find a problem about me going to the Bahamas and having fun in coding, sure, I’ll do that.
But if you’re just going to give me some, you can hire someone new to the job if that’s really defined.
Go ahead. What is the one thing to avoid a hackathon? What’s a pet peeve that just people have?
Is there anything like that?
OK, so here’s one interesting topic about hackathons.
You get people from all over the place.
all sorts of different types of people, all different ages, backgrounds, you name it. So there isn’t, I don’t necessarily have a pet peeve.
What I would say is that assume the hackathon will be disorganized.
Assuming it’ll never be on time.
So plan into your timing some extra time. So let’s say, for example, the hackathon is supposed to wrap up by five o’clock in the evening.
I assume that it may not wrap up by five.
So add a few extra hours there.
But, you know, I’ve been to multiple hackathons, so I said, oh, we’re going to do the judging at noon, and we’re going to hand out the prizes by two, and the prize is being handed out at 5.30.
So assume that the organizers are going to be a little disorganized.
That’s, you know, I’m a trained runner-time guy, and so that’s a really obvious thing to pops up.
But in general, the other thing is, I’ve done a lot of hackathons on the west coast in Silicon Valley in those areas. And so over there, you’ll get a lot of VCs involved because they’re looking to fund startups. So you see a lot of startups participating in hackathons.
Either groups that want to become a startup or even funded startups will go and participate in hackathons because they’re using it as a mechanism both to try out their new product as well as a mechanism to find people to work on their product.
find people who are interested.
One of the things we see, there are two primary hackathons locally that I know of.
One of them is NASA Space Apps Challenge in the fall that used to be much more defined and better put together.
The other is the Hudson Alpha Tech Challenge that is based out of Hudson Alpha. Thank you Hudson Alpha if you watch this for hosting us. It’s in March and that one is a little better put together. You know, it’s sometimes what I see and I’m usually joining in and it’s a different thing for me, a different kind of a technique on it.
A lot of times when I’m working with a hackathon, they always need mentors.
Yes. So as an alpha, I sign on as a mentor.
What do I get to see?
I get to see how all of the different teams work together.
Oh, you know, and kind of look through this, you know, it’s kind of interesting, but they always need that too.
Yeah, if you think about it, if you’re participating in hackathon, there’s probably five different roles, okay?
There’s the hackathon person who’s actually doing the hackathon.
That’s role number one.
There is hackathon observer, just someone hanging out and watching.
Hackathon mentor, you actually got an official role, official reason to watch, as opposed to the observer who doesn’t. Then you’ve got the hackathon judge, this is an observer who’s actually voting on the prizes.
And often they’re also doing the mentor role as well.
A lot of times, also, they’re celebrities. I met all the candidates for Wyoming state governor back in 2018 when I was at the Wyoming Blockchain Hackathon because they were all there trying to say that they had been cool and interested in blockchain. And then the fifth type of person besides the judges is the hackathon organizer themselves. And the other person who’s running around crazy is always three hours behind.
And I’ve been that person, so I know how it goes.
So I used to run a hackathon for a small nonprofit startup that, and we ran several hackathons at Silicon Valley, at Cloudera headquarters about 15 years ago. Cool. All right, so let’s move on. Yeah, we need to move pretty quick. Yeah. Do you use adults? my general advice select something you want to work on don’t do something just because someone else tells you to and that includes the organizers don’t do it even though you know if you don’t want to do it don’t do it this is volunteerism this is not a job definitely think about possible projects before the hackathon starts because you want to hit the ground running um and maybe build your team before it starts or find your team in the first couple hours uh don’t add anyone after the first couple hours If they haven’t found a team by then, you don’t want them. If you’re on a team, presume only half the team’s going to significantly contribute. And the other half will, you know, they’re just there to hang out and give you positive feedback. So be comfortable if you’re doing a team. Don’t be like mad at them for not being able to significantly contribute. And definitely save it so you can use it as a reference in the future. And that’s really what I want to talk about.
One thing I would recommend adding to that last bullet is someone who’s been on the hiring side.
If you are saving your project to GitHub.
make a read me and explain like what you did, your contribution, why it was important. Cause I have dug into many a GitHub where it is just a Jupyter notebook and it’s just all of this code and I have no idea what’s going on and I’m not going to look through that from a hiring perspective. But if you have a read me, which I’ve seen in a lot of people’s like, Hey, here’s what I did. Here’s the contribute contributions I made.
I’ll go through and read that.
And then I’ll go look at the code from that.
So just stay from the other side.
If you are trying to use this as a way to start to get a job or to move into the field, read me. first you’re helpful yeah you always want to have a read me if you’re gonna give someone your github because yeah this no one has time to really read through multiple files especially to like if you’re producing YouTube channels or other videos to explain what your project is which you most likely are doing in an act of thought because you got to explain to the judges what your project is doing because they’re even though they claim they can read your code they will never read your code I’ve been an organizer and a judge, and I’ve never met another organizer or judge who actually read the code. So just telling you that. Now I have met judges who will ask me if I wrote the code, and I’ll say yes.
So they’ll ask you, but will they read it? No.
Because they’re busy. They just don’t have time. So put those videos in your readme, because people may actually go watch the video, even though they’re not really going to read YouTube or no one. Yeah, and I do all this stuff myself. So that was it for my slides.
Let’s stop sharing there.
I need to make someone else a host. Yeah, I would say get back. All right. Where do I make this?
If I can.
Yeah, I might be able to say replay host. I’m a host. Oh, you’re a host.
All right.
That works. Yeah, so with the earlier questions. So when someone sues you or sent you a nasty letter claiming that you have not paid them for using the photos of their baby Then you have to show them where you get the baby photo from and Tell them. Hey, look, I generated this. I didn’t use your photograph Now what they could theoretically do You know, let’s say you used Gemini They could theoretically sue you and Gemini. Probably they’re not going to do that. Probably they’ll just go away. But that’s what I would do is I would tell them, hey look, I didn’t use your photo. Now, do you have any evidence to show them, tell them that you got the photo from Gemini? So that gets to logs. Do you have a log of what prompt you used?
to generate those photos.
So, save your chat logs if you’re generating photos and using the photos or imagery or video in a commercial offering. Right. That’s my advice there. Not advice. Yeah, not advice. That’s my general lecture, right? Well, observations.
Yeah. Not one if you don’t mind. Yeah, sure. Go ahead. open source, I have an open source repo, and when I made it open source, I added a condition for licensing that says you cannot use any of the data on this repo for training AI without written consent. I would that play if I find out one day I’m prompting an AI and it starts pumping out stuff that I know came from my repo. All right, so we’ll take a 30 second pause on that one and we’ll answer the the other question and then we’ll come back.
So again, your issue really comes down to the parody versus, you know, commercial use. And really, honestly, I don’t know that AI really changes that legally at all. From a parody perspective, it’s just like, you know, you drew it yourself or you had the AI generated as long as the intention is to do cracker barrel and it’s a parody. I think it’s just going to be under the basics of the way copyright law normally handles parody. I don’t think AI Anderson did years at all. Now, it’d be different if you didn’t know it was a cracker barrel and that’s a slightly different issue. It changes the wording a little bit and it’s a different block. Right.
It’s not like trying to re-represent cracker barrel.
I know, but it’s still based off of record on your mind. Same color, same style of work. Yeah, yeah, it’s gonna come into the whole doctrine of when it can be a parody or what can be a parody. Now, are you saying that it’s intended to be a new business, it’s not intended to be a parody? Yes. Okay, so if we step away from being intended to be a parody, then you no longer have that safe harbor of being a parody for fair use. So now you’re in a weaker position. Now the argument they’ll use is actually not copyright. The argument they’ll use is trademark. They’ll all say that what you’re doing is close to their brand and they’ll come after you under trademark terms and it won’t really be copyright. And then it gets to be a question of how close are you? Did you change enough that you’re now just a generic business or are you still very, very similar to them?
And that becomes a question of branding and what trademarks do they actually have? Maybe they don’t have the trademark coverage to really cover what you’re doing. So a lot depends then on the specific details of what they’ve done to protect their brand and how close you are. Obviously, we have millions of businesses out there. There’s going to be a lot of similarities. So you just kind of say, well, yeah, we’re not that similar. And you point out the differences. Assured itself as a parody, but it’s not gonna be like, it wouldn’t be like a logo if they come here. Okay. All right. So back over here. So since I didn’t have five minutes to think about it, not sure, I can really answer right at this moment, you know, didn’t do all that multitasking. Why don’t you repeat the question one more time? Yeah, I mean, answer it later. No, go ahead and repeat it. But my repo, when I made it open source, I put a disclaimer on the license entity. Yeah. There’s several conditions and all that stuff, the GPL and that.
But then I put one in there that says that you cannot use anything in this repo for training in AI without written consent.
OK.
All right. So you’ve got the GPL as your open source license.
gnu public level.
But that only applies to certain entities. So certain entities, smaller.
And then you’ve got various other licenses.
And then you’ve got that little repo disclaimer. All right. So one comment I’ll make is that the sort of disclaimer you use has actually been on the internet for a really long time. in another context. So if you go back to the 90s and the dawn of search engines with Yahoo and Google and all the rest, in the early days of the internet, search engines hitting your website was actually kind of painful for your website. It put a burden on your, because remember a lot of people were running websites off of machines in their home or their office. So we started putting little comments on websites in the metadata of HTML pages saying, please only visit this website once a day or once a week or don’t go to any child pages off of this page, only hit the main page, little direct use. to the search engines and how to be a good player when visiting your website, not putting too much of a stress or burden on the website. And those were mostly respected, but not always respected by the search engines. And so in some regards, what you’re talking about is very similar to that.
You’re saying, hey, don’t use this material for training purposes.
You’re saying, you know, Yeah, essentially, this is my private property. I’m not giving you permission. And yeah, you can do that. You certainly can do that. If you think about it from a copyright perspective, the idea of copyright is that you own this property and you can choose who can use it. So here you’re saying, I own this property and I’m choosing not to allow AIs to copy it. So if you discover later that they copied it with a permission, you can charge them a fee, force them to pay you that fee. OK. That’s what I was hoping for.
But that’s like the robot.
Web sites, you got robot.txt.
Yeah. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, that’s exactly what I was saying. Yeah, I think it’s more of a guideline than an actual report. You know what I mean?
I don’t know that it had like legal language in it as far as the penalty if you didn’t score that. Well, it wouldn’t have to really, because if you think about it, what does a search engine do? A search engine copies your website.
It downloads a copy of the website. Right.
So technically, if you copy without permission, that’s called violating copyright. Right. So you wouldn’t have to put anything in there in terms of legal penalty other than to say you don’t have permission to visit the rest of this website, which is what robot.txt does.
But I don’t know if there’s actually been any lawsuits on it.
That and all. Quick question with this.
So if you do not have a disclaimer and the AI references his project, does it actually have to say It’s referencing that project to honor the GPL.
About GPL, you’re supposed to say all the GPL stuff that you use. Yeah, so there’s definitely a possibility of violating the open source license and copying it without, yeah. If you copy it the wrong way, you can definitely violate open source software licenses. Sure. Yeah, it’s just like violating copyright in general.
Same thing. You want to let me make you the host? I can still sell you. Just give me. Oh, you want to be sure to stay in here?
Yeah, I can still sell you.
Oracle, Sells, Java, and they also have Linux. Red Hat, so now it’s open source. Basically, what a license is. It’s a way in which someone can use them.
You could buy my car. I can also let you borrow my car and drive it. They’re both licensed. You are live now. Just as a quickie here, I came across this article just two days ago.
Check out this number here.
Yeah, okay.
Yeah, it’s happening on the last channel. That sounds like a good wage. Not too bad. Who’s heard of SDT?
And you know what that stands for?
What was your question?
SDB, have you heard of that?
Software to non-document.
Software-driven design?
Yeah. Closer to close. Okay. Specification-driven design.
That’s the new thing then.
If you go on YouTube and start looking that up in context of AI, what you will find is things like the prompt orchestration you mark down language, you just get all these out of here.
I have a pet peeve, by the way.
I do have one.
I hate the word orchestration when it comes to agents, but that’s just… Well, I guess it’s like an orchestra. I love the idea of being a conductor up front, but my agents aren’t quite that skilled as like actual players in an orchestra. Just think of it as like an elementary school orchestra. We’ll get better over time.
It’s kind of like herding cats when you’re trying to get your AIs all in line. I have a herd of cats. I know what it’s like. This is all these players, I mean all the big players right here, okay. Cairo, I think, I’m trying to think that’s anthropic, I’m not sure. Microsoft, Pomo, we’ve got GoToLab, I think this is Cloud Code. And everything, it goes to MD files for cloud code.
Everything goes to XML that they devised for the pommel.
These guys, I’m not sure what they use because I haven’t run that stuff yet.
Speckkit is another one that maybe that’s Anthrope, but I get them mixed up. Maybe that’s Gemini. But they’re all doing the same thing.
When I looked at all of these cross support, they’re doing almost the exact same thing. And what is that thing they’re doing?
They’re creating specifications.
And I’ll show you what one of them looks like.
This is a sample from Paul.
It’s basically tagged props, right? So just like if you’ve dug a little bit into some of the training. and fine tuning of AIs and stuff, you feed it little JSON scripts with the prompt and the completion tagged in there.
Well, this takes another step further when you’re prompting the AI, so you give a task.
Well, the interesting thing is that we’re seeing conditions show up.
Instead of just asking, you know, make me a snake game, you can start now specifying layers below that.
like in this case, if somebody comes in and they’re a premium user, then your webpage will say one thing versus something else versus something else.
So why is this important?
It’s because when there’s a couple of good videos out there now, I wish I would have had the link to one of them, but you can email me or whatever if you want to find something. Just start Googling on YouTube. There’s one I think called prompt engineering is dead, something like that. It will get you to a series of videos where they’re basically showing you why it’s dead. It’s basically the premises, you take your prompt, you give it to the AI, you get an output. When you finally get the output you want, you throw away the prompt, which gets you in trouble with that. But what they’re saying now is you don’t want to throw away the prompt. The prompt is now the new code.
Yeah, the prompt becomes that.
The artifact from the AI, the code or whatever, the picture, whatever.
That’s just like the binary.
You want that because that’s what you’re going to use. But you want to keep the source code, which is the prompt. And now what we’re seeing is a new level source code here.
So it’s almost like.
By prompting with a structure, you can make up your own language as you go.
So if you’re in a particular domain, you can just start building prompts and libraries of prompts in that domain and reusing them and sharing them with other engineers.
So what I want to do is I want to show you a quick demo of what this does for you.
start a new chat.
If I just take this, this logic one, toss it straight up, nothing but whatever systems prompts they have built in.
And I dropped it in there.
And it sees what I gave it was a specification.
So it doesn’t really know what I want to do with that spec because the spec is still generic.
So I’m going to say, I’m just going to say, make a UPi for this.
It should figure out I want to use HTML. And it should generate basically, well, actually, it’s doing it in different language now. So I may want to specify make a UPi and the HTML on this one.
Start it up.
OK, so one of the tags is not in that. Specification was the language to use.
You could put that in there as well. Then you wouldn’t have to give it anything else. So if you think about it, this is the new way that coding seems to be moving.
For now, the AIs, the LLMs are like your CP or your compiler, basically.
Yeah, I’ve been thinking about like a style guide, like I need a style guide that I’m going to give the AI so he knows how I really want to do stuff. Yeah, it’s a lot like that. The thing I like about the pod, I came across Pommel Friday last week and what I liked about it was you can include other files into your file.
So if you’ve got boilerblade stuff that you use all the time, you know, hey, If you know the whole kind of don’t make up your own answers if you need information ask me know all this stuff that you can just include and Some of the systems we use a lot If you go look at their system prompt and how many pages worth of oh my gosh And you can now decompose it and put it in a structured form and actually I don’t know it seems seems Yeah, it works.
I lost a little time.
Here’s what just came out of this.
Okay, so this is just one rendition of it, the more you, you know, and then there’s just welcome back. There’s an app and there’s a premium, which is basically what that logic was saying, right? I miss this. Did you save that as a file or is that running in the generator?
It’s running this.
Yeah, it’s an artifact. It’s an artifact running in, on the cloud, in the cloud. Although I, I don’t know about time, but I did build a little bit of Python that uses the API that grabs it and puts it in a file, pops it up for you, but it’s basically the same output.
Okay, so take a look at this other one here.
This is where you’ve got loops.
Okay, iteration is not something you normally think of when you’re prompting an AI right now, but it is important if you’re going to have a detailed specification.
Like say you’re trying to build a really complicated system or a complicated UI That you don’t know what the data coming in is and here it’s gonna it’s gonna loop over all the product data in this case So if we go if we go and maybe start a new one Give it give us the loop And it’s probably I probably should have told it what language to use You’re going to say use html.
That looks like it probably might have figured out what you want to say. If there’s a template that uses loops, let me create an interactive version of the sales summary.
What we’ll just let it continue.
Sometimes you have to help it. Sometimes you have to include that as a part of the specification. These are just straight out of the Microsoft repo. that I grabbed these, so I didn’t really change them. OK, so there’s what it did. And so this basically does what was going on in this. So you’ve got three things. If we added more data or sent more data in.
And just like you mentioned, Jay, you can connect to other things.
with pommel you can connect to mcp servers you can connect to databases you can you can connect to anything you can include other stuff you can do it conditionally you can do it iteratively okay and you can modulate everything so think about the power of that we’re just scratching the surface with this kind of stuff um so when i saw this i go whoa okay this is pretty powerful um The other things I showed you like cloud code does office the same exact thing, but they use MD files like the cloud MD file, which is your so your master, right? You’ve used them Okay, and and there and then those other guys I hadn’t Done a deep dive in on but they look pretty much going the same direction The reason I get excited about this Is because of these kind of funs that I’ve worked with for many a year Let me go ahead and just show you what it does using these other alternative files in each app. Here I have a thing called logic and a thing called loot.
These were actually parsed out of these guys, right?
I’m gonna put logic in here or on it. And again, it may ask me, what?
Okay, so it’s what it’s seen as these these are by the way these I’ll show you what these look like These files are actually flowcharts of what we just saw all Okay, okay, so it’s it’s asking me those questions. Let’s make the hdmi So ask them to do basically this thing again Can I ask a question? Yes. So I noticed like, you know, both times in doing your example, you paste in like PDF first or the document first.
Is that generally the pattern you’re following?
Are you generally following a pattern?
Will you tell the EAI what you’re about to paste in?
Well, I’ve been doing both.
I pasted in, I hit, look at it, give me some feedback.
And in one case, it jumped the gun to making an HTML.
The other case asked me, what do you want to do with this? I see it’s, it notices it’s a specification.
It just doesn’t know what language it should use or what it should do with it.
At the moment, unless I tell it, but sometimes it jumps the gun and just doesn’t.
Okay, so here’s, here’s the, here’s the logic guy, his administrator, premium user, generate greeting. What it’s doing down here, I believe.
Yeah, I don’t know if it’s I don’t need a mistake with the colors because I don’t see anything there unless it’s down here in this little text box There is a experimental rust-based original printer for environments without JavaScript or Python interpreters That’s cool As long as I get to skip the emotional trauma of cascading style sheets. No, now you have cascading, palpable files. It’s not showing the grating.
So, you know, you know, AIs that you have to sometimes get them to come along a little better and make the changes.
So here we go.
Now it’s doing the grating.
Before I think it was all white down here, now we see. Okay.
I’m going to jump to, I’m going to just jump from that one to get the idea to a different one.
Here’s this loop, the same thing that Pommel did a second ago.
Again, I’m just giving it just the specification.
that’s looking at the SPAC, trying to analyze the SPAC, and trying to figure out what does he want to do with this. I probably could have, and I’ve had experimented by putting the prompt in the SPAC, and it’ll usually comply, so make a UI HTML.
I wonder if a custom cloud instruction.
that you put it in for whenever I paste an EFC file, you should automatically make it.
It should follow that.
Well, that’s a good question because the little widget or piece of code I made that used the API goes directly to the AI.
It grabs that whole thing and sends the whole EFC file.
But sometimes I’ll put the extension I want, like I’ll do .html.vfc.
and every single time it figures out what to do with this.
So if I put a .py, .vfc, it’ll create a Python form.
So it’s kind of built in.
Here’s what it just did.
Again, this almost looks exactly like what the pommel had done. Let me show you what these actually look like.
They’re fully graphical.
They’re not text. You edit them, and this is my open source tool, so there’s a plug for my repo. But you basically edit these guys, and you can change them however you want. And here’s the, that’s the iterative action, and here’s the logic. And then here’s one, this one has the HTML extension on it.
I don’t know if you can read that, but this one actually is to make like a little little gooey with some balls that bounce around.
And I’m just going to run that one to see what it does.
One general question I have, what is POML seems pretty similar to XML?
So what are the benefits of POML over like an XML or even just kind of using Markdown in a similar way?
I think they’re all the same.
I’m literally starting to see that there’s a trend. They seem to be equivalent, that the AIs figure out whatever you want.
Now, granted, if Microsoft’s making POML, they’re going to send the specification for POML language.
They’re going to burn it into the LLO.
But it doesn’t even seem to have to do that, because it’s never burned in this BFC spec. And yet, here’s this.
Let’s go ahead and do that one without telling you anything and see if it gets it.
It has the HTML tag in there.
So it’s just going right ahead to build the HTML for that. And if you look what this inspect does, I can zoom in a little on it.
It just shows some balls and repeats forever for all the balls.
when you click it, it bounces five times. And then this is a function, show the balls, which gets called up here.
And so here’s the thing now.
Seems to be having a lot of trouble with color. I can’t see the balls.
Yeah. Or from this format, do you know?
Okay, there’s a softball show up there.
show you the balls for a fraction of a second, like you didn’t say something. Yeah, it’s it’s it’s iterating and then now it doesn’t show it. So there’s something odd happening here with the colors. Try it again and see.
So, yeah, that is one issue, though, that all our alarms are probably always going to have is every time you run a spec, you’re going to get a slightly different answer.
Will you do one experiment for me real quick?
Sure. Paste in this BFC in a new chat and say, please create me a properly formatted prompt to instruct a new instance of your model to create this in HTML.
Okay, so we’ll try that. That’s probably going to be the last.
We probably need to wrap up. So that’ll probably be the last. Yeah, we’ll do that as the last experiment. So this failed because it’s not bouncing balls. We’ll try that.
We’ll start with a brand new chat. Okay, and then we’re going to give it this one. And then we’re going to say, make a prompt.
Do we want to use the PLML? You can just tell it, you know, convert this file into a proper HTML prompt.
Make this file a proper HTML UI or just output. for app.
Okay.
I’d be curious if you just pasted that to a new instance.
Sorry if you told it just to run in if it would. Where’s the, where’s the artifact? Age, love, so I’ll see, I’ll use the up and down arrow, hands.
There’s a copy.
Very bottom. Yep, right on. But usually it goes the artifact on the right.
Yeah, Claude’s not really working to all right now. This kind of reminds me when you refactor code or you’ve got a program and you think it’s a mess you go take this program and define everything it does.
Make me a prompt to make that program again and let it go do it the way it wants to go do it. So you don’t have to rewrite it. Well, it looks like they’re bouncing this way, like we’re looking down.
How much code I end up just work?
The first time I tried it, you got something that looked low like this, but they actually bounced properly.
And so, so that is an issue.
But you’re going to get that issue with any kind of prompting language. So this, I think is. I didn’t ask it to show me the prompt that way. There it is.
Here’s the prompt, create a ball bouncing application.
So there’s where it created the prompt.
Kind of like traditional prompts are right now, right?
Where you’ve got several sections and you’ve kind of done it, everything else.
And that’s really where a lot of this spec driven development is starting to go is, let’s just work with the specification that we can reuse.
and keep modifying it, and if something doesn’t come out right, either look at the specification, see if you can make it tighter or run it again on a different LLL, because you’ll get different results. Ideally, we should get to the point where we can have one specification that can generate whatever we want.
So, if it has all the business logic, say, okay, make me an app for this logic, or make me a website for this object, this logic.
And that’s pretty cool way to go.
We’re starting to go that way. I don’t know how far we are now.
But it’s kind of getting there.
What I discovered was that the graphical representation here is just much easier. And it’s almost like you’re writing your own language.
Whereas if you’re in the pommel space, you’re just writing another language, which is OK for people that code but not for people that don’t code. And so I’ve been teaching some of this to people that don’t code and they love it because they can just lock down a diagram that just captures their logic, give it to the AI.
And if it’s not right, they keep editing that logic until they get it right because they’ll never be able to code. So it basically replaces code, which is weird.
Yeah, it’s so it’s just human language and in a chart but now or in a XML or whatever So that’s uh, it’s about it. I’m thinking out here appreciate it If you want to know more about this, I mean I can get you some here’s a QR code to both The repo if you want to tinker with this or join my open source thing And then here’s a QR code.
I show one at a time.
This is to the repo.
Also, on Thursdays, I’ve been doing a free kind of workshop. I get that QR code.
Thank you.
We reclaimed post real quick after we put this up. This is the repo for the short code for the workshop.
And then you can always email me, Sumo.
Okay, stop sharing. Yeah, I get it back. I might be sure to think we’ll just wrap it up with that.
Thanks everybody for coming and doing extra stuff. It was really nice to not have to put together a full session of things after talking to a whole bunch of people yesterday.