Slightly Caffeinated

Chris Gmyr (00:00)
Hey, welcome back to Slightly Caffeinated Podcast. I'm Chris Gmyr.

TJ (00:04)
I'm T J Miller.

Chris Gmyr (00:05)
Yo, T J. So what's up in your world this week?

TJ (00:08)
man, this week's been interesting. so you know, with with the kiddo in school and and just everything going on in life, I have been, and I talked about this in a previous episode, just like I've been dealing with some burnout across the board, just burned out on on life, like my wife's schedule, like her school classes just started for her semester, so she's super busy now too.

and between that and like Sonori and Halcyon and Prism, like I just I got burned out. And you know, I I thought I was coming back from it and I I wasn't quite there yet, but I'm I'm feeling there now. I'm actually excited to be cracking open editors and like working on stuff. you know, and and I think something we could talk about a little later too is I I finally got around to a Prism release. you know, I had

I just had some mental and emotional blockages that that kept me from getting to it. and kind of what tipped the scale is I had some stuff at Luma that needed some work and I got into that. But that that got me so energized to like get back up and running with things again. So diving back into all my projects and and feeling a lot better about it. so I

I I'm in a really good spot right now. So that that's that's kind of where the week's gone was just kind of investing in that, kinda, you know, protecting that energy and and trying to like nurture it. So yeah.

Chris Gmyr (01:40)
That's great. Well, glad to have you back. And I did see the

TJ (01:41)
Yeah, man. Yeah.

Chris Gmyr (01:45)
notification for the Prism release and there's a bunch of great stuff in there. So we'll

TJ (01:49)
Yeah.

Chris Gmyr (01:50)
dive into that in a few minutes. But yeah, great that you're feeling better

TJ (01:52)
Yeah man. So how about you?

Chris Gmyr (01:54)
and yeah, moving along.

TJ (01:56)
Yeah, yeah. Feeling feeling better than I have in the last couple of weeks. So I'll I'm absolutely gonna take it. Yeah. How about you, man?

Chris Gmyr (02:03)
Sweet. Yeah, we'll take it.

yeah, with me, lots of lots of the same. I feel like just have like a bunch of projects just going on and I don't know. I kinda love it. I kinda am a little burnt out on it, but I don't know. During the day, like working on it, it just feels so good to make so much progress. And then I don't know, it I feel like ideas beget more ideas, and then I'm like, I gotta write this down.

I gotta do this and I gotta do that. And it just like expands a little bit. So I talked about my other little side project jig a while ago with like workbench and it's just like little idea cache and whatever. So I was working on that a little bit, and then kind of this idea popped out of that, which we'll talk about in a little bit. Avicet's moving forward, Hive is moving forward, got like

five or six projects going on at work. I feel like just all over the place. but outside of just AI and coding and tech stuff, super excited to go camping with my son and scouts this weekend. So I have the day off tomorrow to kind of prep and get ready for it. And it's just gonna be out in the woods for, you know, two days and not sure if I'll have cell service out there, which will be fine. But it'll be good to actually have a complete

Break and just sign off and not worry about anything. Try not to think about anything, you know, and just

TJ (03:26)
Yeah.

Chris Gmyr (03:27)
relax a little bit.

TJ (03:30)
Yeah, I I'm very envious. Like I I really wanted to go camping this summer and I just for reasons and it got so hot and humid too, which I like I don't want to be cooped up in a tent with it being like ninety degrees and fifty percent humidity, you know, like

Chris Gmyr (03:48)
Yeah.

TJ (03:49)
just it n it got kind of nasty. So I did I never got around to it, but man, taking some time away in the woods I think would have done me

So good, like at any point this summer or even now. so I'm I'm envious of that, man. That sounds like a lot of fun.

Chris Gmyr (04:05)
Yeah, it'll be a lot of fun. so this will be our last big camping trip for Cub Scouts for him. we have like one little overnight to do later in like January, but then he like transitions over to like BSA and he'll be doing a lot more camping on his own. I'll still go here and there, but they're like a lot more independent, you know, at that time. So we'll just kinda see see how he does

TJ (04:26)
Yeah.

Chris Gmyr (04:26)
and where he goes and go from there. But yeah, we also booked camping trip for just

our family second week of November 'cause it's still relatively, warm or not freezing here. So it's like same thing. We don't want to be sitting in a tent with four people and it's like ninety degrees and high humidity and all that. So have that to look forward to, you know, as well. so it'll be my daughter's first camping trip. So we'll see how that goes with the four year old. it could be super fun, could be super stressful, but we'd only be

TJ (04:53)
Yeah.

Chris Gmyr (04:53)
about like forty minutes away. So it's not too bad if we have to

Come home early.

TJ (04:57)
that's a great, it's a great I I think it's like a really nice distance for that. if you've got a bailout, you can bail out. it's not too far. but it's far enough to make it worthwhile. yeah, and and it's like it's the sleeping in the heat. Like I I can't handle sleeping hot. Like I even here at home, like I've gotta sleep in like cool temperatures. Otherwise it just doesn't happen.

So that's it's the the daytime's fine. It's like

I

Chris Gmyr (05:24)
Mm-hmm.

TJ (05:24)
cannot sleep in the heat. And like the worst is like waking up sweating. So like, you know, I've I've ran into that plenty of times, you know, camping. So I I just I couldn't couldn't do it. But yeah, that's cool, man. I my my son, I think, is like really interested in camping too. My wife is interested in glamping. and so like I'm fine with like a tent and

a blow up mattress or something, no problem with that. But I think my wife is at camper R V levels of of camping, so

Chris Gmyr (05:56)
Yeah.

They might check around you 'cause at least around here and then I know in New York you can rent places like a little cabin or something like that where it's like more

TJ (06:03)
Yeah.

Chris Gmyr (06:04)
clamping and it has like tent camping attached to it, or you know, you're just in like a slightly different you know, camping spot that you can just walk back and forth to. So I don't know, maybe best of both worlds if you can find one.

TJ (06:16)
Yeah, I know there's definitely some little cabin type camping situations around here where you can like rent them for the weekend or whatever. so yeah, I think I think it's definitely an option. but it's just it's just so funny because like I'm I'm no problem backwards camping, but my wife is just not having it. Yeah.

Chris Gmyr (06:33)
Yep, yeah, totally.

TJ (06:36)
So on on the coffee front, man, have you have you heard of Seven Brew or like the hype around Seven Brew coffee? So

Chris Gmyr (06:44)
Don't think so. I don't think it's come up.

TJ (06:47)
we've got them popping up all over here. I think there's like there's one that just opened by my house. There's another one that's been open for a while, and they're they're opening up another one like not too far from my house. and I asked my wife about it, because like there were

lines backing up out of the parking lot and down the road to get into it. And come to find out, have you heard of like the the Nido craze? Like the yeah. So like I guess they had like yeah,

Chris Gmyr (07:12)
Yeah. Yeah. Like squishies.

TJ (07:15)
they're like little s jelly stress ball things.

Chris Gmyr (07:18)
Mm-hmm.

TJ (07:20)
and I guess they were having some like promotion with with Nido or or something. And so people were going nuts about it.

And so I thought there were all these lines over like really good coffee. And so I told my wife about it. I'm like, man, like I kind of want to try this now. And she's like, not for you, homie. their their specialty is like really sugary drinks. And I I came across a a funny reel on Instagram that was, here's my application to work at Seven Brew. And it like cup full of ice and just like him pumping globs of caramel into it. And so

That's that's I guess the pair that with what my wife said. that is absolutely not my my kind of drink. And so I I guess as my wife said, the the coffee is maybe if you're looking to go get like a really nice cup of coffee, that's maybe not the spot. If you're looking for like a big sugary hit, maybe that's it. So that that's I don't know, I was just amused by that. But it's crazy. They're like

popping

up everywhere around here now, so

Chris Gmyr (08:25)
Yeah. I haven't seen any around here, but I'll keep my eye out. Not that I go out and about too much, but when

TJ (08:30)
Yeah. Yeah.

Chris Gmyr (08:32)
I do, I'll keep my eye out.

TJ (08:33)
Yeah. And bean-wise, we're on some Ethiopia Jurgischef, which is one of

Chris Gmyr (08:40)
nice.

TJ (08:41)
one of my definitely up there is like one of my favorite beans of all time. this is from a bigger batch roaster, so it's not it's not like super fresh roasted, but it's still really good. we used to live

in Ferndale, and there was a like high-end roaster who was like always like anything you ever bought from him was roasted within like two, three days. and he used to get in some like really specialty batch like Ethiopia Yurishef and that was maybe the best cup of coffee I've ever had. and so I was nice it's nice to kind of get some of that beans in here, even though it's not quite to that bar.

it's still just such a super smooth cup of coffee.

Chris Gmyr (09:25)
That's cool. I like some yoga chef. That's always been good. Usually I

TJ (09:27)
Mm-hmm. Yeah.

Chris Gmyr (09:29)
do like iced coffee with that during the summer.

TJ (09:31)
Mm-hmm. for sure. I and that's I feel like that's one of the few beans that that works really well, like all the way around, right? Like any way you brew it, you know, iced French press, you know, pour over, it's all it is just great. It performs great everywhere.

Chris Gmyr (09:50)
Nice. Like it.

TJ (09:52)
Yeah.

Chris Gmyr (09:52)
for me, let's see, I think I have basically the same bags that I've had before in the subscription box. but the one that I have in there right now is the darker roast that we got again. I know I mentioned it before, maybe like a month or two ago, that we tried it and it was really good. So, got that again and then on Sunday the kids were like kind of occupied in the morning and it was kinda quiet. I didn't make the drip coffee yet, and I'm just like,

I'm gonna break out the Kemex and kettle and do that.

TJ (10:23)
Yeah.

Chris Gmyr (10:24)
So yeah, broke it out and made a really good cup of coffee. so my wife came down and was surprised. It's like, it'd be fancy this morning. I'm like, Yeah.

TJ (10:30)
Yeah. No, it's so funny. I was

I was just lamenting with my son this morning on the drive to school. I'm like, on one hand, I I love having a French press in the morning, but know, he he brought up we used to have the like a pod system. I can't remember which one. It wasn't it wasn't the popular one, it was some other one. and he's like he was lamenting not having that 'cause he used to use that to make like

Hot chocolate or cups of coffee, here and there. And I was I was like, yeah, just it's expensive and recycling and waste. Like it just I can't, we couldn't get on board with it, which is why we got rid of it. But I was lamenting over like, I kind of miss having an auto drip that I can just wake wake up to coffee because it is such a chore, like waking up, getting the hot water ready, getting the French press ready. Like it's

Chris Gmyr (11:19)
Mm-hmm.

TJ (11:19)
And just sometimes waking up to coffee already ready to go is just so nice. but I I I don't know if I could give it up 'cause I do I do just love a French press. So

Chris Gmyr (11:31)
Yeah. Yeah. and yeah, ours I mean, it takes a little bit of work 'cause we still like grind the beans. We like put the water in, like it's not like a fully automatic, like some people have. But I don't know if I would want to I don't know, grind the beans like at night or whatever, put it in the morning and I don't know. It's it's fine just doing it like that. So it's like a little bit of prep still, but it's fine. But it's not as much prep as probably like the French press and the ChemX for sure.

TJ (11:57)
Yeah.

Chris Gmyr (11:58)
But

it does make a really good cup of coffee. I do miss it. So yeah,

TJ (12:00)
It does. I do like it. I I I do really like it still.

Chris Gmyr (12:05)
I don't know. Maybe I'll work that into the weekend mornings when we're actually around. So can't do it this weekend. We're gonna have camp coffee, which is not as smooth for sure.

TJ (12:12)
Mm. No. But there

is something special about camp coffee though. Like I don't know, it it it's never as good, but it it definitely hits a certain way.

Chris Gmyr (12:24)
Yeah, yeah. Especially just waking up and it's a little chilly and it's like, Okay, adult coffee time, let's go.

TJ (12:28)
yeah. Yep.

Chris Gmyr (12:31)
Cool. So yeah, we we mentioned it before about prism release. Let's see

TJ (12:35)
man.

Chris Gmyr (12:36)
what happened.

TJ (12:36)
Yeah, so I man, like I said, I just I've had this mental emotional block of getting back to it. And it's still some of the same stuff I was wrestling with around Lyricon and just like dealing with some like guilt and shame around abandoning it and and then at the same time like having this huge list of issues, huge list of PRs to be dealt with and

it just it felt like, an insurmountable amount of work to like pick back up and sift through and like you know, stressing out about the decision fatigue from having to make a ton of different decisions. So I sat down, part like I said, part of this came from needing to make some updates to enable some work at Luma. So like I've now got some exterior.

motives to get some things updated because we want to use some newer anthropic models which require some updates in Prism. So I I kind of just sucked it up and sat down with Opus 5.5 and I had built at Laricon Time a Laravel app that was supposed to help me triage it. Like I basically set it up so that

it like sonnet could rip through every single issue, every single PR, and start to like find seams and find like easy wins and you know separate stuff that is like needs decision making from stuff that could just probably get merged as is, ready to go. and so I started looking at that, but I'm like, well we've got all the analysis done, but like

It still felt like a lot of work and a lot of content to sift through to look at all these things that you know Sonnet laid out for me. So I've really come to to trust Opus 5.5 quite a bit. And so I'm like, now that we have the analysis, like help me get through this. And before this, I really hadn't looked at the repo at all because it just it it

For some reason, pulling up the repo was giving me a pit in my stomach. So I hadn't really I hadn't looked at it at all. But what Opus found was that Wishborn, and a huge shout out to Wishborn, had been maintaining a community fork of Prism. And the way in which he had done it was merging in these PRs.

that and getting the the first look at a bunch of these PRs merging it into the fork and it was done in a really clean way and so we opus and I took a look at it and we're like well we can merge like after some discussion like Opus decided that like we can merge up to a certain point of this fork and then we've got to make some decisions moving forward on the rest of the stuff. And some of those things may need some tweaks too because

not only was this merging other PRs and resolving issues, the fork also kind of extended upon that with support for things like OpenAI and Anthropics file APIs. So op uploading and you know managing you know context files for the providers. And then also added in like telemetry, which is definitely something that I've

wanted in Prism and have approached two or three times and done some iterations on that. so I have to look at that stuff a little bit further. But it was really cool that we were basically able to like wholesale take you know, a bunch of this fork and like merge that stuff in. And then we cherry picked some stuff from later commits in the fork that were some like bug fixes and things, kind of

sidestepping some of the additional feature work that I'm either not ready to merge yet or or anything else, but massive, massive shout out to Wishborn for maintaining this

Chris Gmyr (16:35)
Yeah, that's so awesome.

TJ (16:36)
and like having it prepped and ready to go. And like it was so cool because I was able to take all of this stuff and create a new branch and merge committed. And so all of the commits like then started closing out PRs and like everyone got

credit for their contributions, which was important to me that people who did open PRs, and especially Wishborn got authored credit in GitHub, in the commits for the work that they've done. I normally squash commit everything, but doing a merge commit this way like maintained all of the authorship and and history. So yeah, Wishborn made it really easy for me to be able to like just do a

of stuff right off the rip. So the the release was 0.101 and that essentially merged like 35 PRs. So it closed out 35 PRs. and we cherry picked some bug fixes along with that. and so it just had like a ton of stuff, lots of fixes. we dropped layer of 11 support, which is something that

I wanted to do anyways. and there's a couple known issues that we gotta we gotta work around along with it, but you know, kind of eyeing up from here is the batch and file updates, like that's kind of next on the list. And then I I I was working on this before I dropped off the face of the earth with it, but I was working on human in the loop.

and it looks like Wishborn had added human in the loop support to in like one of the commits in this fork. So I kind of wanna either take a look at that as inspiration or maybe just run with that. So that I've got I've got some n stuff right next in the canon. but man, this it felt so good to get a release out because O dot 100 was like I I tagged that March twentieth.

and so it's well, it just felt so good to be back in in the weeds with it and like felt really good to go from a place of being like guilty and like ashamed of and embarrassed dropping off of it to like look we we picked up with a huge release, a ton of stuff. and I've got some plans to like continue working on it over the next like

week or so, hopefully I can get everything up up to snuff. And I basically want to get all this stuff in, zero out issues, zero out pull requests, and tag like hopefully like sit at a release with all of the stuff in it for maybe a couple weeks and then tag a 1.0 and then I've got some ideas beyond one point

maybe even into a 2.0 territory. so you know and you had asked an interesting question like some other people have brought this up and I I talked about this around Laracon too. you know what does it mean for Prism now that Laravel has their AI SDK and that that's super successful and first party support and you know I I've

I I think there is room for both of them. I think the Laraville AI SDK is a really high-level abstraction, with some some, you know, levers and stuff to drop down to a little bit lower level, but it's not nearly as bare metal as Prism is. And so I think there's room for both worlds. You know, if you want to do something that is highly custom or in a a very particular way.

Or if we support a provider that Laravel AI SDK doesn't. you know, I I just I think there's room for for both of them and you could even mix and match them. I could see of like, you know, for a handful of your workflows, AI SDK works great. And and maybe for a few custom pieces that you want to build, you've got to drop down a little bit lower level and like prisms there to support that.

And for a 2.0, I might move towards tighter framework integration and a little bit, you know, offer some little bit higher level support, a little bit closer and in the direction of AI SDK, while still maintaining its its low-level support for things. So I've got I'm I'm starting to get some fresh ideas and some new inspiration for it, which feels really good, honestly. So you know, I'll

What I'll pro I I you know, one of the other things I gotta look at like how much it's realistically gonna be used to and like I've gotta kinda balance that with putting effort and time into it. So if download numbers, you know, are are still there, especially for the newer versions, then I'm gonna definitely continue to put time and inspiration into it. if I if things really like stall out, then maybe I don't put more work into it, but at least for now

I'm I'm in like I'm reinvested into it. I'm re-energized by it. and and it just felt so good to get a new release out there. And I'm that just being able to tag that release and have an immediate plan to move forward on the rest of the features that I I really wanted Prism to have and what the rest of this fork also contains.

it just it it's it's got me in a different mindset and resolved some of this mental, emotional blockage that I had with it.

Chris Gmyr (22:06)
Yeah, yeah. I think that's a really good plan forward. And at least right now I still feel like there's room for both the Laravel AI SDK and then Prism. and that's how like a lot of their integrations work too. It's like, well, if you want mu more of the file system, you can, you know, call that lower level or even, you know, Redis or other things that it, you know, patches or has like managers for. You can also, you know, go back down to more of like the the base layer, you know, of that. So

Yeah, I think the the plan is solid. keep an eye on the downloads and the interest, as you cut your releases and see where people are updating or if it, you know, stalls out and then you can just decide then. But yeah, I think it's a solid plan moving forward.

TJ (22:49)
Yeah, as of right now, it's Lairavel Lairav removed first party support. they removed it as a dependency back in April. I kind of I I totally understand why that happened and you know, just directionally things started to

move away from each other. So it just it made sense for them. but that was back in April, they pulled that as a dependency. And that was also the the highest number of installs for Prism. I think for the month of April it had like 900,000 installs. since then it has fallen off, but we're still maintaining somewhere between 700 and 800,000 installs a month. So

it's it's definitely still getting installed plenty. you know, and so if we if I keep seeing these numbers stay steady, especially with like more updates coming in and and really hitting you know, getting to a place where we can tag a a a very stable and predictable one point then I'm I'm still invested in it. I still use it day to day at Luma. So like I've got some

I've got I've got some other investment in it, but I'm if we if like I said, if we maintain these numbers, then I'm on board to to keep keep rolling with it. So

Chris Gmyr (24:04)
Yeah, that's awesome.

TJ (24:06)
Yeah, the it's crazy. looking at it, because I hadn't looked at the repo, I didn't really know where those numbers were at. But like you know, checking it this morning, it's it's been downloaded almost seven and a half million times. Like that's insane

Chris Gmyr (24:20)
Ink.

TJ (24:22)
to me. Like absolutely insane. so yeah, I'm I'm reinvested, I'm re inspired, I'm re energized by it. So

We can keep looking forward to more releases and more features coming out. like I said, I'm hoping to get to like parody with the fork, maybe excluding or including like telemetry and human in the loop, like within the next like week, most two weeks. so yeah. I'm I'm I'm back at it and feels good. Feels good again.

Chris Gmyr (24:53)
Love

it. Love it.

TJ (24:55)
Yeah, so moving on from Prism, I know you've got some new updates with Avaset. And I would love to hear some more about that because I just recently updated I I think I was a couple of versions behind and and now I'm fresh up to date. So yeah, let's

hear a little bit more about Avaset.

Chris Gmyr (25:11)
Sweet.

Yeah, just been cranking on this too. one of the main focuses for you know the side projects going on. So have been doing a lot of updates. So pushed in like auto update features and functionality before. So when you open it, it'll, you know, periodically check the server and you know say that there's an update available. so you can, you know, download and it'll refresh that automatically.

wanted to this was a a few weeks ago, like open it up to like pricing and more like beta users and things like that. So I w wanted a way to for people to be able to report a problem. So now there's that in the settings, like about in debugging area. So like if you have a problem, you can you know just type in some text. You can optionally put your email address, you can optionally add in any sort of like debug logs and system information in there.

Because also trying to stay like privacy focused, as well. so not enabling those by default. You kind of like opt into all that stuff and then, you know, write whatever, you know, problem or issue or feedback that you're seeing. then add l like a handful of fixes and improvements. So now there's like a global mic mute shortcut. So I have that to like hyper for mute and unmute, which is super nice. So you don't have to do any like mouse or

whatever juggling during meetings, you can

TJ (26:32)
Yeah.

Chris Gmyr (26:32)
just hit that. more like inline updates. you can delete parts of the transcripts, words, sentences, whole blocks. You can restore them. you can undo them. there's a lot more like better like error handling and messaging. So like if you there's recording limits. So I think the default is like two minutes. So if you start a recording

You end it in under two minutes, it'll just delete it because it's like, well, a real meeting isn't gonna be less than two minutes typically. So if you see that, it'll have like a a warning message of like, hey, you know, your limit is, you know, set to whatever, how many seconds? this recording was under that. So it was discarded. there's a bunch of like cleanup with that as well. worked on your echo fix a little bit too. So that is working a lot better now and did like a whole bunch of

Testing with that in various mics and setups and Bluetooth and wired, laptop speakers, laptop mic, all the things. So that's working a lot better. And we basically like match up what you are saying on like the me like primary track. And then if there's like X amount of words that match, we'll basically take that out of the system audio.

And we're able to remove that double echo on there for you speaking into the system because it's the mic is picking up both. So it kind of alleviates any of that like echo problem. So if you ever run into another meeting where you're using the laptop speakers and mic again, hopefully it'll work a lot better for you.

TJ (28:05)
Yeah, yeah, that's awesome. I I was really curious how you were gonna fix that because that's just such a tricky problem to solve. I was wondering if you were gonna do something similar to this or you know, rely on the voice printing stuff that you have, of like being able to take like an imprint of a voice and where you're doing speaker matching and then basically matching it against my own voice and then saying like if it doesn't match your own voice.

on your track then we zero that out. but this is a great way to approach it.

Chris Gmyr (28:36)
Yeah, yeah. There's a couple of trial and errors and we had three or four different ways that we tried and this one was most successful with my trial. So we're we're going with it for now. So if someone reports it.

TJ (28:47)
Great. Yeah. Yeah. This is great.

Chris Gmyr (28:49)
Yeah, so if someone reports it, you know, again, you know, we can continue to to work on it. yeah, and then was it the other other day, two, three days ago, the thirteenth, launch launched the full launch.

with pricing for Avicet. So decided on fifty-nine bucks a year. There's

TJ (29:04)
Mm-hmm.

Chris Gmyr (29:05)
a 10 meeting trial. there's now a download like trial link on there. and then if you go past those like 10 meetings, it'll say like, hey, you register max like 10 meeting trial type of thing. Here's where you can check out. handling a bunch of that with checkout and using like Polar SH, which you're using for some of your apps, which was like

TJ (29:25)
Mm-hmm.

Chris Gmyr (29:26)
super easy. It was fantastic to work with. So

Thank you, Polar SH,

TJ (29:30)
Yeah.

Chris Gmyr (29:30)
for all that. It's a super cool service. and yeah, been handing out links and having more people, you know, try it. been posting it online, socials and in a couple of Discord servers. So just trying to get some traction and feedback for the app, the landing page website. all the docs and change logs are on the website, you know, too. So kind of seeing what I can do.

to support, the launch and assuming it would be quiet because I haven't written about it that much because I've just been busy

TJ (29:59)
Yeah.

Chris Gmyr (30:00)
building. So eventually here gotta turn into more of the the marketing, you know, engine on there. But have a bunch of things like that are kind of sitting on my system too that haven't been released yet. So like cleaning up like old meetings, kind of command bar improvements, content.

fixes for different windows and stuff like that. and then working on a new like playback engine that will allow you to control speed. so like I listen to podcasts and all all that stuff on like two, two and a half X. So like being able

TJ (30:32)
Yeah.

Chris Gmyr (30:33)
to like zip through meetings and like unassigned speakers even faster. it it's very fast with just like the one X in AvaSet right now. I feel like

Even for the speakers that I don't have marked yet, I'm able to like assign them and get out of the meeting in like thirty seconds or less, which is really nice. which is not something that I experienced with Crisp or others.

TJ (30:57)
Right.

Chris Gmyr (30:58)
but this will make it like even faster to just scrub things, and, you know, kinda have whatever preference that you want for like the playback and being able to control that yourself through the app has been yeah.

TJ (31:10)
It's sick.

Yeah, man, this is awesome. I'm so

Chris Gmyr (31:12)
So yeah, well that's a lot of

TJ (31:15)
I'm I'm so excited for you, man. I'm I'm so stoked that you were you got this out there and like pricing and and the the checkout and and license and trial. Like this is I

This

has been such an awesome journey to watch, but the app is so solid. I use it. we've got daily stand-up, weekly standing meetings, and this has worked so well. I even used it outside of work. I had

some calls about insurance and other things. And it was so nice to have the the specifics from that call just easily available for me to copy and paste into my notes or, with with the insurance stuff, there's pricing and terms and all sorts of wild things. And it was great to be able to just have that available you know, to go talk to my wife about it. And it just it it like

It was some web portal call to so like not not Zoom or anything standard and Avaset just picked it right up and recorded it and I was able to like label it right off the bat. Like it it's been such a great piece of software to use. So and massive congrats on getting this launched.

Chris Gmyr (32:23)
Thank you. Yeah, I appreciate it. yeah, and hopefully like more people use it and can enjoy it. So we'll see we'll see where it goes. not expecting, you know, too much, but hopefully it'll be a nice surprise if people, you know, enjoy it and continue to use it. And I don't know. I've had like a little bit of I don't know, worry, maybe not like anxiety about it, but it's like, well, with all the people, you know, vibe coding and kind of ripping off software and things like that, it's like, well

Couldn't someone else do this? And it's like, well, maybe for maybe like the basics, but there is like so much stuff going on behind the scenes to make it a really nice experience to make sure the speaker matching is correct, making sure like the engine of it works. It's like I don't know, unless you're like really into this, why you would build

TJ (33:09)
Right, yeah.

Chris Gmyr (33:09)
something like this. So I'm trying to be a little bit more positive, you know, with it, but it's hard.

seen all the AI news, you know, and posts that I see

TJ (33:20)
yeah.

Chris Gmyr (33:20)
of like, well, you know, people are ripping off you know, Adobe and Microsoft and these other things and it's like, well, I don't know. So

TJ (33:28)
Yeah, and I've

I I totally feel that. Like I've had some hesitation around tonori as well because there's a bunch of open source versions of it now and you know similar solutions everywhere. And then yeah, everyone's like vibe coding like their own versions of things. And you know, I think there's something to be said. you you absolutely I think nailed it that there's a level of care and like detail that, you know

you or I would put into our stuff that maybe isn't getting picked up by a slop canon or by open source. And you know, there's I I set the bar really high. If I'm gonna be like asking people to pay for this, I've set the bar really high for myself to put out something of quality. And I don't know how much you can say that for these other solutions. Like you then maybe or maybe not, but like

I know at least for me, yeah, you're you're buying something that I've put a lot of care into. you know, and there's something to be said for that. And there's like that level of domain knowledge that like you have about this and that I've got about Sonori that, you know

someone just putting together their own version open source maybe hasn't paid attention to all those details as as tightly. so I I think there's still definitely room for this stuff. And there's also plenty like we're in a bit of a bubble. Like we're we're surrounded by tech people who have the ability to do this. I was talking to my therapist about Sonori and she was so stoked because she's been wanting an AI like notes

thing that she can take voice notes for, but it needs to be like HIPAA compliant. And I'm like, look, it's doesn't

Chris Gmyr (35:07)
Mm-hmm.

TJ (35:07)
leave your computer. And she's like, that is incredible. And she's not gonna be someone who's gonna go vibe code something. So there's like these untapped

Chris Gmyr (35:13)
Yeah, exactly.

TJ (35:15)
markets, like I wanna I don't want to say like maybe untapped, but there's markets for these things outside of our developer bubble that aren't being vibe coded. She's not gonna go grab an open source version of of you know that that's just it's not realistic.

for

a certain market for these these solutions. So like same could be said for my wife. I she would love AvaSet and like would use Sonori and like she's not gonna go use something open source or, you know, like vibe code something herself. So I think I think for us part of the challenge in making these products successful is tapping into those markets outside of our like immediate circle, which are all developers who can

or could build their own version or might reach for open source. yeah.

Chris Gmyr (36:05)
Yeah, that's where I'm coming back to too is just finding that market or group of people. Like even if I have to switch like marketing gears to something else, it's like, fine, you know, as long as it's working

TJ (36:17)
Right.

Chris Gmyr (36:17)
for me, working for you, people that I know, you know, whatever. and it's working for this other group, T P D in the future, you know,

TJ (36:24)
Yeah.

Chris Gmyr (36:24)
that's fine. 'Cause I I don't know. I think it's a pretty solid piece of software. So yeah, and only being, you know, sixty bucks whatever a year.

I don't know. Can't

TJ (36:33)
Super reasonable.

Chris Gmyr (36:34)
beat it.

TJ (36:35)
No, that's so reasonable. And it's not, you know, I what crisp is like somewhere between 10 and 15 bucks a month. You know,

Chris Gmyr (36:43)
Mm-hmm.

TJ (36:44)
and this replaced it and not only did it replace the functionality, but it performed so much better. and it's you just you cannot beat that as a yearly price. Like it, yeah.

Chris Gmyr (36:56)
Yeah, a hundred percent. So my plan is to keep the price, you know, here. Not sure if it'll be just like a a beta, you know, pricing temporarily, but we'll keep it here and see, you know, where it goes. but yeah, if anyone is interested, it's live on the website at avacet.dev. link will be in the show notes and would love to have you try it out. and again, 10 meeting trial, so use it as much as you want for those ten meetings.

And it doesn't matter if it's a half an hour meeting or a three hour meeting, just counts as one meeting. So we'll love people to try it out.

TJ (37:28)
Yeah, absolutely go check it out.

Chris Gmyr (37:30)
Sweet. so we're getting a little high on time. We have a couple other projects. you want to talk about new meta type project or you want a quick

TJ (37:40)
Yeah.

Chris Gmyr (37:41)
rundown of five new features? Maybe we can get through both.

TJ (37:43)
I'd love to hear

I I'd I definitely need some updates on Hive because I I would love to I I'm I'm staring at working on Prism with Hive and with with getting back into Sonori and Halcyon with Hive as well. But I am so curious about this AI meta project and I just like I I saw your tweet and it's I I I feel like it's something I need right now, actually. So

Chris Gmyr (38:09)
cool. Okay. So hopefully we can get through this quick because I definitely want to get through hive stuff too. So I'm calling this like my AI like meta project. Like I've I capture so many things on Twitter, on Instagram, blog posts, whatever. And I usually just like throw it into my to do us like inbox. It's just like I'm I'll check this out later because a lot of it is like, well, here's this like new skill or updated skill or different cookbook from

anthropic or this prompt to change, whatever setting, or to look into this. And it's like I feel like I need to be on my computer, like run through these, like spend twenty minutes to an hour like vetting these new things that I'm seeing or, you know, whatever. And they just keep piling up because I never get a chance to really kind of comb through the list at all. So I'm like, it would be great if I could just dump this into like an AI project.

And it would go through, it would research everything safely, it would vet things, it would look into the projects that I have running locally, anything that I wanted to do like in the future, looking through transcripts of past, you know, chats and things like that. I was like, this is this actually like viable or would you use it or not, or possibly. so that's what I built. so I can basically still share things to to doist. I set up a project that it's like the AI.

like meta like inbox and I can just send it whatever sort of links that I come across or videos or whatever. It'll take all that stuff. And it will search through that project and to do list every hour, bring things in, triage it and research whatever the thing is, and then look into my transcripts and projects and all the things that it knows about me already. And

At the end of the day at about 5 30, it'll take whatever's new for the day and put it in this like kind of newsletter, newspaper like format. And it'll give suggestions on like, hey, we should really adopt this for like system-wide or the specific project. or like maybe we could trial this, or like, hey, this is kind of garbage or like not needed anymore, maybe with new models. so like let's just completely, you know, drop it. So

yeah, it it spins up this clawed artifact. So I'm able to like click buttons, I can chat with it right in there as I'm reading it. because I basically wanted to like kind of read this at the end of the day, like waiting for the kids to go to bed, like at bedtime, and just be able to like kind of zip through all these things and be like, yes, like let's approve this for this project. And it'll using Hive, it'll communicate with that lead once it's back up and running. and then

Say, hey, here's this trial, and then kind of take action from there. So it's a really easy way for me to like vet these things, kind of behind the scenes, and then I give it like final approval or overrides of like, well, you said like don't worry about this, but I actually want to try that because of XYZ. I can override that and then you know can do whatever sort of trial implementation, whatever from there. so not only will it take like my ingested data links, blog posts, whatever.

but I also pointed it to the cloud code, GitHub releases and the codecs releases. So it will go through all those releases on a daily basis and then see if any like new features or config changes or whatever are related to any of the work that I do on my personal system or on the work system. so at work I mentioned we use AWS bedrock with the AI gateway. Things are a little bit different over there.

a little bit slower with some of the newer features for Anthropic. So there's some things that I can use on my personal system that I can't use on the work system, or it's delayed for, you know, a few weeks or a month or so or something like that. So it's nice to know, like, hey, I can actually use this feature now in bedrock because it was released. So it'll capture all this data, make a new little newsletter, newspaper thing for me that I can kind of click through and take action on and like then instruct the agent to then spread out.

and either make this change or make this skill or adjust the skill or create this config change locally or make a note for me to to make the change on my work system for like bedrock. So it's like kinda all inclusive of like AI ingest triage action.

TJ (42:30)
Yeah, that is absolutely something I've needed. Part of part of my burnout that I've been dealing with is just so much is changing so fast. And it's a lot of things that I feel like I need or should care about. And I I I I have lists and like things littered and like bookmarks and stuff all over the place of like

things to research or look into or curious if it applies to what I do or what I'm doing or how I work. And I cannot keep track of it all. And it is I I get that similar kind of block to it where it's just like, I now have so much to go through and it just keeps growing that I avoid it.

Because and I I never look at any of it and then, you know, I find out something weeks later that it was a setting I should have set and it was transformative to things and I just I I missed the boat because it got lopped in with like a whole bunch of other stuff.

Chris Gmyr (43:32)
Mm-hmm.

TJ (43:32)
so this this is something that I definitely feel like I'm I'm needing now. so that's that's awesome. I love I love the way you've approached it.

That's that's so cool, man.

Chris Gmyr (43:44)
Thanks. Thanks. Yeah. Love to see I don't know if there's a way to actually package this up. Cause it's all like kind of just pros instructions, you know, right now in

TJ (43:54)
Yeah.

Chris Gmyr (43:55)
a new hive project. But yeah, I can I can maybe tinker with that or at least send you the instructions.

TJ (44:00)
Yeah.

Chris Gmyr (44:01)
but yeah. yeah. So only like first day with it. I have like a whole bunch of things to triage like my first you know issue here. But

Yeah, I think it's gonna be super helpful long term and especially bored on the weekends and just perusing things. It's like, okay, ten things got added to the the triage board and just knowing that that AI will kinda take care of it behind the scenes and I'll get like a a debrief at the end of the night or the next night. yeah, super helpful. So looking forward to seeing how it works out.

TJ (44:27)
And I wonder it makes me wonder how much you could integrate recursive self improving of it it you know, figuring out the stuff that that like after the fact that you do find useful, or you ultimately end up finding useful from everything that it's like gassed at based on, you know, what you're introspecting on your projects and everything else.

You I wonder how much there's that self improving aspect of no, I did really find X, Y, and Z useful. This thing was like tangential and not really useful. and like kind of feeding that back into the system. But yeah, this sounds incredibly useful.

Chris Gmyr (45:06)
Yeah, thanks. And there's a little bit of that built in. So I was running a bunch of trials within the Hive project, but I kind of extracted all that stuff into this new project. So anything that is like, well, it like here's a strong like adopt, like you should do this. And then here's things that you can like trial out. So we've been doing a lot of trials on just like a week or two of like haiku five point five came out.

Let's use that for a bunch of sub agent, like read-only, like non-reasoning tasks to see if we can like cut our costs. So there's like a two-week trial on that. hey, let's use this, like change to how we do like plans and the plan template using some like different frameworks. let's put a trial on that for two weeks. So we can like then kind of look back at like the sessions, the projects, the tasks, the whatever over that like trial period and let's

Yes, that's, really worked, or no, this didn't work. Let's, toss it off to the side. But everything is also getting tracked in this project as well, as like basically journal notes, entries for decisions and research logs and all that stuff. So we can basically go back and see like what what changed, what we adopted that didn't work out, or or vice versa, something that was maybe like parked or skipped, but it we actually did a trial, it worked out, and then now it's like

fully adapted. So hopefully a lot of that stuff will be built into like the system as well.

TJ (46:31)
Yeah. That's so neat, man. I'm I'm really digging it. You're you're crushing it on projects right now, man. Yeah.

Chris Gmyr (46:37)
So many projects. So many projects.

cool. So we are at time. I feel like we have a lot to talk about with Hive,

but maybe we can leave that

TJ (46:44)
Yeah, we c we covered a lot.

Chris Gmyr (46:46)
there.

TJ (46:47)
Let's yeah, let's leave it here and we can we can pick up with Hive and some other stuff next week.

Chris Gmyr (46:52)
All right, sounds good. So thanks for listening to the Slightly Caffeinated podcast. Show notes and all the links and social channels are down below and also available as slightlycaffeinated.fm. If you have questions for us or have content suggestion, go to the ask a question page on our site and we'll feature it in an upcoming episode. Thank you all for listening and we'll catch you next week.

TJ (47:10)
Thanks.