Reporting from the Vienna Calling FileMaker Conference

Reporting live from the Vienna Calling unconference: five conversations from the only European FileMaker unconference, recorded on the ground in Vienna. Organizer Philipp Puls on why the format works, Charles Delfs on rebranding to Klai and what an actual AI strategy looks like, Johan Hedman on the next big European conference, Vincenzo Menanno on technical debt and using AI to analyze your code, and a turnabout where Javier Durá puts Matt in the hot seat.

⏱ CHAPTERS
0:00 — Welcome: what is the Vienna Calling unconference
0:24 — Philipp Puls (72solutions): inside the unconference format
13:50 — Charles Delfs (Klai): the rebrand & an AI strategy that isn't just "chat"
36:42 — Johan Hedman (Square Moon): EngageU goes three days
41:33 — Vincenzo Menanno (Beezwax): technical debt, iteration & AI code analysis
1:00:52 — The tables turn: Javier Durá interviews Matt Navarre

🎙 IN THIS EPISODE
• Why an unconference is capped at ~100 people, and the "if three people care, make a session" ethos
• FileMaker's collaborate-over-compete culture
• Charles on why "code as product" is fading in the AI era, and governed, observable AI access for non-technical staff
• Semi-deterministic / hybrid apps, FileMaker as a scalable backend, and the new Klai Run (on-prem, partly open source)
• Johan on EngageU expanding to three full days in Malmö with a business track
• Vincenzo on defining technical debt, putting complexity at the right layer, and recursively analyzing whole script trees with an MCP
• Measuring real performance: the hidden cost of open-record and set-field steps
• Javier interviews Matt: the Navarre.ai rebrand, free weekly AI classes, and why insert-from-URL + the web viewer are the real AI unlock
• What's next: OpenClaw / Hermes as an "ultra robot" with a FileMaker database as its memory

Links
72solutions (Philipp Puls):  https://www.72solutions.eu
Klai (Charles Delfs): https://klai.studio
Square Moon (Johan Hedman): https://www.squaremoon.se
EngageU conference (Malmö, Sept 30 – Oct 2): https://engageu.eu
Beezwax (Vincenzo Menanno): https://www.beezwax.net
Afterdata (Javier Durá): https://www.afterdata.es
FileMaker Magazine ES: https://www.filemakermagazine.es

Transcript

SPEAKER_04 0:00

Claris Talk AI. Reporting live from the Vienna Calling Unconference in Vienna, Austria. I sat down with several of the people at the conference and the organizer Philip and talking about what this conference is about. Enjoy. Vienna Calling, I'm here with Philip. This is your event, sir. Yes.

SPEAKER_00 0:24

And a very fun event it is. It is. I really enjoyed doing it.

SPEAKER_04 0:29

I I don't know why this is my first time attending, because Vienna is my favorite city in Europe, because classical music.

SPEAKER_00 0:35

It is. Lots of classical music options here. In the summer, it seems like much more touristy, but you know, nowadays Vienna is a very touristy place anytime. But during the summer, especially in August, we have that uh open-air film festival on Rathausplatz where you have operas every evening. So that might be a time to come next time. It sounds so lovely.

SPEAKER_04 1:00

This event has been amazing. You pulled 70 people or so this year. Yep. Um from all over Europe, some from the US. Um I'm the only one from Greece, which is weird. Uh but it boosts our statistics. It's yeah, it's true. That one extra checkbox for that one extra country. And I would say it's an unconference. How would you describe it briefly to the people who need to know about it for next year?

SPEAKER_00 1:25

An unconference is a format where we provide a safe space for your ideas, your grievances, your uh hopes and and ideas. And uh every participant is expected to chip in. Like uh it's not a conference to sit in the back and uh uh dabble on your BlackBerry. Oh god, that's how old I am. Anyway, uh but it's a conference uh that really intends to get you going. Like either you present or you have an opinion in a discussion or you or you show what you've what you've missed last year or whatever. So everybody can chip in their own sessions. The calendar is open, you just pick a room, you pick a time. Uh you you might have an idea during lunch with a couple of of other participants that you join in. And uh once there are three people interested in a in in whatever, yeah, you can just pick a room and book a time, and maybe five other people will join in. And even if not, you'd still have a nice place to discuss what what you have to discuss with your three people.

SPEAKER_04 2:42

Yeah, that's one huge difference, right? It's it's not a curated thing. You're not you're just making the space available because you know that there's going to be great content. Yeah. And some of it's along the same lines of what you would see at other conferences. Um my session, for example, was something I presented on a couple times before.

SPEAKER_00 2:59

Sure.

SPEAKER_04 3:00

Um, and I think several of the other ones were as well. Vincenzo Manano um on technical debt. Yeah. A subject that he knows a lot about and cares a lot about, and is really important for the community to see.

SPEAKER_00 3:12

I think those issues are all not just very important, but they're also initiators for the other discussions that go on. So we always need a few of those more or less standard uh uh presentations to get people going. So normally on the first day, it's more or less a one or two-track uh conference. Yeah. Uh but then it spreads out because then the discussions happen in the cafeteria, in the in the bar in the evening, then you decide to make a session the next day. So that's also why we have a three-day format. Because if you were just warming up on the first uh and leave, it would just be another conference. I was actually sort of unsure like today, the last day, when it's supposed to end. We will we will have uh I planned the closing session today. Okay. Uh and we'll have it at uh midday, uh which is approximately uh the same every year. But we can still switch it. We can still move it back or forward as we please. But the calendar has filled for Saturday now already, and I think people are are good to good to be released into the wild again.

SPEAKER_04 4:28

I think I might uh this is one of the podcast segments I'm doing, but I want to do some more, so I might actually just put a little segment on there, like come to that, and we'll do a 10-minute podcast episode.

SPEAKER_00 4:40

Um I think the the real big difference for Vienna Calling or unconferences in general, but Vienna Calling is the at least only European unconference in FileMaker uh at this time, uh, is that you uh do not get people on stage that you normally see, but you get people on stage who maybe have never spoken at a conference before. There's been a few great ones like that. And and those people are genus. Uh not everybody is is in his or her setup prepared to speak in front of 250, 500 people. Uh not everybody wants that pressure, seeks that attendance, uh but uh every one of the attendants here uh brings knowledge, brings uh their um their experiences, their customer stories with them. And I have not seen a single one on the conferences that hasn't at least brought uh one or two great and amazing stories over coffee. And if they are good and amazing for three people, make a session. Yeah, that's always the answer. That's really interesting, make a session out of it. And you don't need a presentation, you just open your laptop, you're you sit down, you talk. Yeah, yeah. Presenting just means putting your idea forward.

SPEAKER_04 6:20

And also the length is something you can pick, right? So like if your idea is half an hour or an hour and ten minutes or whatever.

SPEAKER_00 6:26

Yeah. Yeah. Uh if it if you just want to get something off your chest, you do a 15-minute session. I think that's the smallest that our calendar allows, and I think that makes sense. What are your plans for this conference for next year? Um we are expecting to be uh around the same time. Um end of May, beginning of June, as we were historically with.fmp Berlin. Uh the spot is is more or less fixed in the calendar for 15 years now. So we intend to keep it because people expect it uh and might have might have arranged their travel plans for ages now around that. Um we we hope to gather um a comparable crowd. Um the the on-conference format we think is best kept at a hundred participants. We we don't want to have more. Yeah, 300 or whatever would be different conference. And it that's that's intentional because over a hundred people uh subgroups form and communications start to change. Yeah, yeah. So those subgroups would then always stick together and not intermingle, and that's not that's not the vibe that we that we're in.

SPEAKER_04 7:47

That's different. It's organized. I don't think that happens that way. But I think at an unconference, I think you'd write. I start long time ago, I was uh involved in um hosting pause on error conferences when they were in the Ace Hotel and when the room was full, and so a couple people were in the bathtub and sitting on the floor, uh, you know, very uncomfortable. It wouldn't be good like in your 60s now. But um I don't know, may maybe that's not really true because they were it was just that great we're all in this together feel. Exactly. And share whatever you have on your mind. This definitely really captures that. Yeah, yeah, I think you're right. So keeping it around that size. I also perceive, and I'd love to know your opinion on this, this seems to be the most technical of all uh the conversations. Is that just because Germans are just more technical than Austrians?

SPEAKER_00 8:32

No, I think because we attract uh historically a very technical crowd uh and the people are very um intentionally focused on being cutting edge and sharing that without showing off. It seems a character trait, but that seems to be a character trait in FileMaker in general. Um that this this sharing and caring for each other, we could be contestants on the same marketplace, uh but for some really funny reason that might uh might be worth a sociology study, um we seem to to function differently. We seem to collaborate more and profit more from the collaboration than from competition. Yeah, that's great.

SPEAKER_04 9:31

I I don't I'm I'm wondering if that's such a unique thing, but I think in our space there's really not a lot of competition for the same customers. I know that that does happen some, but it's not the rule. Um also there's a lot of people here who are internal developers and therefore very strongly motivated to share and to learn from others who do what they do. Whereas if it was just you know partners or something, that you know, there would be more of that competition.

SPEAKER_00 9:57

It might be that, yeah.

SPEAKER_04 9:58

Yeah.

SPEAKER_00 9:58

I'm just happy that it is that.

SPEAKER_04 10:00

Like exactly, let's just not let's not stare a gift horse in the mouth, as it were.

SPEAKER_00 10:05

Yeah. Um and uh we are we we need to think now about uh the size for next year and the rooms for next year and where to what where to be. We might be in the same space. Uh the the museum's quarter here is uh lovely central. I have it kind of described the place, right?

SPEAKER_04 10:25

Like it's right right in the dead center of where the museums are. So across the street from like the uh Natural History Museum, uh Art Museum, close by is the Albertina, which is one of the best museums in the world. Um and also just all the other culture here. And the room that we're in, you look out the window, and that's your view. You have this unbelievable view of the center of the city.

SPEAKER_00 10:48

Yeah. It's it's Baroque Vienna, uh right from your windowsill. Yeah. It's beautiful and it's calm and it's easy to reach, and all those are perks that I really like. On the other hand, uh the rooms are a bit too conferency for our taste because we've we have good beamers here and the seating is very st or seems very strict. Um and I think we could enhance the unconference feeling by a bit of a a more low-level uh venue uh that gives more of a workspace vibe than a conference vibe, and that might change dynamics again.

SPEAKER_04 11:33

Yeah, and actually something we're gonna see where there's like a I don't know, in a hotel, but one of the things that was great about Paws and Era was it was actually in a hotel, so a lot of people could stay there. And so there was a lot more of that.

SPEAKER_00 11:45

So, like my where I'm staying is an Airbnb type place, uh, I don't know, two kilometers away because for room rent pricing and um but there are plenty of hotel options in five minutes distance here. So um it's a personal choice where you go like one central, I think would be. It's just it's just one of those points where where we are able to keep the conference cost, the conference ticket cost really low. Yeah, exactly. Uh is that because we don't have to care for your hotel, we don't have to care for your travel plans.

SPEAKER_04 12:19

And yeah, you also don't have to guarantee by paying like a lot of other conferences, they have to uh have their contingent paid for contingent paid for 70 people, but if we only have 50, then we're on the hook to pay for a bunch of hotel rooms. That's a lot of risk.

SPEAKER_00 12:35

Exactly. And we're a small team, uh, so we have to mitigate that.

SPEAKER_04 12:40

And a great team, the four of you.

SPEAKER_00 12:42

Yes, yes. It's beautiful. It's great working with all four of them. Uh Stefan, my my partner, uh is is in that with me now for 15 years uh or about. Um uh Julia and Emmanuel are very strong, uh very uh very focused and and great developers and organizers. Uh we wouldn't be able to do all that without them.

SPEAKER_04 13:10

I only get to see their organization amazingness so far. Uh but yeah, just presume they can also be great developers too.

SPEAKER_00 13:19

So we're very lucky. It's a good team, and we are we are uh uh in the right size where everybody still works on projects and you don't have too much overhead in quality control or or uh that dishing out uh tasks, but you get you get to program yourself, you get to you get involved with the customer. So we all enjoy that.

SPEAKER_04 13:42

Yeah. Well, I'm very much looking forward to it next year, we as we round the last day here. Thanks so much for putting this on. It's awesome. Thank you. It was a pleasure. I'm here with Charles Delps. So you have rebranded your company to Clay. Let's talk about that.

SPEAKER_06 13:58

Yeah, so this is something that's been actually in the works for for quite some time. And with the advancement of AI over the past few years, it's something that we've seen the writing on the wall where we're going with generation of software. And I mentioned yesterday in one of the talks, the thing that you offer, if you know, as a developer, if we're offering code and that code becomes highly accessible, we don't have a product anymore. And that is the way I think with every developer we have to really sort of think the way or change the way we think. So we've been doing the same thing. Uh, internally, we're thinking, how can we change what our offering is? And FM Better Forms has always been a product that has given the developer access to advanced technology without having to learn that technology. And uh as such, that we we we realize that now we have access, yeah, every every every developer, not even a developer, every citizen, has access to the same technology stack now, for the most part. That doesn't mean we don't have a product. So we have to say, well, what is it that we can offer? What can we bring to the table that the average vibe coder, which I personally hate that that term, um the average vibe coder can can can um can use. And that is a few things. It is infrastructure, is governance, right, and is enterprise grade ability to to host applications. Like deployment, yeah. And deployment, yes.

SPEAKER_04 15:40

Which is not easy, uh, you know, as someone who's vibe coded, whatever, uh several different applications. We were talking a little bit about a little bit uh ago about um using Superbase. I set up uh a starting application with Superbase, and uh it set up without me intending it. Every table was open to the world, like sort of by default. And then it gives me like a warning of all the things you can fix, and they can't be fixed by a prompt. I'm like, ah, so that's exactly that's one of the things that um that you I actually see some other things you really bring to the table.

SPEAKER_06 16:14

Well, there's a few, there's a there's a few more. Uh let me just sort of dig into that a little bit deeper. You know, um we were talking, had a really interesting discussion uh the other day with uh uh Nicholas Franco, and we talked about you know where is what is what is useful, and it's a little bit maybe slightly different opinion, but if we want to have an AI strategy for your business, right, it's not just about buying software. It's not about I'm gonna buy the latest pipe drive that has AI in it, or I'm going to buy the latest product, or I'm gonna get everybody a subscription to Cloud Code because that cannot scale. And you don't want uh you don't want interfaces that are all always chat based. You know, I I see I look at the Filmaker community, for example, and we see people people um demonstrating, demonstrating and saying, hey, look, I can connect an MCP up to my application, I can I can chat, and I can get my sales report. But that sales report is a sales report that you get every week. Why are you generating that? Right? So we see the future as being more what you may want to call uh semi-deterministic or hybrid type apps.

SPEAKER_04 17:23

Semi-deterministic?

SPEAKER_06 17:24

Yeah, so so if you think about I'm typing a prompt into to to your to your um quad code or or GPT of of any sort, and I'm saying generate me Monday sales report. Right. That goes, it queries the database, it's generating some HTML, and it's it's rendering this report. I don't know if I call that semi-deterministic.

SPEAKER_04 17:49

No, no, that's fully determined. That's fully determined.

SPEAKER_06 17:51

Well, that's no, that's not that's I would cut I wouldn't say that's deterministic at all because we're relying on file. Okay. But what we really need is that person who's generating that sales report, let's call them the uh the sales agent. They need the ability to have control to be able to generate that stuff, but not generate it, first of all, they need to be able to generate safely at an enterprise grade. So that means somebody who's not a technical worker has to have full access to the that technical stack. They don't want to generate that code every single time, every single Monday. Right. That's not practical either. So how do you solve that? Well, you need to be able to give that sales agent access to technology. I need to be able to give the receptionist access so that so that they can go and say, hey, I want to I want to connect all of our company's contacts and send them out an invitation to an event, right, and build a little app for that. Right? That's a that's a massive security breach in a reasonable-sized company that has any type of technical infrastructure. So what we're leaning to solve is to solve that by bringing access to the technology to that person, but have that in a highly governed and observable manner. So now they can say, hey, I want to have access to our uh pipe drive, for example, to uh gather all of the contacts, I want to have access to Notion, I want to create some kind of user interface, and I want to send it out through an email. So those are four, uh one, two, three, four, three or four conduits. Right? Those conduits are not controlled by that same AI. And that's the fundamental premise of where we're going. Bringing that access to that level of AI integration, AI agentic engineering, but I'm not gonna say the word engineering in this particular case because that person's not really thinking that way.

SPEAKER_04 19:45

No, but that's what's ended up happening, I guess.

SPEAKER_06 19:47

Yeah, yeah, yeah. And bringing that all the way to the end user, right? And that's I think where the future of AI is gonna be. For companies who have uh a true AI strategy.

SPEAKER_04 20:00

Yeah. I think you said something the other day uh that I really loved about sort of the definition of what an AI strategy is. I really like some of the words you had there do you put it.

SPEAKER_06 20:11

Yeah, you know, uh it's it's I think I think if you don't have a strategy, you obviously don't have a strategy. Right. But I don't think it's just throwing out AI at every single problem, and I don't think it's the chat interface.

SPEAKER_04 20:24

Yeah, so like the if you have a if you use a chat interface and you throw AI at every problem, you don't have an AI strategy. It was, I don't know, you put it really poetically, but I can't remember the way you did it with it.

SPEAKER_06 20:34

Um but yeah, I I think that's uh you know, uh there's a demonstr there's a uh we're we're here at at uh um Vienna Calling, and there was a talk, yours, that you mentioned um you showed you showed a number of really simple examples. I think the AI right now, you know, it's such a massively new thing that it's really, really interfaces with everything. But like with any technology, it's gonna start to to subside and it's gonna fade into the background and it's gonna be just part of our fabric of ritual. Yeah.

SPEAKER_04 21:11

And definitely what I'm what I'm uh focusing on, like with the free classes I do and stuff like this, is to build little things. Uh you know, buttonize it was a theme I was using early on, um, so that they just become part of it, just like all the other functions do. I think those are the biggest wins, and people aren't doing them. No, I know. Right? They're not doing that. Well, they are starting to. Like in Europe, I think they're, you know, the polls I've been taking, I think there's actually some things. But but also I think people are doing really big AI implementations, which are much harder, much more, I don't know if I'd say brittle, but a lot more high maintenance.

SPEAKER_06 21:43

Um there is a vulnerability or uh a fragility, I guess you could say, in in the big ones. Like we're we've done we do consulting as well.

SPEAKER_04 21:52

And yeah, see, that's actually one of the other things I kind of wanted to get to you. The huge the huge benefit that you bring to your clients at Clay is Is your great development staff and your personal interact, you know, working with customers and whatnot to make sure that they're getting all of the security and all of the right answers and all of the right technology and all that stuff so that they so that someone doesn't make the pretty big mistakes uh of vulnerabilities. I don't know.

SPEAKER_06 22:18

From my perspective, you bring that to the table big time. Well, you know, we have a lot of experience running infrastructure. Yeah. Right? And running infrastructure, running a software as a service, and the consulting side. And that's why our our new direction, when we changed the company, we said, well, how can we where are we going from here? Right? Where we our old uh our old product, FM Betterforms, was our it was a terrible name. Yeah. And it was really a working name. Yeah, it was. But it started becoming known throughout the community. So it's people just started calling it BetterForms. Yeah. I don't know. Yeah, we and we've totally dropped the FM part. Now that doesn't mean FileMaker is not a first-class citizen, but it's not the only citizen. And FileMaker understands this as well because you know half of the presentations I see on FileMaker now are all about integration. That is probably a good 60% of everything.

SPEAKER_04 23:11

Integrating with other technologies, other systems.

SPEAKER_06 23:15

And we we see that as well. So, you know, why don't we integrate our platform and our uh our expertise into that and try to predict, knock on wood, try to predict where where our future is going to be with AI. So that's that was the that was the impetus to uh to to change, to change not only the name, but also the direction. So yeah, and this was only what, a couple of months ago, the change? Yeah, we I guess we did it what about two months ago or so, a month and a half ago, right? And so it's it's an ongoing process as we as we restructure things.

SPEAKER_04 23:51

Yes, yes. Canadian access. You can always nail the Canadian accent.

SPEAKER_06 23:55

I love it. Yeah, yeah, yeah. I still think about is sounds normal. A boot. It's not a boot. And you're gonna hear this on the recording as well. It's about.

SPEAKER_04 24:03

That still has still has a little bit of an accent to it.

SPEAKER_06 24:06

I should hope so. I need to differentiate somehow. But I mean, I don't know.

SPEAKER_04 24:09

From California and Pacific Northwest, I have probably identifiable accents too.

SPEAKER_06 24:16

I think just going back, sorry if we could though, just going back to the small, the buttonize it sort of strategy. You know, I think a lot of the developers are only or maybe not even doing that as much as they could be, because it's not as exciting. No, it's not as. You know, it's it's five lines and five lines of script in FileMaker, which I think is absolutely incredible because there's no other environment that I know that you could put maybe four or five lines and get that kind of result with.

SPEAKER_04 24:45

Yeah, I kind of go back, like one of the examples I use is the uh export to PDF script step when that came in FileMaker, or the send mail script step. You know, like the export export PDF in one step could like, you know, set a file on disk, create it, put data into it, have it everyone in your pages, and create an email and attach it. I'm like, wow, that's you know, that's kind of an example of of some of the stuff I'm trying to go for. Um I'm gonna turn this right back around on you though, because one of the other things that you really, really offer in your, I don't know, I don't want to be all cheering about your your system. But you can actually, and you do bring a lot of these little buttonizer kind of things automatically with your technology because they're so much easier to integrate in the modern web world where you live.

SPEAKER_06 25:30

We do. We have a lot of things. Um Clay Studio is is now FM BetterForms, right? Or sorry, yeah, FM BetterForms is now Clay Studio. And Clay Studio has a lot of um a lot of history built in um based on on features that people need. Right? We have things like messaging, so real-time, real-time messaging. Oh, yeah. Either peer-to-peer. You know, if you really wanted, we've had uh we've had some of our team build pong games that that you know. Yeah, real time. You know, real time, real-time messaging. Um system optimization caching, so d to deburden FileMaker. You know, we have one user, he's got almost a million users. You know, a million users powered by you know, effectively one file maker server. Yeah. That's pretty crazy. Yeah, but you know, you're not you're not gonna see that with the most I've ever done was 900,000.

SPEAKER_05 26:23

I'm kidding.

SPEAKER_06 26:25

That's pretty good. That's that's some serious scale right there. So, you know, then maybe there's not they're not always on concurrent, like, but they have around the clock, around the world, there's you know huge, many, yeah, hundreds of people, hundreds of people all the time. And that type of infrastructure is something that that's it's not a buttonize it thing, but it's it's a type of thing that that it's a solved problem, right? And it's not something that you're gonna just vibe code out into an app. No. Right? That that that uh that experience. So that's one of the things we bring there. Um going forward now, what we're gonna be selling is, as I mentioned, it's the gonna be the governance, it's gonna be the ability to have code repositories, uh, safe deployment of AI code, bringing that code really, really close to the end user, which is the person who has the knowledge of the job. So the domain worker now suddenly can create the things that they want to create safely and enterprise. Your IT department's gonna be super, super happy with that because they can control and they can monitor everything. There's no unknowns there. There's transparency into the entire stack, everything is highly sandboxed as well. So we'll start with that, and the biggest, most exciting thing is it'll be able to run on-prem, and it'll be some of it will be open source, you're not gonna be blocked to our company. You're generating standard code and you can take it away with you as well.

SPEAKER_04 27:50

Yeah, that's an amazing, that's another awesome kind of file maker benefit that is really otherwise hard to do in a lot of these platforms because they're just cloud, period, you know.

SPEAKER_06 27:58

So that's where the direction is. Um that's gonna be part of our second product. It's called Clay Run. Uh it'll be released, it'll be released next quarter, roughly. Okay. For for beta. Um, this is not something that we can just vibe code, and it's not that we don't have the experience to, right? We we do all agentic engineering internally for for new development now. Um, but there's so much thought that we have to put into architecting the the infrastructure, the fabric that we're building. So yeah, for sure, for sure.

SPEAKER_04 28:27

And that's yeah, again, yeah. I think one of the really big clear messages is what are the limits of of this vibe coding? And one of the things I really love to quote is you can vibe code something and you can like bring it to the customer, look at it, and it's like, oh my god, this is 90% 95% done, and then you realize that you have another 95% to do to finish it.

SPEAKER_06 28:47

Yeah, yeah, there's uh there's a double proto principles kind of thing going on there. And you know, vibe code, and and this might change, but today and probably for the next year, we can get to we can get to V1 really, really fast, right? We can get the version one of a of an app by vibe coding.

SPEAKER_04 29:06

It's really well that's actually what I'm saying, is it's not really realistic. No. Like, like when you get, you know, my joke, which I think is actually really true, is you you vibe code it and it looks like 1.0. Yes. But then when you actually um, you know, when it meets the enemy, um you realize that there's another 95%. That last five percent is not the small part. That's the you know, that's the the proof of concept now when you get to that.

SPEAKER_06 29:31

When you get to that. We can get a proof of concept out the door.

SPEAKER_04 29:34

Yeah, quickly.

SPEAKER_06 29:35

And the proof of concept is totally fine if you're not exposing it to too many people, if it's if it's there's no nothing exciting to be hacked in in it or anything like that.

SPEAKER_05 29:46

Yeah.

SPEAKER_06 29:46

Now, um getting to version 27 is a whole different story. Yeah. Right? And you can ask any engineer that's worked on that's using AI on a serious app that has you know a sprawling code base, and sprawling not because it's old, but because it does a lot. Sure. Right, you can't just easily implement the feature. You can't just go and say, hey, I want to add this new feature. I mean, uh arguably sometimes.

SPEAKER_04 30:11

Yeah, I haven't done it at that scale. I would think that if the if the app is properly structured, well, I don't know, you can tell me how I'm wrong here. But I mean, if if you're doing it with a modern, not necessarily modern, but like a normal way of development where everything's on GitHub and you're checking and there's all segmented out, and and you can, you know, uh check out and check in code segments, and everything is version controlled. Um and when you're working on a new feature, it's a discrete thing. Yeah. You can you can test it in a dev environment. These are all things that you can't really easily do with FileMaker. Um, that it would be more possible. And I I would also say that as AI comes out, there would actually have the ability to do things like look at the entire code base and kind of factor what the other thing uh Yeah, well uh let's just to pause there, you said look at the entire code base.

SPEAKER_06 31:01

Yeah. AI skims the most LLMs or most agenda cardnesses will skim the entire code base. Yeah, yeah, yeah. Okay. So they'll say, okay, these are all your APIs, these are all your services that you have coming into your your back end, let's say.

SPEAKER_04 31:13

So that's actually not obvious, it just skims it as deep analysis.

SPEAKER_06 31:17

Correct, right? So unless you're looking at each one of those, you know, you're looking at the contracts for every one of those services that your backend is is is providing. And I'm talking specifically about backend because front end, yes, there's vulnerabilities, but the front end is fully hackable if you're talking about a browser. You can open up everything, you can inspect everything. Yeah. A lot of people don't realize that. They think the security lives there, but the security has to live in the back end. Yeah, yeah, yeah.

SPEAKER_04 31:42

Which is another thing FileMaker very much brings to the table, right? The security at the back end, you looked dubious.

SPEAKER_06 31:49

I mean, well, I I think it does, but FileMaker has a few benefits there. I mean, it's pretty opaque, right? Like we we we at these conferences and things like that sometimes talk about security, and it's it's it's uh it's pretty walled off that we don't see exactly how everything works. And I think I'm not saying security by um uh obscurity is a good thing, but knowing less definitely is better than having something um yeah semi-open. Fully open source is a different story.

SPEAKER_04 32:21

Yeah, yeah, yeah. I guess I'm thinking about like like security up to making sure that your your clients only have access to stuff that they're supposed to see, no one can go in and delete all the data.

SPEAKER_06 32:30

Um so this is interesting. So there's some there's some you know talk. Somebody mentioned in in the most recent talk, they said, well, how come FileMaker doesn't just rewrite their stack into Mongo? Yeah, or not Mongood. And it's not yeah, and yeah, first of all, it may not be the best, yes, best database. Um and it's just a popular one. It was the popular one a few years ago. Superbases was the popular one when when a GenTech uh um companies like Lovable and such came out.

SPEAKER_05 32:59

Yeah.

SPEAKER_06 33:00

And now we're starting to see a move right to pro uh to Postgres, right? Uh even though it's a derivative. So, or vice versa. And Postgres is actually number four on the on the database charts right now, right? I think Oracle Oracle's one and and I think they're like the top five. So that changes from time to time. But I I wanted to ask them is what is it specifically that you want out of that database that you don't have now? That you don't have in FileMaker? Because we've been using FileMaker as a database purely as a back end for the last eight years. That's how since uh F better farm's inception. And it's been a phenomenal, phenomenally great database. I mean, there's a couple quirks, but you have those quirks in any any system. It doesn't matter what database that we're we're using, and you have to learn how to work around that.

SPEAKER_04 33:52

I mean, I to my answer to that is pretty clear, and that's FileMaker is a single point of failure, FileMaker Server is a single point of failure, and if you have another system that is an appliance, um like Mongo and systems like it, everything is redundant, it's kind of always up, and you can you you know you don't have a single thing that you stop and start. Better or worse, but I mean that has to be an issue for that client you have, it has a million you know users hitting it. Um, you know, there's a there's necessary times a file maker server has to be brought down and brought back up. Um, although in your case, there's probably really good ways you can work around it because you can because you control the front end. But like in a normal file maker service, when people are accessing it with the file maker client, um you know that that's definitely an issue. And I'm not saying it can't be worked around, but but like me, what I've dreamed about for years was a way around that where I didn't have to have uh to have the expertise and worry about file maker server, because I think that's kind of anyway what I've dreamed about different.

SPEAKER_06 34:54

Yeah, I think it's the fact that we do have really deep knowledge right there is when something's off, something's weird about performance or something's weird, like we have accessibility into it, we we understand it. Yeah, right? We can contact other people and so on. So so I think anyway, yeah, just returning, file maker, I think it's a great backend. Yeah, yeah, yeah. And you know, even for the companies that are that are thinking, well, well, maybe we should switch to something else because I was walking through the airport and I saw Oracle or Salesforce, you know, you know, I I I think we have to start to realize now that there's amazing things that we're gonna be able to do now, even more so with just a combination of integration. Yeah. Bring the web outside, you know, whether you're using Clay Studio or you're scratch coding something, whatever you're doing, um, or a new product, which I'll maybe sometime in the future we'll talk in deep in depth about it. But um it's been it's been great. And I'm actually pretty pretty positive, pretty optimistic about the future there.

SPEAKER_04 35:54

I mean, I really am too. I mean, I was you know fasting a little trade or whatever, but uh I definitely see a huge advantage to having FileMaker Server back in his earth. And and you're like we were talking about, it's been your it's been a pillar of success for you for these last many, many years.

SPEAKER_06 36:12

So yeah, and and hundreds of our users.

SPEAKER_04 36:14

Yeah, exactly. Yeah, yeah, yeah. It's always great to talk with you. I'm so glad to to see you. Thanks for for coming all the way over here. Uh thanks again for your time. All right, man. Cheers. Yeah, and it's always nice to catch up, sir. Likewise, likewise. So uh you're here at Vienna Calling, and then I my favorite conference in the world, of course, is engage you. I don't want to like, you know. Thank you very much. Yeah, yeah, but I wanted to like, I guess we have two questions. Like, what did you think of this conference and what is the big plans for the changes coming up?

SPEAKER_01 36:42

So Vienna is always nice to meet up with fellow uh developers, but also uh it's an on-conference conference, so it's a little bit different to what we do at Engage U. But I love the atmosphere and and that's just looking out of the picture, being in a historical park, museum quarters, it's fantastic.

SPEAKER_04 37:01

Yeah, for sure.

SPEAKER_01 37:02

But uh yeah, we do have something else. We do have what I think is the biggest English speaking uh clearest platform community conference taking place in the fall.

SPEAKER_04 37:12

It doesn't have to be what you think, it actually factually is.

SPEAKER_01 37:15

It might be, it might be, yeah. But I I think that's gonna be the case this year. And and therefore I think uh we've raised it to become three days, three days of a conference, three full days. So first I will have one full track out of three that will be focusing only about businesses because we get so many requests uh from partners and from players to talk of what do we actually do with the web platform? What what are our benefits? And now with AI, where should we go? So we've started to talk to clients, we've started to talk to customers, and then starting to invite them to our conference so that they can actually see what we're up to. Yeah. Um, so we're looking forward uh to see other things coming out of this conference, not only technical parts.

SPEAKER_04 38:03

Yeah, that especially makes sense to you uh running a large partner like a platinum partner company because you have such a direct connection to so many businesses from having helped them over the last 20 years. I don't know how long you've been around, long time.

SPEAKER_01 38:15

Yes, very long time.

SPEAKER_04 38:17

Yeah, yeah, yeah.

SPEAKER_01 38:18

Uh so yeah, so that's day one, then what? So day one will still be like two tracks of tech. And Claris will be there with full five people talking in and out. Uh they will be present throughout those three days. Nice. Do you have the right five this time? Or everything. Yeah. So uh we have this perfect spot in Malmou. It's three minutes' walk from the train station. You fly into Copenhagen, you take a minute uh train and leaves every 20 minutes, and then super central of Malmö is a super nice hotel. News, conference, uh, and uh the hotel. You stay at the same place. We eat at the same place.

SPEAKER_04 38:56

We I talked about it, I've talked about it before, but like the things I love, love, love about DEF CON from a long time ago was that everyone is there. So the social aspect, as good as the conferences and the sessions, there's there's so much more benefit, honestly, in having the chance to connect with your peers and get ideas and ask them questions about their presentation or just compare notes until, well, when I was younger, two, three, four in the morning, whatever. Now I I stop at one.

SPEAKER_01 39:25

I think when we were younger, it was hard to grasp everything from a compliance because there was so much new. Right. You went to comp at different sessions and then you're trying to grasp it all, but it's just becoming like when you try to produce a sausage, that the thing was just big enough for one. But nowadays you're trying to grasp these small things from different sessions. But the overall is to meet people and see what they are doing right now, where are they heading? Yeah, what they think about AI, what do they do with servers, whatever options they have on the table right now.

SPEAKER_04 39:55

You know, it's funny, I as we talk about it, I only just now realize that there was obviously in the younger part of my career, there was times going to conference where I was learning, learning, learning, soaking up information. Then, from having been a developer for so long, there came a point that I the sessions didn't have a lot of new content for me because of the amount of stuff that I've been doing. And know what? Now I'm learning, learning, learning again because there's so much new information.

SPEAKER_01 40:21

Um it's definitely a new chapter with AI in so many different directions.

SPEAKER_05 40:26

Yeah.

SPEAKER_01 40:26

Uh and now that we also see that Clarice is making some small steps towards what we are kind of looking forward to, I think this conference would also be locked by the new stuff that's hopefully gonna be presented in that conference.

SPEAKER_05 40:40

Yeah.

SPEAKER_01 40:40

Um, so I hope there will be a nice debate going off after the release of 26 and all the way up to our conference gets started of what's gonna be there, what do we need, and what can we do? And then it will be uh like a used party of like new things brought into the table to do with AI. What are the dates of the conference and the website? Yeah, so we start on the last day of September, so 30 of September, and then the two first day of October. Okay. So that's a Wednesday. So get in early, so then we can show you around in Malmore or Copenhagen or whatever you'd like to say, and then stay the full week and enjoy it.

SPEAKER_04 41:14

Yeah, staying the whole week is the way. And uh, for the listeners who are in the US, North America, get over here. It's you really need to come over to Europe and go to this conference. And like, even I don't know, there's just so many, so many reasons why it's great to get over here.

SPEAKER_01 41:29

Thanks for your time, it's always great to chat. Thanks, always a pleasure.

SPEAKER_04 41:33

I'm here with Fendi. Ciao. Ciao. Hello? Oh, yeah, yeah, of course. So uh conference here in Indiana? Uh so you uh watching your current role with Packs. You're you have an awesome title, don't you? Chief Ieration Officer. I like that. What do you take from this conference? What do you what do you think?

SPEAKER_03 42:05

Um, I gave a presentation on the first day, and uh my favorite thing is a lot of that topic. And one of the themes was about iteration. Iteration is still a superpower for us. It only gets amplified with the effects of AI. I would say that uh a lot of us that are exploring AI, like as many people said, like Amy Chris yesterday or this morning said, just be curious and keep exploring, keep learning. And coming to a conference for me is also an opportunity to learn. I picked up quite a few things that I didn't know. And uh and and as in the uh the post that uh Jay Wellshofer did on LinkedIn, which I noticed like a couple of days ago, and I pointed out one of the things he says kind of like argue for uh you know your point of view strongly, but hold it loosely in a sense, you know, like you're passionate about what you believe in. Strong opinions loosely held. Yeah, exactly. So, you know, there's some points of view here that were shared that I don't really agree on, whatever. But you know, and uh you point you you point your your point of view out, and um yeah, you you see how it goes. Definitely learn some new things.

SPEAKER_04 43:24

So iteration's one of your favorite things. What about some other ones?

SPEAKER_03 43:27

Technical debt.

SPEAKER_04 43:28

Technical debt? We've never talked about that before.

SPEAKER_03 43:32

Yeah, uh it's been a fascination for me for a long time.

SPEAKER_04 43:37

How would you define it briefly?

SPEAKER_03 43:39

What is technical debt?

SPEAKER_04 43:40

How would you like a brief description of it?

SPEAKER_03 43:42

I mean, uh it's it it was coined by um the person who uh uh no, I forget his name, but uh long time ago? Uh not yeah, quite a while ago. And it basically defined as you you you you trade moving quickly for uh you know. Technical debt. And think of it like this like you're exploring a new idea, you build a layout, you add some fields, and then you you add all these fields by by like a number, like almost like repetition fields, but they're numbered, and you think, oh, I'm gonna do this calendar thing, and I'm gonna have all these fields, one for each day, and then you realize, like, oh, there's a better way. And what you do is don't clean up after yourself, and you just go to this other thing and you keep going and you leave everything behind.

SPEAKER_04 44:28

I've seen solutions that had like a set of 50 layouts from the first version of it. And then when they upgrade it to the next version, they have like the v2 set of layouts, and then the v3 set of layouts. So there's like 150 layouts in there that are not used. Yeah.

SPEAKER_03 44:41

Maybe. The best one I remember is seeing a su a solution where I would see ID, and then I would see another field called the really ID, and then I would see another field called the really really ID. Oh man.

SPEAKER_04 44:55

That's so that's cringe with V.

SPEAKER_03 44:58

I I just didn't know if there was another really, really, really ID to be aware of.

SPEAKER_04 45:02

So the thing is, like, the really really ID is not the one that's actually in use. Yeah. And you gotta go look at the graph. Yeah. Sometimes I'll make like friendly ID, like for an invoice number or something. Um yeah, because it's hard, you know, a lot of older databases didn't have that.

SPEAKER_03 45:17

But you know, like in general, like I've been doing this, you've been doing this a long time. What's been fascinating to see over this so the over so many years is to uh to see that um sometimes we've had these hacks and we've we've added schema to to to execute the hack, right? And now we've seen that like, oh, well, you don't need to do that because now you have this high condition calc, or you have uh layout calculations or something. So what what is beautiful is to see that schemas getting distilled down to schema, like real essence, real exactly.

SPEAKER_04 45:53

Yeah, but I mean technical debt there, you can't go deleting you know togs and relationships and things like that that have been in the that have been in a managed database easily because finding them, getting the getting the use. This is one of the reasons why um one of the things I preach is to put the complexity at the right layer.

SPEAKER_03 46:12

Right.

SPEAKER_04 46:13

Um which is an automatic way of handling technical debt in a lot of ways. So if you use the schema and you use the calculation fields, um which is the way that it used to be, like in older versions of FileMaker, there there weren't other ways to do it other than using a calculation field. And if you wanted a portal sorted, you had to have a second, a second um relationship that was sorted. Right. And then the other ones like for different reasons.

SPEAKER_03 46:39

Yeah. Um yeah. So you know what's coming, what they showed, because we we listened to the Clara's presentation yesterday. And uh, what's coming is the ability to build web-based layouts, etc. And you know, with the beauty of web development, you can do things you never even had the possibility of doing. So that's really exciting. And I think that will also provide us with an opportunity to reduce even more technical debt.

SPEAKER_04 47:11

Um yeah, so like putting it at the script layer and putting it on the interface. Um, I think there are places where I think it's maybe easier to manage it. So let's talk specifically about AI and how it uh manages technical debt. Because I have a couple opinions there too, but I want to hear yours.

SPEAKER_03 47:29

Yeah, I think uh I think AI is definitely gonna have an impact on how we code, that's for sure. Yes, and it's and it's uh already has and we're gonna get we're gonna get proper tooling and plumbing to do that. Um I think the beauty is that we can ask it to review some code and tell us like, hey, you know, like can I make this better? Or or what's the problems with it? As a matter of fact, when I started exploring uh extending inspector with an MCP, I had uh this idea that I wanted to give it a script and then um you know I had a script that I would call through the MCP to say, I'm gonna give you the file name, the solution ID, and the script in that in that file.

SPEAKER_05 48:15

Yeah.

SPEAKER_03 48:15

And then I want you to return me all the steps in that script and then analyze the script steps and see if there's any problems.

SPEAKER_05 48:23

Sure.

SPEAKER_03 48:23

Where it gets interesting is normally like if a field is missing, you'll see field missing. Yeah. Perform script unknown, go to layout unknown. Those are easy things to find. And you can spot them by just looking at the script. But where it gets interesting is the logic, the subtleties of like, hey, you have this if block where it's checking for this condition, and further down in the script, you're doing another check, and you don't need to do that because you've already checked it and you're within that block. You would those things are really hard to catch. Yeah. And it's like, oh, okay, I can simplify my code, I can remove that, or you've declared a variable and the variable is misspelled somewhere. Yeah. And you know, those are really hard things to catch.

SPEAKER_04 49:08

Jason uppercase, JSON lowercase. Yeah, exactly. I bet that's a tough one.

SPEAKER_03 49:12

Yeah. So uh so the other aha moment I had with with AI, once I got that working, I thought, oh, okay, so I'm returning the steps of the script. And uh if I wanted to, I wanted to I wanted to have AI look at the whole execution path of the whole script. Like this script calls this subscript and this other subscript, etc. And I was under the impression, like, oh my god, that's a lot of work. I'm gonna have to give all the, I have to traverse the step the the tree structure of every script that's called and return all the steps. I said, wait a minute.

SPEAKER_05 49:46

Yeah.

SPEAKER_03 49:47

AI already knows how to fetch the steps of a script. And I said, if this is the root script I'm calling, and you can analyze it and see what other subscripts I'm calling, just recursively call the same MCP tool and get yourself all the steps for those other subscripts.

SPEAKER_04 50:02

That makes sense. Yeah, yeah, yeah.

SPEAKER_03 50:03

And then all of a sudden it's building the whole tree and analyzing the whole execution path. And it's like that was an aha moment to for me to leverage AI to get those benefits, right? So like it's yeah, it's it's it's spending the time with AI to realize that like it really becomes an assistant, like you know, a helper.

SPEAKER_04 50:23

Yes, definitely.

SPEAKER_03 50:24

Yep.

SPEAKER_04 50:25

So I guess what I I don't know, one of the things I've encountered and wish for, like in that example of the uh 50 layouts for version one, 50 for version two, some script somewhere still uses one of the layouts from version two or whatever. And so if you can't, you can't just go delete those layouts. So there's not really a way to know what what stuff is actually in production that uses some of the old stuff. So you can't just go deleting and cleaning. That's what's really hard. And I I kind of file that under managing technical debt as a way to delete the things that are that are um no longer in use.

SPEAKER_03 50:58

Well, I mean you you have to balance that that that uh uh that need or that or that uh expectation because uh like the tools that are out there, inspector, base elements, and fm perception. So with these tools you can you know look at look at where the references are to a layout or a script, right? Yeah, so that'll that'll tell you things that are unreferenced, which is great. Yeah, and that's huge. But you can also go to a layout by calculation or by number, yeah. Or calculation by name or something, right? And uh and that's indirection, right? So you have other spots of indirection. But the beauty of that is you can scan your code and say, Am I using any indirection? So your confidence level on deleting something that is not referenced goes up significantly, right? So if the tool can analyze and say, hey, I looked over the code base, there's no there's no references of you using go to layout by indirection, right? So, and this layout is not referenced anywhere, that means these are the layouts you can delete, safely delete, right?

SPEAKER_04 51:58

And uh unless some other application calls in a bit. I guess it can't.

SPEAKER_03 52:02

Well, like put it this way if if you do build layouts that are API layouts for other systems to access it, right? And a data API layout or whatever. Yeah, what you can do, and if you if you're meticulous and you know you you want to document these things as you build them, you can write a script to say go to layout. This and the comments at the top of the script, you say that this layout is only for purposes of documenting direct you know dependencies, exit the script at the very beginning, and then you add your code, like go to layout such and such, and now that layout is referenced in your system.

SPEAKER_04 52:37

I see, I see, I see.

SPEAKER_03 52:38

And so when you go to look for it, it's gonna say, hey, this these layouts are referenced.

SPEAKER_04 52:42

That never occurred to me. That is a really good idea.

SPEAKER_03 52:44

No, no, that that's it. That's a very, very useful tip. Yeah, I hope.

SPEAKER_04 52:47

I actually put a bunch of stuff after the out after the end if or after the exit layout in the script. Like a lot of stuff that I'm kind of working on. This might work, this is another version bladder. I put it down at the bottom. But I never thought about doing it like specifically to aid documentation.

SPEAKER_03 53:02

Yeah, you could you could put it there. Uh, as a matter of fact, as we move into this new paradigm of working with AI and AI building web-based layouts, you may end up in a situation where um, oh, I I need to need a layout for this purpose, yeah, or I need a script because I'm I've told the AI to put a button to trigger a certain script in that UI. Yeah. So you can document all those in a script.

SPEAKER_04 53:30

As a matter of fact, you could probably feed I kind of say that a different way. Like as more of your application goes into, goes out of FileMaker and goes into JavaScript code and a web, um, then you can manage, uh you can clean it and manage it using AI.

SPEAKER_03 53:46

Oh yeah, of course.

SPEAKER_04 53:47

Yeah, so that's that's I think that's actually we finally have an agent that has time to go do that because we would never be able to prioritize that ourselves.

SPEAKER_03 53:56

But I'm just saying, like if you if you use a tool uh as a central place to document it and you want to be methodical, you could put it in there. But like you said, you could use AI or you know, uh this stuff is gonna be accessible, you can scour it and and uh inspect it that way.

SPEAKER_04 54:12

Yeah, cool. What are what were some of your other favorite things in there?

SPEAKER_03 54:17

Um well I I talked about timing. That's another one of my favorite topics, right? So I've been lately I've been with Hanzo and Richard Carlton Live TV, right? And uh we talk about performance. Measurement. And uh you know, I brought up the topic of measuring things and understanding the performance implications of uh of different script steps and uh and how and and so what what it's interesting about it is I showed a very simple example where I said, look at uh how much time it takes for this from this script step to this script step using get UT UTC uh microseconds, right? And uh and so the crowd answered the questions and all this, and then I showed another example where I put I bookended those steps around a set field step. And I was kind of it's like a look like a tricky question.

SPEAKER_04 55:09

It was a tricky question, which I got wrong.

SPEAKER_03 55:11

Yeah, well, I mean it was point it was pointed there so that you could see that like it isn't the screw the set field issue that is the option that's gonna take time. Which is what I'm sure it was. It's it's a it's a it's an in it's an implicit, it's an implicit on record open that happens at the time that you set the field. Yep. So like if you open the record before and then you measure the set field, you'll see like, oh, that's really so little time.

SPEAKER_04 55:37

Yeah, even on a hosted file, right? Right, right. Which which then presumably means uh, well, I guess two questions, right? Nerdy it nerding out. So if you do an open record and then a series of set fields, they're all gonna be super fast. Yeah. But what if you just do a set field, set field, set field without an open record?

SPEAKER_03 55:53

The first one takes the hit. Once the record is open, then the other ones are or the record is open now. Those ones are gonna be less less of a hit.

SPEAKER_04 56:01

So it's not there's not like an inherent value in doing open record.

SPEAKER_03 56:05

Uh there well, I mean, unless unless you you want to programmatically take the steps to know that you can open a record, etc. But you know, yeah, if you do things transactionally, then it's a whole other animal.

SPEAKER_04 56:17

Yeah, so like open transaction, probably the same kind of situation.

SPEAKER_03 56:21

Yeah, and and then you you'll see the difference, right? If you run that locally over a WAN versus you run it on the server, and that the the whole point I was trying to make is use these steps or techniques to learn how much overhead you're you're getting and how much overhead it's taking, right? Because we're talking microseconds to like a third of a second. Yeah, that's a huge difference.

SPEAKER_04 56:44

This is one of the things you and I argued about a little bit is which of these things do you measure that are academic and which of them are actually meaningful? Well, an open record is not academic, and because that one, you know, the time it takes to set a field on a hosted file is is noticeable to the user. Right. Um if you have a if you have a table, and we encountered this in older solutions, that has you know several hundred fields in it, many of which are indexed and with auto-entered, just creating a record. I've seen customer solutions take three, four, five seconds. Right. And you know, um yeah, so they're like, why does it do that? Right?

SPEAKER_03 57:20

So that's just yeah, but then the whole point of my my bringing up the topic was uh to get people to be curious, right? So this leads you to curiosity to say, oh, what if this? What if that? Like and I showed an example where I said, okay, what about refresh window? Do you think it's gonna be much more expensive over a WAN, over a LAN? They were the same.

SPEAKER_05 57:41

Yeah, okay. They were the same.

SPEAKER_03 57:43

You know, so like it's like it informs your knowledge and you become more, you know, kind of like, oh, okay, well, uh you become more confident in the code that you write because you know, like, okay, this is not gonna be impactful or or whatnot.

SPEAKER_04 57:59

Good stuff. Yeah, yeah, it's like you know, it doesn't have to always only be AI in this world. Um, kind of, although saying that I'm kind of wondering how much can you use AI to track and understand these things? But stuff like this, probably not, right? You have to really go be curious and test it yourself.

SPEAKER_03 58:13

Yeah, yeah, exactly. I think the only um the main issue also for me was, and and I I think Hanza would agree, is that currently we have to modify our code to get those insights. And I would say it would be nice in in the future if we had like some profiling tool that we or some developer enhancements where you could say, you know, profile the script and just you just run it, and then you could look and see like from here to here, how long did that take, or from here to here, and you're never modifying your script, right? And yeah, the other beautiful thing, if we could have it one day, is to turn on some kind of logging so that you could log every time a script got executed, how long did it take, did it get any errors? Yeah, yeah. So then you could like it's been running for six months. Let's look at all the scripts, digest the log or from or something, and then do a heat map on all the scripts that you have in the system, and then you discover, like, hey, this script never gets run by anybody. Maybe it gets only run once a year or some other weird thing. But so far, this script has never been run.

SPEAKER_04 59:21

So, how how close do we get that with a checkbox on the server to do the, I can't remember the name of the checkbox. It basically kind of does an audit log, it checks the performance logging. Oh, it's logging at 2000. We're not there.

SPEAKER_03 59:31

What are you talking about? We don't have that there.

SPEAKER_04 59:33

Well, there's a there's a checkbox on the server level that you can actually do a bunch of tracking and it tracks all the performance of things going on in a file maker database. It kind of uh, and I'm not sure if it logs all the stuff we're talking about, but that's a function of FileMaker Server that's been there for years.

SPEAKER_03 59:50

So but I've never seen a way to log script execution times.

SPEAKER_04 59:54

I don't know if it does that. I think it might it might track other things. That's my question, right? It might take like layout loading and other things you can you can analyze it to get a bunch of other things, and I think that may be the layer that this stuff gets.

SPEAKER_03 1:00:05

Maybe what you're referring to is top call log? That's one. Oh, okay. So uh but uh you know if a if a script if a script doesn't fit into the the category of the top 25 scripts, it won't get logged, but it got executed.

SPEAKER_04 1:00:21

Yeah, so yeah, yeah. You're talking about a much more deep low-level log. Yeah, yeah, yeah. Which which would by its nature affect the uh performance.

SPEAKER_03 1:00:28

Right, but that's the point. But that's okay. But that's the point. Like if in development or staging you turn this on and you go through the system and you do some testing, then you can uh double check things to make sure that you know things are are working correctly. Awesome.

SPEAKER_04 1:00:44

Yeah, Enzo, it's always great to chat with you, sir.

SPEAKER_03 1:00:47

Nice to chat with you, man. Nice to see you here in Vienna.

SPEAKER_04 1:00:52

In a little bit of a turnabout. I'm here with Javier Dura. And you are gonna be interviewing me rather than me interviewing you.

SPEAKER_02 1:01:00

Yes, yes, I am. Well, we are here with uh Mandabar for FileMakerMagazine.es. So uh first of all, Mandabar for people who don't know you, tell us who you are and what you do.

SPEAKER_04 1:01:14

So I am a file maker developer for gosh 30, 40 years now. I've been podcasting uh for 20 years in the file maker community. I've uh started and run companies and sold them and then moved to Greece uh four years ago with the thought of slowing down and that didn't work out, so I'm full tilt in FileMaker still. Um yeah, so developer and mainly really training right now.

SPEAKER_02 1:01:38

Yeah, you have been in the platform for a long time, as you mentioned. What gets you hooked? I mean, what do you think that uh get excited about the the platform, about FileMaker?

SPEAKER_04 1:01:49

I'm actually more excited about what I do than I remember ever being, maybe I don't know, a really, really long time anyway. Um and it's definitely about AI and AI integration. And I have the one area that I really focus on is um is getting AI integrated into FileMaker databases. So everyone is using AI to like write code and to test code and to plan and things like that. And that's obviously great, but I'm really focusing on how can we make the applications um better for our clients. You know, so for internal developers, what can they do internally to have better quality? And for partner companies, what can we do for our customers to solve problems in really easy ways, you know, a page of code, a few hours of work. That's my that's my jam.

SPEAKER_02 1:02:35

You also are well known for your roundtables and your debates and sessions. We are now in Vienna, in Vienna calling. So what do you like this type of format type of way less prepped?

SPEAKER_04 1:02:49

So that's one one part. But I think also because of doing podcasting for so long, I really enjoy interviewing and kind of asking questions and kind of getting the information out. Um yeah, you know, I there's so many people with so much more expertise than me, one of which is you, and and getting a chance to ask questions about it, um uh to take kind of the spotlight away from me for sure. And also I learned so much that way. Starting the whole podcast with Chris Upper Light with AI was so I can pester him and ask him questions about it and learn. And that worked out quite well. Yeah. You also run uh free AA classes Zoom.

SPEAKER_02 1:03:27

I attend those classes.

SPEAKER_04 1:03:28

Pretty much every week. I do on on Thursdays. Uh this is great with kids running by. I do um I do a free class um on a given AI topic in FileMaker integration. And I've rebranded, so I was Navarre training. I still really do training, but now I'm Navarre.ai, so you can kind of find out about those classes there. And uh so yeah, that's been really, really fun to do.

SPEAKER_02 1:03:52

And are you giving these classes for free? That's a lot of work. So what's your final purpose?

SPEAKER_04 1:03:58

Um so they are for free, but the um part of it is just to get back to the community and also people make appointments with me um to do dev. So I'm training uh individual people, you know, on an hourly rate basis, uh, to kind of start with AI. So a lot of people just really don't know where to start there. And to get from zero to starting using it is not obvious. You can't really use AI to kind of get to that point. So I'm really trying to get people from from here, you know, from nothing to the point where they can start using it, seeing the benefit, trying an integration, and letting that light bulb kind of start going on. And get them to the place where, like, okay, now I understand what this is, now I can actually use AI to teach me AI. You know.

SPEAKER_02 1:04:45

I mean, we can consider you more or less an AI expert. So uh where do you think most FileMaker developers are using AI in the wrong way?

SPEAKER_04 1:04:57

Um I think I don't know if I would say that they're using it in the wrong way. I think they're just not using it. Um and that that's the biggest problem, I think. Uh they're not using it to solve little daily problems. Uh I think maybe the temptation of developers to use it the wrong way is to try to do things that are really big. Um that are not like, yeah, from scratch, like, oh, let's gonna rewrite our whole organization and we're gonna try to take you know these people out of the mix and to try to thoroughly automate the system so we can push a button and a whole entire project comes into existence. And it seems like that could work, but it it takes a huge amount of work to do that, and there's a lot of stuff you have to do to get it. I don't know if I've seen anyone try to do that, but I've talked to a lot of people who've that's what their plan is. And then a plan that's that big doesn't really ever get started. So that might be something people are doing wrong.

SPEAKER_02 1:05:50

And what kind of brilliance do you think that AI make no sense at all? Um, that's also a really good question.

SPEAKER_04 1:05:58

So I think there's a Lot of databases that the industry has been using FileMaker for 20 years. The company's been using it for 20 years. And the problem is solved. Like the FileMaker database works the way it needs to work, and they don't really need any other things in it. And you know, the like for any kind of a any kind of a thing where they're tracking physical work, right? So it's tracking people who go into the field and install things, um, or repair things, or inspect machines, or uh run machinery, a lot of internet of things type things. I think that there's a lot of situations like that where if it's just working, it doesn't need you know this modernization. Yeah. Or little things only, but not a big rewrite. And there's other organizations, other types of companies that really can benefit from large changes. Anyway, so that's that's my thought on that.

SPEAKER_02 1:06:50

AI moves fast and Feimaker moves slow from my point of view. Yeah. Um we only have one new version every year. How can you think that we can solve this situation and to be uh to follow the AI rhythm?

SPEAKER_04 1:07:08

Yeah, I like that. So uh one of the things they came out with, uh a new script step that allows you to do a direct AI API call. Um that script step, when it came out, um the most current model at the time from OpenAI was like uh GPT-4. And and it you only use a certain endpoint and it didn't give you a lot of visibility. So when I realized those limitations, I basically switched back to insert from URL, which does everything. So insert from URL and the web viewer open up such a crazy amount of possibility for AI that I don't think it, you know, anything that FileMaker does, as long as I don't break those things, it almost doesn't matter because the the world that's open to you from those two capabilities give you almost everything you need. I mean I'd wish for more, but really whether they come out with new features or not doesn't matter.

SPEAKER_02 1:08:06

What would be your advice for someone at the FileMaker community that has never used AI at all?

SPEAKER_04 1:08:12

Go to my website and make an appointment with me. Um yeah, and I do free 30-minute appointments. So if someone just wants to talk, they want me to copy and paste code from the stuff I show in the demo, definitely do that. It's great. I I love the community aspect. I live uh in a rural area in Greece, in Greece on the island of the Crete, and I love, love, love the community and connecting with people. Um all of my career has been connecting to people one-on-one or one-on-small group. Um so that's you know that what that's the specific way I can help people. But I think if they just don't know, I think they have to try something to create something, maybe with, maybe without FileMaker. But the creative capability of using something like Cloud Code is so fun and so rewarding, and it just really unlocks the creative joy.

SPEAKER_02 1:09:02

We are now in Vienna in the Vienna Calling event 2026. There's been a lot of discussion about AI, because uh especially this is an conference. Uh, what are your thoughts of this kind of event and a conference when where everything is a little kind of chaos?

SPEAKER_04 1:09:20

Yeah, yeah, yeah. No, no planned sessions. Well, planned sessions in advance, but they can change on a daily basis. I'm really used to this format because many years ago, 10 plus years ago, I think we did pause on error conferences. Um, and those are big in Seattle and LA. LA we did one there, New York was one. Anyway, so I did uh several of them in Portland, and I really like that format a lot. Um and I think I was talking to Philip about it actually today. Uh but if there's a conversation, his point, if there's if there's a topic that three people are interested in, then it should be a session because there's gonna be a lot more than three people who would find it useful. Um and I think that's I think that's definitely the case for a lot of these questions. Um you know, letting the community decide what we want to talk about and share and ask questions about.

SPEAKER_02 1:10:09

So let's wrap this, let's wrap up this interview. Um what's coming next from Nabar? What are you planning to do? Where are your next what are you working on?

SPEAKER_04 1:10:19

Gosh, the thing that has me most excited lately, um, the two things, the two things that I'm preparing, uh are use of the claw things like OpenClaw and Hermes, things like that. Um, which I sort of see as an ultra robot. Like we've had FileMaker robots in the past, a long time ago, right? Where you have like a machine that's logged into FileMaker that does stuff. Well, OpenClaw is basically a machine, a dedicated Mac or whatever uh uh VPS instance that's running all the time, that is yours to control, that is private, and what I'm working on is having that machine um uh having access to a file maker database via MCP or V the Data API or whatever. And so its work is actually communicating with FileMaker. Its memory is the FileMaker database, yeah, and then different FileMaker databases can assign tasks to that machine using FileMaker, and then that machine can use its ridiculously powerful local sandbox and all of the other hundreds of things that it can do and do all that work and then send the results back to FileMaker. So I've been playing with that a lot, and I think there's uh I think there's untold capability down that road. I'm still kind of even visualizing the everyday things that that can do that are gonna be super useful to people. I think it's kind of early for Open Claw. I think you know uh it's rough. Like it's not something you should set up unless you know what you're doing.

SPEAKER_05 1:11:45

Yeah.

SPEAKER_04 1:11:46

Um for today. But Google or Apple could be in the position of saying, oh, we're gonna turn this switch and now we're gonna give this capability to you. Uh, and then it's gonna open up and become a whole huge thing. Uh yeah. So anyway, that's what I'm working on next. Sounds interesting. Yeah, it's fun.

SPEAKER_02 1:12:04

What are you working on? Well, uh I'm working in some integrations using AI plus make to make some kind of interaction with real customers via social networks like Instagram, um any any application from Meta Instagram, Facebook or WhatsApp. Okay. So my customers will be able, not my customer, the AI will be able to answer direct messages from all these platforms. Nice. But uh I'm working on it right now. Yeah, so having fun? Yeah, it's really fun. I mean, testing these new platforms and all this stuff is uh it's really interesting for us that we have been doing more or less the same thing with Feedback for a long time.

SPEAKER_04 1:12:55

I'm in a ridiculously lucky position that I have a lot of time to just experiment and try and test and and like uh distill is really the main thing I'm trying to do.

SPEAKER_02 1:13:04

Yeah, and then share. People like you and me that um like to invest in new things and to talk to people, we need to invest our time trying new things, not working every day in the same task in the same boring boring things. And ruling out all the ones that fail, which there are many. Okay, so thank you very much for your time, Matt, and uh we will meet soon in one of your Zoom calls.

SPEAKER_04 1:13:32

That's it for this episode. Thanks for tuning in.