Latest / North Meets South Web Podcast / OpenClaw, Arbor, and horseless carriages

Transcript

OpenClaw, Arbor, and horseless carriages · North Meets South Web Podcast

• 0:00
  -Hey. I'm Michael Dyrynda. -And I am Jake Bennett.
• 0:03
  [claps]
• 0:05
  And welcome to episode 186 of the North Meet South Web Podcast. [upbeat music]
• 0:10
  We are back on the Spindrift kick, folks. It is back and better than ever. Raspberry
• 0:16
  -lime. -Raspberry lime.
• 0:17
  -So good. -Sounds good.
• 0:18
  -Yeah. -There was-
• 0:20
  -It's a tasty one -... there's a, there's a brand of, I think
• 0:22
  it's like Swedish or something cider, Rekorderlig, and they've got like a-
• 0:26
  -Oh, okay -... strawberry and lime flavored cider-
• 0:28
  -Ooh! -... which is quite... It's those things,
• 0:30
  when I was younger, they were great, but they're so full of sugar, I can't drink
• 0:33
  -them anymore. Too old for that stuff. -Yeah, I wonder, I wonder how much sugar
• 0:36
  this one has. I don't, I don't think it actually has... It's 1 gram.
• 0:40
  -Yeah. -1 gram of sugar, and it's like, it's like
• 0:43
  supposed to be zero grams of sugar added. So it's, the ingredients are literally
• 0:47
  carbonated water, raspberry puree, and lime juice. That's it.
• 0:50
  -Nice. -So-
• 0:51
  -It's just fruit. -It should be, you know... Yeah, not too
• 0:53
  bad, not too bad.
• 0:55
  -[inhales] Oh, my word! -None of that corn syrup that you guys
• 0:56
  -like. -Okay. Yeah. Yeah, none of that, none of
• 0:59
  that corn syrup stuff. Just, just straight up raspberry puree and, you know, lime
• 1:05
  -juice, whatever. -Amazing.
• 1:06
  -Um, yes, it is- -I've got 40 minutes before my, uh, my
• 1:09
  -Claude Pro, [chuckles] my to- my, uh- -Before your-
• 1:13
  -... my limit refreshes- -Before your-
• 1:13
  So, uh, we've got 40 minutes of- [chuckles]
• 1:16
  So tell me real quick, you said Claude Pro is on sale for, like, 100 bucks or
• 1:19
  -something? -Yeah, I think the price has changed. It's-
• 1:22
  Okay
• 1:22
  ... uh, Claud- I think the Claude Max subscription is now $100 a month, and-
• 1:27
  -Hmm -... I think,
• 1:29
  -just looking at- -Rather than $200?
• 1:29
  I only looked at it very quick. Yeah, I, because there's two options now. So
• 1:33
  there's, there's, like, the free one, which gives you whatever the limit is.
• 1:36
  -Yeah. -Then there's Claude Pro, which is 20 bucks
• 1:39
  -a month, and it gives you- -Okay
• 1:40
  ... five times the limit of the free one. And then there's Claude Max,
• 1:45
  which gives you the option now of either five times or 20 times the limit of-
• 1:51
  -Yeah -... whatever. So.
• 1:52
  Yeah, I think that I've gotten... Let me see here. Adjust usage. 'Cause I have,
• 1:57
  right now, I have, hmm, I have the Pro plan, but I have the 5X more
• 2:04
  -usage of the Pro plan. -Yeah.
• 2:05
  -So I'm paying 100 bucks- -Right
• 2:07
  ... for the 5X Pro, and then there's Max, which is, like, another... Yeah, so it's
• 2:12
  like 50% off, it says. It's usually, it's usually 400, now it's 200, I think is what
• 2:16
  -it is. -Mm.
• 2:17
  -Maybe? -Yep.
• 2:18
  -I don't know. Yeah. -Yeah, I saw that.
• 2:19
  -Yeah. -I just went... I don't remember why I went
• 2:21
  to look, but I went to look the other day and saw that it was
• 2:25
  only $200 a year, so I signed up finally for, for Claude. So,
• 2:30
  -but I've, I've been using- -$200 a month, not a year, right? Not a
• 2:33
  -year. Not a- -No, no
• 2:33
  -... not a year, a month. -Claude, Claude, Claude Pro is $200 a
• 2:36
  -month. Claude Max- -A month. Yeah, yeah.
• 2:40
  -Claude Pro, $200 a year. Claude- -A year?
• 2:45
  -... Max, yes. -Oh, yeah, yeah, yeah.
• 2:47
  -So Claude- -Sorry, sorry. Yes.
• 2:48
  -Yeah. -Yes.
• 2:48
  -So Claude Pro- -Sorry. Thank you
• 2:49
  -... is like $20 a month or $17 a month- -Yeah, yes
• 2:52
  -... if you pay for a year. -Thank you.
• 2:52
  -Yeah. -Yeah, yeah.
• 2:53
  And then Claude Max is $100 a month, or I guess $200-
• 2:56
  -Got it -... is still an option there. Yeah, yeah.
• 2:58
  -So I'm not, I'm not, uh- -Mm
• 2:59
  -... I'm not down to pay $100 a month for- -Okay
• 3:03
  -... Claude Max. -Yeah.
• 3:03
  But I, I ponied up for the, the Pro annual. We'll see how that goes.
• 3:08
  -Yeah. -So.
• 3:08
  I, I went, I went with the $100 plan, so I'm on the $100 a month plan, and,
• 3:13
  [inhales] uh, it's great. I don't really have to care so much about my token usage.
• 3:17
  Just us- use Opus for everything and then let it roll, let it roll. Um, I think I
• 3:22
  -have hit token limits once, maybe. -Mm-hmm.
• 3:24
  But other than that, I just let it roll. So-
• 3:27
  -So, I mean- -I've, I've only switched to Sonnet, like,
• 3:29
  -once, but- -Mm-hmm. Yeah, so I... Because... I mean,
• 3:33
  for work, we've got team plans, and, and those are fine, and so I, I still tend to
• 3:38
  prefer
• 3:41
  Opus for planning.
• 3:43
  -Mm-hmm. -'Cause, you know, it's got the deep
• 3:45
  reasoning, it thinks about things, it goes and figures stuff out. And then I use
• 3:48
  Sonnet for
• 3:50
  the actual implementation, 'cause it doesn't need to think. It just takes
• 3:53
  whatever the plan is and runs with it, so-
• 3:54
  Take the plan and go do it. Yeah, yeah, exactly.
• 3:56
  -Yeah. -Yep.
• 3:56
  -Um, but I had been using- -That's a good, I think a good call
• 3:59
  -... GLM, because that was- -Okay
• 4:01
  ... $200 a year, and that seemed to work fairly well. And then I think-
• 4:04
  -Is that Zed AI? -Yeah, Zed AI. Yep. So that one seemed to
• 4:09
  have been working okay, but I thought I would just give it a spin because I was
• 4:12
  looking... Minimax is quite good. I, I played with that on the open code, Zen,
• 4:16
  when they had that for free. I played with Kimi as well, which, which seems really
• 4:20
  good. Um, [chuckles] GLM seems to hallucinate into Chinese. Like, it-
• 4:26
  -Okay. Oh, yeah -... it shows you the thinking, and then it
• 4:27
  just starts spitting out, um,
• 4:30
  Chinese characters or Mandarin or whatever, whatever it is. So that, you
• 4:34
  know, it's fine. It's just funny to see that every now and then. But, uh-
• 4:38
  -Yeah -... I've, I've been, like, multi-wielding.
• 4:40
  I've been, I've been doing, um, Laracon things, I've been doing work things. I've
• 4:44
  been working on Arbor, which, uh, is in a, like-
• 4:47
  -Yeah -... very close-
• 4:48
  -We should talk about Arbor -... to, to usage state. I've been, I've
• 4:52
  been using
• 4:53
  Claude, I've been using... Well, I've been using Opus and Sonnet. I've been using
• 4:57
  GLM, I've been using Kimi, I've been using AMP, all of these things. I'm, I'm right
• 5:01
  into it now. Um, still haven't installed OpenClau or MulkBot or Claude Bot,
• 5:06
  -whatever- -Yeah, I, I want
• 5:07
  -... it is today. -Okay, I want to talk about OpenClau as
• 5:09
  -well. So like, okay- -Yeah
• 5:10
  ... so I want to talk about Arbor, I want to talk about OpenClau. Um-
• 5:12
  -We have to be careful about turning- -And then I want to talk about-
• 5:14
  ... this into an AI show as well. [chuckles]
• 5:15
  Well, I mean, so here's the deal. It's like, that's just kind of like what's
• 5:17
  -happening right now. -Mm-hmm.
• 5:18
  It doesn't have to be an A... It's not an AI show. It's just like-
• 5:20
  -Mm -... it's, it's kind of like the most
• 5:22
  interesting thing that is going on right now.
• 5:25
  -Mm-hmm. -And so I want to talk... Uh, we can
• 5:28
  briefly touch on those things, and then I can talk about a couple other things if
• 5:30
  you want to. Um, but then I also want to talk about horseless carriages. Have we
• 5:36
  -talked about this before? -Horseless carriages.
• 5:37
  -Horseless carriages. -We have not.
• 5:39
  -Okay, I want to talk about that too. -That's a tease-
• 5:41
  -Okay, so let's- -... for later in the show. [chuckles]
• 5:43
  It is. It is indeed. Okay, so couple things here, um, OpenClau.
• 5:48
  -Mm. -So do you want to talk about OpenClau or
• 5:49
  -Arbor first? -Let's talk about OpenClau. Let's not talk
• 5:53
  -about me. -Okay. Yeah, so I saw Taylor post the other
• 5:56
  day that like: "Hey, Forge, you can now provision an OpenClau thing," whatever.
• 6:00
  -Mm-hmm. -So what is that? I don't even know what
• 6:02
  that is. I, I, I was gonna go look it up, and then I was like, "Eh," I didn't have
• 6:06
  time. I was, I was working.... working that day.
• 6:09
  -So- -So anyway, tell me about OpenClaw.
• 6:11
  OpenClaw was previously, um,
• 6:17
  previously MaltBot, which was previously Claude Bot.
• 6:23
  -Okay. -So it's the... I think the idea behind
• 6:26
  this is that it's, it uses whatever model
• 6:31
  -under the covers. So you, you set it up- -Right
• 6:33
  ... as like a tele- it uses Telegram as a chat interface, or you can connect it to-
• 6:37
  -Oh, interesting -... Slack, WhatsApp, whatever.
• 6:37
  -Telegram, okay. -And so it's just-
• 6:39
  -Sure -... there is a bot, and then it uses, um,
• 6:43
  Claude, uh, you know, Opus or Sonnet or GLM or Kiwi. You can connect it to any
• 6:48
  -provider- -Sure. Okay, okay
• 6:49
  ... or model under the hood that it supports, and then it, and it uses that.
• 6:52
  -Bring your own key? -Bring your own key kind of thing. Um, and
• 6:55
  so you host it on a VPS, in a Docker container on your Mac, if you're brave.
• 7:00
  There was, like, a, a run a couple of weeks ago, like, everyone went out and
• 7:03
  bought Mac Minis to, to host their, um, Claude Bot instances. Um, and, and the
• 7:09
  idea is that it's, it's like a, a personal assistant that-
• 7:13
  -Okay -... you ask it to do things. You get it to
• 7:15
  schedule tasks, you get it to do whatever. So it's like ChatGPT on
• 7:19
  -steroids, but I think the, the main- -In Telegram.
• 7:22
  -Yeah. -Like, so that's your interface.
• 7:23
  -Yeah. -You choose an interface, right?
• 7:24
  -Correct. -And so I guess Telegram is a common
• 7:25
  -interface. -Yeah. Telegram is probably the easiest
• 7:27
  one, I think, from what I've seen, to, to get up and running. Um,
• 7:32
  and
• 7:33
  it, it's basically, yes, Siri on steroids. Like, if Siri actually worked. Sorry to
• 7:37
  everyone whose
• 7:39
  -home assistants- -No, it's true
• 7:40
  -... have just triggered there, so. Um- -It's freaking true. Yeah. It's ridiculous.
• 7:43
  Uh, yeah, so you can get it to do
• 7:45
  whatever. You can... You know, people have been hooking it up to their email
• 7:47
  accounts, their calendars, reminders, whatever. Like, all of this stuff that
• 7:51
  -feels kind of scary, and I think- -Like, using MCPs?
• 7:55
  No, just, like, "Here are, here are my username and password. Here are my, you
• 7:59
  know, keys," whatever, just giving it access to these, these different tools. So
• 8:04
  it's, uh, interesting. It- a- and people go, "You know, it's self-hosted, it's all
• 8:07
  local." I think, well, the, the bot software itself, you know, the wrapper
• 8:11
  around it, is self-hosted, but it's still shelling out, you know, using the remote
• 8:17
  -models. So it's not- -Totally, yeah.
• 8:19
  It's not as self-hosted as you think. There's, there's the ability to use-
• 8:23
  -You're not running Ollama. -Yeah, there is the ability to, to host
• 8:26
  -your own models- -I mean-
• 8:27
  ... but I don't think most people are doing that.
• 8:29
  -Right. Right. Okay. -But there's, uh-
• 8:32
  -Interesting. -Yeah, there's, like, a whole-
• 8:32
  I, I would like to see... I would be interested to know kind of what people are
• 8:35
  -doing with that, yeah. -Yeah.
• 8:36
  -Um- -Jeffrey, like, Jeffrey's got some videos
• 8:38
  up on Laracasts about it. He's been talking about it. Um, Dreis has been
• 8:42
  talking about it. Like, a whole bunch of the Laravel team have been using it a lot,
• 8:47
  and, and so now, you, you know, we've got one-click deployment of it on, uh, on
• 8:50
  -Laravel VPS, so yeah. -So instead of po- instead of hosting it
• 8:55
  locally on a little, on a little MacBook Mini-
• 8:57
  -Yeah, instead of- -... or, yeah, sorry, a Mac Mini
• 8:58
  ... hosting it on your own device or, or spinning up a VPS manually and doing it,
• 9:01
  -you just, you know, Laravel VPS- -Just go deploy it
• 9:03
  ... comes up in five minutes, then, you know, OpenClaw comes up in five minutes
• 9:08
  -on, on your- -Hmm
• 9:09
  -... um, environment there, so. -I mean, it makes sense. I mean, it's like,
• 9:13
  if all you're doing is hooking it up to another location... Like, again, if you're
• 9:16
  not using Ollama, you don't really need it, the bunch of, of hardware.
• 9:21
  -No. -It's like if you're just literally taking
• 9:23
  it and using it as a place to, like, say, "Okay, my chat interface is over in
• 9:26
  -Telegram." -Mm-hmm.
• 9:26
  "I want you to use my key for this particular thing, and then I'm letting you
• 9:30
  do particular other pieces on my behalf," and just kind of keeping it all there.
• 9:35
  -Yeah, okay. Interesting. -Mm.
• 9:37
  -I can see that. -Yeah.
• 9:38
  -Hmm, cool. -I, uh-
• 9:39
  -Yeah -... I still haven't dipped in yet. I-
• 9:41
  I'll just say I'll be a late, I'll be a late adopter to that.
• 9:43
  -I'll be- -I'll figure it out someplace
• 9:43
  ... I'll be a late adopter once all the security concerns and things are figured
• 9:46
  -out. So it's, uh- -Mm, mm
• 9:48
  ... it's definitely interesting, but, I mean,
• 9:50
  my, my biggest thing with it is that, you know, when Siri ca- and I, and I don't
• 9:54
  know, I'm gonna prefix this with, I don't know if I don't use she who shall not be
• 10:01
  -named, because they're not very good- -Yeah
• 10:05
  ... or because, like, I just have no use for it. You know, the, the only thing I
• 10:09
  really use it for is setting a timer, and even that can sometimes be hidden.
• 10:11
  -Timers- -Like, I-
• 10:12
  -... reminders- -Yeah
• 10:13
  -... calendars- -Yeah
• 10:14
  ... sending text messages, making phone calls, right?
• 10:17
  -Yeah. -Um, use it for all those things.
• 10:19
  Like, stuff that it does well enough that it's, you know... But
• 10:23
  g- getting it to do all this other stuff, I just... I'm interested to see how it
• 10:28
  shakes out, but I'm not interested enough to actually use it just yet.
• 10:32
  Okay, so, so l- let's, let's play the game here real quick, okay? Hypothetically, I
• 10:38
  -wanna make sure- -Hmm
• 10:38
  ... I'm understanding this correctly, all right? The reason why I asked about the
• 10:41
  MCP thing is 'cause that seems like the natural play here. Like, isn't an MCP
• 10:46
  basically a new... L- like, so if I have a, um, I don't know, a Gmail MCP, right?
• 10:52
  -Mm-hmm. -That's installed.
• 10:53
  -Yeah. -And then I authenticate that with my
• 10:55
  -Gmail, right? -Mm-hmm.
• 10:57
  It now has access probably to my calendar, and to my emails, and to my chats-
• 11:01
  -Yeah -... whatever.
• 11:01
  -Sure. -Right?
• 11:02
  -Mm-hmm. -And it is exposing to my LLM, "Here are
• 11:05
  the tools that are available to you through this MCP." Yes?
• 11:08
  Yeah. That would be the same way of doing it.
• 11:09
  Why would I... That's, that's how I would do it, right? I mean, like, what are we
• 11:12
  talking about? They're putting their username and password into, like,
• 11:14
  something else. It's gotta be an MCP, right? For OpenClaw.
• 11:17
  Well, does Gmail, does, does Gmail have an MCP?
• 11:21
  -I don't know. -Like-
• 11:21
  I don't know. I would assume that they would.
• 11:23
  -Oh, yeah, it does. -Google?
• 11:24
  -End of- -Yeah
• 11:25
  ... December, they, they, they enabled it, like-
• 11:27
  -Okay. -But again-
• 11:28
  -So, so, yes -... using an MCP only gives you whatever
• 11:31
  -the provider gives you. -So what is this doing? Like, this, what
• 11:35
  are they... It's OpenClaw, like, is going on the browser for you and doing UI
• 11:38
  -interactions? -Yeah.
• 11:38
  -Is the idea? -People... I've seen people do, do all
• 11:40
  -kinds of things, um- -Okay
• 11:42
  ... where it's going into uncharted territories, and it's, you know, clicking
• 11:46
  through... It's using Playwright's MCP to go and click through browsers-
• 11:49
  -Oh, interesting -... like, websites-
• 11:50
  -Oh, okay -... and doing all this kind of stuff, so.
• 11:52
  You'd have to program them all up front, right? It's the idea.
• 11:55
  Yeah, you'd... Well, you tell it what you want it to do, and it goes and figures it
• 11:57
  -out. Um- -Okay, so-
• 11:58
  But I think the, the, the, the, the thing that separates it from something like just
• 12:03
  using ChatGPT is that it, it uses its own memory store. So it's got, like, a memory
• 12:07
  file and a soul file, and all this kind of stuff.... and, and so it has
• 12:12
  -contextual- -You can, you can-
• 12:13
  --historical information. -You have full, you have full control over
• 12:16
  -the prompt, probably- -Yeah.
• 12:18
  -Right? -Yeah.
• 12:18
  -Like- -Yeah, you tell it what to go-
• 12:20
  -... interesting --and write in there. You give it, you,
• 12:21
  like, you give it it the personality you want it to have. Do you want it to be
• 12:23
  -sarcastic? -Yeah.
• 12:24
  Do you want it to have... You know, I've seen, uh, you know, I, I mentioned Kitze.
• 12:27
  He started this, like, Tinkerer Club, which you can pay for access to. It's a
• 12:30
  Discord thing. But he, he's got, like, 10 different
• 12:34
  OpenClaw instances running, assuming personas of, like, Dr. Cox from Scrubs,
• 12:39
  -and, and, like- -Cool. Okay
• 12:40
  ... Chandler from Friends, and all of this kind of stuff. He's got like a- he's got
• 12:43
  Saul Goodman there acting as his lawyer. So you give all these different instances
• 12:48
  of OpenClaw, their own personas that are responsible for their own things, and you
• 12:51
  give them access to their own things. Um, so you've got, like, an-
• 12:54
  -Yeah, okay -... accountant that has access to all of
• 12:56
  that kind of stuff. So
• 12:57
  it's, it's definitely, like, he's got a whole army of them. But the, the main
• 13:01
  thing is, like, you can use ChatGPT and do all this stuff, but each ChatGPT-
• 13:04
  -Well, sort of -... conversation is kind of isolated,
• 13:06
  -right? -Yeah, right. Well, and also, you can't
• 13:09
  -install MCPs on ChatGPT. -No.
• 13:12
  Right? You can only do what ChatGPT is allowing you to do.
• 13:16
  -Right. -Now, if they define or create a new tool,
• 13:18
  great, and I can now use that tool. But because it's not a local instance, I don't
• 13:22
  have control over my own tool chain. I only have access to what they give me,
• 13:26
  -right? -Yeah.
• 13:27
  -Okay. -Yeah.
• 13:27
  So th- this is, this is the sort of thing that I'm, that I'm... I think I understand
• 13:33
  it correctly, right? But this feels like the magic that MCPs will bring, which is,
• 13:41
  I have access to all... I have all these services that I use that are on the web,
• 13:44
  right? Um,
• 13:46
  everything's now accessible through the web. I mean, it used to be like, there
• 13:48
  were so many things that were stored locally, right? And then we... Remember we
• 13:50
  had these netbooks for a while, where it was like there was no operating system,
• 13:54
  basically. It was very thin, and just, like, the idea was like, you're going to
• 13:56
  do everything in the web. And then it was like-
• 13:57
  -Mm -... well, this is a crazy idea. And then
• 13:59
  it was a Chromebooks came out, and now it's like, okay-
• 14:00
  -Yeah -... everything's actually on the web.
• 14:01
  Like, there is no OS. It is all the web, right? And so great, now everything's on
• 14:05
  the web. Wonderful. Um, but what if I want to interact with those things, not
• 14:10
  through a UI, but we want to do it programmatically? Great, now we have APIs,
• 14:13
  we have GraphQL, blah, blah, blah. MCP, I don't want to have to write every
• 14:16
  connector. MCP is the promise that, like, you just tell your LLM, you have this
• 14:20
  tool, it will figure out what thing it needs to use, and it knows how to talk to
• 14:24
  us, and then it'll hand- handle it on your behalf. So what I can do now is I can
• 14:28
  say, "Hey, my local instance of OpenClaw, you can interact with me on my Gmail. You
• 14:33
  can interact with... You can interact for me on my calendar. You can interact for me
• 14:37
  through Yelp. You can leave reviews, you can make reservations, you can an-
• 14:41
  interact as me in GitHub. You can interact as me, blah, blah, blah. Here are the
• 14:45
  -MCPs that I'm giving you access to." Now- -Mm
• 14:47
  ... my OpenClaw instance has, when I ask it to do something, it's going to look at
• 14:52
  its tool chain and say, "Are any of these things relevant to what I'm trying to
• 14:54
  accomplish?" "Yes, Jake asked me to make a calendar reminder for whatever."
• 14:58
  -Mm. -"Okay, great. I'm gonna go talk to that
• 14:59
  MCP. I'm gonna go do that thing." So, you know, the previous iterations of, like,
• 15:03
  same thing, like, you have she who shall be not, not be named, right? The girl in
• 15:07
  the box, Siri, Alexa, all the things, right?
• 15:11
  -Mm. -Um, Google, all those. But the limitation
• 15:13
  that they had, again, was like, you... There was no way to integrate that unless
• 15:17
  you were using the service. So if you're using Apple, fine, you can use their
• 15:21
  -calendar reminders. -Mm.
• 15:22
  If you use Google, fine, you can use their stuff. But there was no way to do, like,
• 15:24
  cross-stuff or to do Yelp stuff with- using that assistant. But now you can,
• 15:30
  right? You can roll your own. So, like, the idea of, like, JARVIS that Tony Stark
• 15:35
  -had- -Mm
• 15:36
  ... where it could just do things as him or for him, was previously thought to be
• 15:40
  like, "Oh, that's, you know, that could possibly exist at some point," but MCPs is
• 15:44
  what, to me, unlocks that.
• 15:47
  -Yes. -I, I think. You know what I mean?
• 15:48
  But the, but the MCP still relies on the provider offering it.
• 15:53
  -It does. -And, and, and I suppose-
• 15:53
  -It does -... it does.
• 15:54
  But, like, I rely on the provider giving me those tools anyway.
• 15:58
  -Right. -Like, if it's not-
• 15:59
  -So this is- -... exposed through an API, there's no way
• 16:00
  it's gonna be exposed through an MCP. So it's like, uh, is, you know, the full
• 16:04
  functionality is likely gonna be available at some point through that MCP-
• 16:09
  -Mm -... um, which means I don't have to write
• 16:11
  all the tooling for it. It's done. Like, the LLM-
• 16:13
  -Yes -... figures out the tooling, right?
• 16:15
  Yeah. So the... Well, this is then where the difference comes from. So they've got
• 16:18
  the notion of ClawHub, which is skills and plugins and things like that, that you
• 16:22
  -can- -Yes, yes
• 16:23
  ... imbue with, um, different things. So there is, like, a CalDAV for calendar, and
• 16:28
  it'll go off and, and then it knows how to deal with calendar. But
• 16:33
  it's got browser control, so
• 16:35
  OpenClaw can, can browse the web, it can fill forms, it can extract data from
• 16:40
  -anything. -That's super cool.
• 16:41
  -So it doesn't, it doesn't- -Yeah
• 16:42
  ... need a, like, a Google Skill might encapsulate this.
• 16:44
  -Doesn't need a plugin, yeah. -But it's not talking to an MCP necessarily
• 16:48
  or, um, you know, using an API. It is going, like-
• 16:52
  -It's using the open route -... it's opening the browser.
• 16:53
  -Yeah. -Yeah, it's u- it's, it's opening a
• 16:55
  browser, um, it's
• 16:57
  putting in your username and password, and it's going and logging in, and it's
• 17:00
  reading your emails, and it's doing things in there, so...
• 17:03
  -Super cool. -It's, um-
• 17:04
  -Yeah, you have the full power- -Yeah, the, right
• 17:05
  ... of the web a- available to you at your fingertips.
• 17:07
  Yep, but it, you know, it can, it can post to Twitter, it can watch and transcribe
• 17:12
  YouTube videos for you. It can do all kinds of different stuff, which is
• 17:16
  -interesting. Um- -Yeah.
• 17:18
  But yeah, I'll let, I'll let everyone else kind of figure it out and, and bring
• 17:21
  back, you know... It's interesting watching Jeffrey's journey. He's tweeting
• 17:25
  about it, and he, he had it set up like a daily vocabulary challenge, 'cause he
• 17:30
  wants to learn, you know, new words and things like that. So
• 17:33
  very good. So he told it, "Every day at sometime, send me a new word of the day
• 17:38
  and tell me about it and how I would use it, and what- whatever." And it decided
• 17:41
  one day to just text his wife instead. You know, he sent the message-
• 17:44
  -Oh, my gosh -... to Jeffrey's wife.
• 17:46
  -[chuckles] -And, and he goes: "Why did you do that?"
• 17:48
  You know. He asked... He asked... He set up a, a scheduled task to go and
• 17:53
  look on Reddit and Hacker News and Google and whatever else, and look for people
• 17:57
  talking about Laracast and, and seeing if there's, like, anything negative out there
• 18:01
  and, and bringing back, like, a summary. And he, he tweeted yesterday that the
• 18:05
  summary that came back
• 18:07
  had, like, some story that OpenClaw had just made up.... but it didn't exist-
• 18:12
  -Completely fabricated. -There was no link to it. Completely
• 18:14
  fabricated it, because it... You know, these things are all designed to kind of
• 18:18
  be useful. You know, that's their, their purpose in, in existence, is to be useful,
• 18:24
  and so it didn't wanna tell Jeffrey, "I've got nothing to tell you." It just
• 18:28
  made something up, so it looked like it did something.
• 18:28
  'Cause he sort of investigated it. He basically, like, did like, you know-
• 18:32
  -Yeah -... a deposition with this thing. It was
• 18:34
  -like, "Why did you do that?" -Mm-hmm.
• 18:35
  And it's like: "Well, because I wanted to be useful, and, like, I already... You
• 18:39
  -know, I was trying to- -Mm
• 18:40
  ... regain trust from, like, this mess-up I made yesterday."
• 18:43
  -Mm. -"And so I didn't wanna, like, give you a
• 18:44
  bad news that I don't actually have any information, so I just made something up
• 18:47
  -to make you happy." -Yeah.
• 18:48
  -And yeah, that's crazy. -Yeah, so it's, uh, it's cra... It's--
• 18:51
  yeah, it's definitely crazy. I know, I know that, um, Simon Frasciootis has been
• 18:56
  using it, uh, to build out
• 18:59
  a tool to help him. So he's filming AAU basketball games up on the North Coast,
• 19:03
  where he lives, and so he's taking all of the clips, throwing it into a folder, and
• 19:08
  he, he had OpenClaw build a native Swift app that would go through, and it would
• 19:13
  pluck, um, you know, stills from the thing that you can scrub through to see, you
• 19:18
  know, what happened. It then looks at the audio and, and figures out if there's any
• 19:22
  peaks in there, to kind of figure out, was there some play that got everyone
• 19:27
  excited? Was there... You know, and then it goes and figures out, okay, groups it
• 19:31
  by team, looks at the colour of the, of the uniforms. Okay, these are orange, this
• 19:34
  is blue, and, and does all this kind of stuff for him. Then he's like, he's built
• 19:37
  this. He's just asked OpenClaw to go out and do it, and it's gone out and done it.
• 19:41
  So there, there's definitely, um,
• 19:44
  time and place to do it. I think having it in Telegram, that you could just message
• 19:48
  it from anywhere, and off it goes and, and does whatever, is kind of cool. He's
• 19:51
  given it his own, like, gi- he's given OpenClaw its own
• 19:57
  GitHub user account, and it goes and just creates repositories-
• 20:01
  -That's cool -... and shares all this kind of stuff in
• 20:03
  there. So it's isolated and can't be disruptive to Simon's things, but it has
• 20:08
  access to, you know, push code and, and have it reviewed and things like that. So
• 20:12
  there are, there are definitely lots of, uh, interesting s- things that can be done
• 20:16
  with it, but lots of, uh,
• 20:19
  interesting, as in, I don't know that I would do that, kind of things that people
• 20:23
  are doing with it. So I, I'll let other people figure out the, the rough edges in
• 20:27
  the meantime, but there's... You know, for me, being fairly late to this, but at
• 20:31
  least on board now, um, I'm, I'm okay opening up
• 20:35
  Ghosty and firing up OpenCode and just typing in there, like, "Go and do this
• 20:39
  thing." I'm okay going to ChatGPT and, and doing stuff in there, so
• 20:45
  it's, uh, it's interesting times for sure.
• 20:48
  Yeah, it is. It's awesome. I, I'm, I'm here for it. Um, I've had some friends
• 20:53
  recently be like: "Man, I would go back to pre-AI stuff if I could."
• 20:56
  -Mm. -And you guys are crazy.
• 20:58
  -Yeah. -That is absolutely crazy.
• 20:59
  -Definitely, and I would- -I, I would, I would not wanna do that.
• 21:01
  -I mean... Yeah. -It's like a power tool.
• 21:03
  Yes. I think still the biggest issue with it is, like, you can, you can do so much
• 21:09
  more, so much faster, but in a corporate environment, we still have
• 21:13
  responsibilities, and we still, um, you know, compliance, quality,
• 21:18
  maintainability, there's all of that kind of stuff. And because it can write a lot
• 21:22
  of code very quickly, you know, in a couple of minutes, lots of stuff happens,
• 21:26
  but you still need to go and spend half an hour, an hour, whatever, reviewing it,
• 21:29
  -going back and forth, refining- -Yeah
• 21:31
  -... et cetera, et cetera. So- -And so, like, and that, that's kind of
• 21:33
  where I'm at with my guys, too, is it's like, listen, here's the deal. The AI is
• 21:36
  helping to write some of this stuff. You really have no excuse for not making this
• 21:40
  code really beautiful, like, and, like, very clean.
• 21:43
  -Mm. -Because it's getting you all the rough
• 21:45
  stuff out of the... Like, the stuff that you used to have to do, just to sort of,
• 21:48
  like, get the first version out of the way, is done.
• 21:50
  -Mm. -Like, it's already-
• 21:51
  -Yeah -... doing that. It's writing the tests for
• 21:53
  you, for goodness' sakes. So now just go to go refactor the code and make it look-
• 21:56
  -Yeah -... really nice, and make it feel
• 21:57
  maintainable, like, or not feel maintain- make it maintainable, right?
• 22:00
  -Mm. -It's already got the for each statements
• 22:02
  -in there. Now go use a collection. -Mm-hmm.
• 22:04
  Now go... Like, don't just be lazy about it. Like, take the time-
• 22:07
  -Yeah -... to actually go through, and now you
• 22:09
  -get to be the senior developer. -Mm.
• 22:10
  You need to be the... You're the one who's coming behind. This is not just save you
• 22:14
  time, so you can go play
• 22:15
  -Counter-Strike or whatever. -Right.
• 22:17
  You know, it's like... Yeah, actually go through-
• 22:19
  -Counter-Strike. -... and [chuckles] yeah. Anyway, um,
• 22:23
  okay, there's that. Arbor. Let's talk about Arbor quickly.
• 22:26
  -Mm. -And then we're gonna talk horseless
• 22:28
  -carriages. All right. -Yeah.
• 22:29
  Arbor.
• 22:30
  Uh, I've just put some finishing touches on it before we jumped on the call, and,
• 22:35
  and we're going to start using it at work for managing our work trees. It took a
• 22:40
  -little bit longer to- -So tell us what it is. Yeah, tell us what
• 22:42
  -it is before- -Yeah, so Arbor is like a work tree
• 22:43
  management tool, that kind of sits on top of Git and kind of provides the ability to
• 22:50
  clone down repositories and set them up in such a way that you can defer to your
• 22:55
  agents to then work on multiple branches and different features and fixes
• 22:59
  -simultaneously. -Work trees. We talked a little about work
• 23:02
  -trees last time, right? -Yeah.
• 23:03
  So this is just a work tree management tool for you to be able to spawn off
• 23:07
  multiple agents at the same time to go work on parallel features, essentially.
• 23:10
  Yeah. There's, there's, there's two ways of doing this. You can clone down the
• 23:12
  repository into different directories multiple times.
• 23:15
  Totally valid, nice and easy. Stay with what you know. Arbor and, well, work trees
• 23:20
  allow you to have one directory that then tracks all of these different branches at
• 23:26
  the same time and allows you to have them all in one directory simultaneously. And
• 23:30
  Arbor kind of then gives you the ability to
• 23:33
  manage all of these things. It, uh, helps you keep them in sync. Uh, you know, if
• 23:37
  you've got your, your main branch, you can then sync your feature branches easily.
• 23:41
  This is just a command, and it will track, "Okay, is this your main branch? Do you
• 23:46
  wanna do a rebase or a merge?" And then every time, you just Arbor sync, Arbor
• 23:49
  sync, and it'll pull down rebase and, and do what it needs to do there for you. So
• 23:53
  I've, I've started adding all these little niceties on top of it. It's got the
• 23:56
  ability to scaffold your applications, but it also has this notion of presets. So it
• 24:00
  will detect that you've got a Laravel application, and it'll go through, it'll
• 24:04
  copy the env example to .env. It'll generate a key, it'll run the migrations,
• 24:09
  um, and seed it as if they exist. It'll look to see if you've got a package-lock
• 24:13
  file and run npm and npm run build, and all that kind of stuff, you know,
• 24:16
  because-... as Taylor has, has said recently, Laravel really is a good
• 24:21
  framework for agentic development because it is conventional, because things are in-
• 24:26
  Yes
• 24:26
  - certain places. And so there are assumptions that can be made about a
• 24:30
  Laravel project based on what is and isn't there, so that you can then run these
• 24:34
  -things nice and easily. But then, I- -But wait, according to Uncle Bob, there
• 24:38
  should be nothing obvious about the fact that your app was a, application is a
• 24:41
  -Laravel app. -Yeah, but-
• 24:42
  It should just be your domain. That's all that matters. Larav... The fact-
• 24:46
  -Yes -- that it's a Laravel app should be
• 24:47
  -completely invisible. -Yeah, yeah.
• 24:49
  -All right, eat your words, Uncle Bob. -Mm.
• 24:51
  Do you remember Jeffrey Way and Uncle Bob got in an argument about that, Laracon,
• 24:55
  -little while ago? -Yeah. I think, yeah.
• 24:56
  This is never gonna make it to Uncle, Uncle Bob's ears, so I don't care, but-
• 24:59
  No, it's fine. It's, uh, I think, I think
• 25:01
  it's possibly that he has maybe had his time in the sun, and development is-
• 25:07
  [laughing]
• 25:09
  -Moving past him -... moved significantly. But yeah, so
• 25:11
  it's, you know, it knows the default things, but then for our application,
• 25:15
  we've got multiple databases. So from the preset, we can then go, "Okay, go and
• 25:19
  create this extra, these extra databases. Go and seed these extra databases. Go and
• 25:23
  write these extra environment variables." I've got it hooked up to one password, so
• 25:26
  -it'll go and inject all of the- -Nice
• 25:28
  -... credentials- -That's cool
• 25:28
  ... out of one password into the ENV file. So basically, this, this scaffold
• 25:33
  process, 'cause we've got so much to do, and we build our assets for multiple UIs
• 25:37
  and all of this kind of stuff, it takes two minutes. But what we're doing is
• 25:40
  running Arbor work, branch name, and it will go and do all of that stuff. Um-
• 25:45
  -That's super cool. -And it's like when you go into that
• 25:48
  directory, you can run composer test or Artisan test, or, you know, PHP Artisan
• 25:52
  test, or Vendor bin, pest test, or whatever you're doing, and know that it is
• 25:56
  ready to go,
• 25:58
  and it will just run the test suite. That all of the keys-
• 26:01
  -That's super cool -... that need to be there to function is
• 26:03
  -there. -Yeah, I love that.
• 26:03
  That everything is kind of just set. That it'll, it'll link into Herd because it
• 26:07
  detects Herd is there. It'll create databases with unique suffixes so that,
• 26:13
  -um, you know- -Ooh!
• 26:14
  -... it's isolated- -That's interesting
• 26:15
  ... this branch is there. So if you're using SQLite, it's easy. You just create a
• 26:18
  new database.sqlite file, and that's all self-contained-
• 26:21
  -Yeah, smart -... within that branch. But if you're
• 26:22
  using MySQL, we can't use the same database.
• 26:25
  -That's what we use. -Yeah, you can't-
• 26:25
  -This is a trick for us, too -... You can't use the same database.
• 26:27
  -I didn't know, yeah. -Hmm. Because if you've got multiple work
• 26:31
  trees, if you've got four branches at once, and you run-
• 26:33
  -Right -... tests in all of those four at the same
• 26:34
  -time- -Tests, yeah, you're clobbering all of them
• 26:35
  ... You're gonna blow away. Yeah. So we've
• 26:37
  got, like, a noun-adjective list, and it will just, like, smoosh them together, and
• 26:41
  then it'll go, "Okay, this is the key for this,
• 26:44
  um, for this work tree," and it'll create a database.
• 26:46
  Does it clean up after itself? Does it kill your database-
• 26:48
  -It does -... after the fact?
• 26:49
  -No, it does. Yeah, yeah. -See? Okay, so here's what's interesting,
• 26:51
  right? So, uh, I feel like there's been, like, three tools that have come out
• 26:54
  around this at the same time. So Conductor is a tool that David Hemphill uses, and
• 26:58
  -he- -Mm-hmm
• 26:58
  ... did a whole write-up about how he uses sort of hooks and stuff in Conductor to
• 27:02
  -do kind of what you're doing. -Yeah.
• 27:04
  Um, which may be similar, but like, you know, yours is a completely customized
• 27:07
  solution. You're not at all beholden to the creators of Conductor-
• 27:11
  -Right -... to build in tooling that you want,
• 27:13
  -right? You could just do it- -Yeah
• 27:14
  ... which is pretty cool. Uh, software for, uh, an audience of one, right?
• 27:18
  -Mm-hmm. -Which is the wonderful thing about AI. Um,
• 27:20
  there's also another tool that we talked about on Laravel News last week, which is
• 27:23
  doing similar to what you're doing, but I'm sure not exactly what you're doing.
• 27:26
  -Mm. -Uh, it mentioned the whole SQLite thing.
• 27:28
  MySQL is a very interesting wrinkle, and I love that, that you're doing that.
• 27:31
  -Yeah. -So super cool. All right. Interested to
• 27:33
  -see Arbor. I think that's really neat. -Yeah.
• 27:35
  -And, um- -Yeah, so I was, um-
• 27:37
  -Let's check that one out. -It's all... I think I've got it to, like,
• 27:39
  -version 0.9.1, and so- -Did you see Parfait, by the way? I'm being
• 27:44
  -ADD right now. -I did see-
• 27:45
  -Did you see Parfait? -I... Well, Parfait is... I saw it.
• 27:48
  -Leonard, Len Woodard? -Yes, Len. Yeah, yeah, he reached out to
• 27:52
  -me. -Like the T-
• 27:52
  -Yeah, I did- -The Tui crap?
• 27:53
  ... The Tui thing that he's been building, yeah.
• 27:55
  -With Arbor? -I did. He-
• 27:56
  -Yeah. -I spoke to him the other day.
• 27:56
  I'm saying, I'm saying, like, I wonder if that would be helpful with Arbor.
• 28:00
  Yeah. Um, so I've got a, I've got a ver- I mean, it's a v- it's on GitHub now, um,
• 28:06
  and you can pull it down and use it. So I was j... I've got it to a place where
• 28:09
  it's, it's ready to go
• 28:12
  into our work repo to be used, so the, the configuration's in there. But I was
• 28:16
  writing the documentation for it
• 28:19
  to, you know, tell the team, "How do we, how do we spin this up?" And I was, I was
• 28:23
  writing these grim things in there like, "You've got to pass the OP_volt and
• 28:28
  OP_item environment variables
• 28:32
  when you run Arbor work, because we need those variables to go and figure out where
• 28:35
  do we pull the credentials from to then inject into the-
• 28:38
  -Okay -... ENV file." And I'm like...
• 28:40
  -Yeah, sure. -"Now, that's, that's not good developer
• 28:42
  experience. I shouldn't have to tell that, and people have to remember it." So then
• 28:45
  I went
• 28:46
  and I said to Claude, "Hey, let's put some pre-flight scaffolding steps in here with
• 28:51
  conditions that, like, these environment variables need to exist. This environment
• 28:54
  -file needs to exist." -Yeah.
• 28:55
  And so did that, built it all out. So
• 28:58
  got that all done, rewrote the documentation again, and then realized if
• 29:03
  you, if you go install and you don't pass the flags to it, it doesn't actually set
• 29:09
  the, the version properly, and it doesn't set the commit and all those kinds of
• 29:13
  things. So I'm like: Okay, how do we get this running in, in Brew? So it'll be
• 29:18
  available in Brew, which then involved creating, like-
• 29:21
  -That's cool -... an organization to put it in and
• 29:23
  figuring out how to set up a Homebrew tap. Which, of course, um, Claude made all of
• 29:29
  this very easy, 'cause I just asked it what I wanted to do.
• 29:30
  -Of course, yeah. -And it told me. And it told me, you know,
• 29:32
  "Go and create a key here, call it this, put it into your-
• 29:34
  What did you write it in? Was it Go or something, or what did you write it in?
• 29:37
  Uh, Arbor is written in Go. The Homebrew stuff is-
• 29:40
  -Yeah -... is Ruby-based, so yeah.
• 29:42
  -Okay. -But it's gone and done all this, and then
• 29:43
  it's gone and created a GitHub workflow. So when a release is published, it'll go
• 29:46
  and, like, pull down all of the binary information, all of the checksums. It'll
• 29:49
  go and update this
• 29:51
  Homebrew tap thing, and then you can just do an Homebrew update, and it will upgrade
• 29:54
  whichever it is, and it'll pull down the new version. So-
• 29:57
  -That's super cool -... excited to, uh, get that going. And
• 30:00
  like, you know, and now, uh, the, the Codex desktop app comes out, has work tree
• 30:03
  support, or Conductor exists. Like, there's all these things that I'm like,
• 30:07
  "No one will use this except me at work," which is fine, but it was just a fun
• 30:11
  experience. As I said, at... When I, when I published the video about it, uh, last
• 30:17
  week or the week before, it was-... it was less about making something that I
• 30:21
  expected the world, quote, unquote, "to use," and more about learning how to use
• 30:25
  -AI to, to get it to- -Yeah
• 30:27
  ... do something that I think is of a, of a decent quality from a, from a software
• 30:33
  perspective, without actually knowing the language that it was written in. Like, I
• 30:38
  wouldn't have written this in Go, but I could-
• 30:41
  -Yeah -... read the Go that was being written to
• 30:43
  understand what it was doing, and then learning to use different tools to kind of
• 30:47
  validate the idiomatic practices that were being implemented, that it was, you
• 30:51
  know, consistent with itself. And, and, and that kind of... W- it was more a
• 30:56
  learning process of: How do I build this using what's available
• 31:02
  as AI tooling? More so than
• 31:04
  building a tool for managing work trees.
• 31:07
  -So- -Yeah. Yeah, makes sense. I get it. I love
• 31:10
  -that. -It's, it's there now. So yeah, after,
• 31:12
  after we finish recording, when my lunch break is over, it will, uh... my Claude,
• 31:17
  [chuckles] my Claude limit for the five-hour period will have reset, and I'll
• 31:21
  -be able to finish that off. -Love it. Okay, we are at 32 minutes, so
• 31:25
  we've got a couple minutes left here. Not very long- not very much time. All right.
• 31:28
  -Tell me about the car and the horse. -Horseless carriages. All right, so this is
• 31:31
  an article. I might have referenced this at some point, but it was basically
• 31:34
  talking about this idea of... Let me, let me just see if I can pull it up.
• 31:41
  Horseless carriages. I just did this today.
• 31:48
  Ah, there it is. "AI Horseless Carriages." Okay. Here is-
• 31:52
  -Peter Koomen? -... sort of, uh, let me see here. Yep,
• 31:56
  -Peter Koomen. -Go.
• 31:57
  All right, so here is what he said: "Whenever a new technology is invented,
• 32:01
  the first tools built with it inevitably fail because they mimic the old way of
• 32:05
  doing things. Horseless carriages refer to that early motorcar design that borrowed
• 32:09
  really heavily from horse-drawn carriages that preceded them." So they have an
• 32:13
  example in the picture of this carriage, this massive wheel with, like, basically
• 32:17
  this steam coal, you know, engine on the back of it.
• 32:21
  -Mm. -"The brokenness of the design was in-
• 32:23
  invisible to everyone at the time, and laughably obvious after the fact. Imagine
• 32:27
  living in 1806 and riding on one of these for the first time. Even if the wooden
• 32:30
  frame held together long enough to get you where you were going, the wooden seats
• 32:33
  and lack of suspension would have made the ride unbearable. You'd think, 'There's no
• 32:37
  way I would choose an engine over a horse,' and you would have been right, at
• 32:41
  least until the automobile was invented. I suspect we're living through a similar
• 32:46
  period with AI applications. Many of them-"
• 32:48
  -Mm -... "are infuriatingly useless in the way-
• 32:50
  in the same way that Gmail's Gemini integration is. The old world thinking
• 32:54
  that gave us the original horseless carriage was swapping a horse out for an
• 32:57
  engine without redesigning the vehicle to handle higher speeds. What is the old
• 33:02
  world thinking constraining these AI apps?" Okay.
• 33:05
  -Mm. -So he goes through and he says, "This is
• 33:07
  how we've been doing it," right? "We write a program, and then we basically say: How
• 33:11
  can we slap AI on top of this to make it better," right?
• 33:17
  And that's essentially the horseless carriage. And we've been trying to do
• 33:20
  this, I think, at work. So what we've been trying to do is we've been trying to take
• 33:23
  our data entry process for some of our claims and say: How can we speed this up,
• 33:27
  make this better, using AI? And what we've been doing is essentially trying to use
• 33:32
  AI to do almost like OCR extraction, which is fine, but like with OCR extraction,
• 33:38
  you kind of only have, like, it's either accurate or it's not accurate, and that's
• 33:42
  -it. -Mm.
• 33:42
  That's what you get, right? Um, you can't... Like, it's either right or wrong.
• 33:47
  It, it's either 100% or 0%. There's no real in between. Um, and so that's what
• 33:52
  we've been trying to do, but it's been difficult and, and sort of frustrating
• 33:55
  because it's like, okay, well, unless we can get it to a really high percentage of
• 33:59
  accuracy, we can't really use it, um, because it's just gotta be accurate. So
• 34:06
  the trick for me was thinking, like, okay, is this a horseless, horseless carriage
• 34:10
  design? We're taking what used to be, like, a just simple screen where people
• 34:15
  can enter in data and saying: We're going to pre-fill that data with the AI.
• 34:19
  Instead, what I think we need to do is essentially redesign the user interface to
• 34:22
  say, "No longer is that the case." What we're gonna do instead is we're gonna say,
• 34:26
  "AI is really good at categorising documentation and making guesses at what
• 34:31
  party it thinks these particular locations are referencing." So like, "Hey, here's
• 34:35
  an address block. Do you think this is claimant one, claimant two, or is this an
• 34:39
  insured party? We think it's claimant one. Do you agree?"
• 34:43
  -Mm. -"Yes." Okay, if you agree, well, now what
• 34:46
  it's going to do is it's not gonna make you fill that information into those
• 34:49
  -fields. It already knows what the names- -Yeah
• 34:50
  ... of those fields are and how to fill them in, so it just fills them in. Now,
• 34:54
  was it accurate 100% at the start? No, but it doesn't have to be-
• 34:58
  -Mm -... because the human is really good at
• 35:00
  recognising, "Oh, in the context of this document here, I can see this is an
• 35:04
  estimate. This is definitely going to be the insured party." Click Insured. They
• 35:07
  -know that- -Mm-hmm
• 35:08
  -... right? -Mm-hmm.
• 35:09
  The system probably even knows that, and if it gets enough thumbs up in a row, it
• 35:12
  can stop even prompting and say, "I'm 99% sure this is the insured party because
• 35:17
  you've told me it is. Every time it's been on an estimate and it's been this way,
• 35:20
  it's the insured party. I'm not even gonna ask you anymore. I'm 100% sure on this
• 35:23
  one." However, there are these other two, and how do we now use what AI is giving us
• 35:29
  to be able to create new user interfaces that allow it to do what it's really good
• 35:33
  at, and then allow the humans to do what it's really good at?
• 35:35
  -Mm. -So instead of just slapping it all into
• 35:38
  there, and then now you have no differentiation between, like, well, what
• 35:40
  did the AI fill in? What was just part of OCR previously? What did somebody else
• 35:44
  fill in before I did? I don't know. I have to go parse through that information and
• 35:48
  just go look through the docs myself anyway to confirm it. No. No, we're not
• 35:52
  gonna do that. We're going to grab the section of the page, we're going to show
• 35:56
  it to you, and we're gonna ask you to tell us which one it is and/or if you agree
• 36:01
  with our suggestion or not. And then after you click that, it does all the manual
• 36:05
  labor that you would have had to do before of filling all that in, and now, instead
• 36:08
  of taking three minutes to load a claim, it takes 15 seconds because-
• 36:12
  -Mm -... you're just clicking through yes, no,
• 36:13
  -or whatever, right? -Yeah.
• 36:15
  So I think this is an interesting thing. Um, I think as we're all pursuing this
• 36:19
  idea of, like, how can we use AI effectively, I really do think we have to
• 36:23
  -sort of reinvent it a bit. -Mm.
• 36:25
  Like, what are the things that we're trying to slap it onto that it's like,
• 36:28
  "Okay, rethink how the process could be done if you had-"... this when you were
• 36:35
  first designing it? You know what I mean? Um, and so anyway, it was, it's a really
• 36:39
  interesting read, and he has some really good examples of how he would design Gmail
• 36:44
  if he was starting over, right? This Gemini integration, instead of it just
• 36:49
  being like, "Hey, we'll do this for you," it's, "How could you use it, uh,
• 36:53
  effectively to help the user?" One of the things that he brought up that was really
• 36:56
  interesting is that the pre-prompting for Gemini is fixed. So if you tell it to
• 37:01
  write an email for you, you can't imbue it with any of your personality or any of
• 37:06
  your preferences, really. It's sort of pre-programmed already for you. So what
• 37:11
  his suggestion was on that is, it would be really nice if I, as a user, could write
• 37:14
  my own prompt to say,
• 37:17
  I w- you know, I'm, I'm a 34-year-old guy with, like, two kids and a wife. I enjoy
• 37:21
  -this type of thing. -Mm-hmm.
• 37:22
  I've been working in this environment for, like, 10 years. I'm experienced, I'm a
• 37:26
  professional. Um, my emails will commonly have misspellings, and that's okay, I
• 37:31
  don't care. It doesn't really matter, and I'm okay with the misspellings, 'cause I'd
• 37:34
  rather save the time. Uh, my emails are terse, but they're not rude, right?
• 37:38
  Whatever. Okay, now, with that context, my daughter is sick, I can't come in today.
• 37:42
  -Tell my boss. -Mm-hmm.
• 37:44
  Okay, well, that puts out a much different response than if it just uses Gemini's
• 37:49
  default prompting, right? And so this idea was, "Hey, I need to be able to write my
• 37:53
  own prompt," and you can't currently do that in Gemini, uh, in the Gemini
• 37:56
  integration. So anyway, it's an interesting article. I would suggest
• 37:59
  reading it, and it's, it's really helpful, and I've just been thinking about it a
• 38:02
  lot. So I'll let you know how it goes. Um...
• 38:04
  -Yeah, cool. -Our Andy, Andy Hinkle on our side has been
• 38:07
  the genius behind building this integration on our data entry product.
• 38:10
  It's super cool, the stuff that he's doing. It's unbelievable. I don't
• 38:14
  -understand half of it. -[chuckles]
• 38:15
  Every couple of days, he's telling me about these new breakthroughs that he's
• 38:17
  had, and, um, it's really fun. So yeah, I'll definitely keep you updated with our
• 38:22
  -progress- -Yeah
• 38:22
  ... and kinda how we're going on it, but it's been fun.
• 38:25
  Cool. All right, I like it. Sounds like a good place to wrap.
• 38:28
  All right. I agree, I agree. Uh, tell me the episode number, Michael. What are we
• 38:32
  -on? -186.
• 38:35
  186, folks. Uh, check out show notes for this episode at northmeetsouth.audio/186.
• 38:39
  Hit us up on Twitter, on X, on Bluesky, @jacobbennett, @michaeldirindo, or
• 38:44
  @northsouthaudio. And if you have any questions, we'd love to hear from you.
• 38:47
  Rate it up on your podcatcher of choice. Five stars would be incredible.
• 38:51
  [upbeat music] Thanks so much for hanging out with us, everybody. We'll see you
• 38:53
  -again in two weeks. -Bye.
• 38:54
  [upbeat music]