FileMaker DevCast: Everything Claris FileMaker
Jul 28, 2026 · 44 min · 12 segments
In this episode of DevCast, our devs share their firsthand experiences using AI tools like Claude and Gemini for FileMaker development. They spotlight where they excel, where they fall short, and what…
John NewhoffHostZandenPanelistEvanPanelistFileMaker Developer PanelistPanelist
It can be pretty challenging to, especially historically, to look at the FileMaker Server logs and, and, and s- pinpoint the smoking gun of, of what, why a server is having a problems, whether it's performance problems or, or other kinds of issues.
I, I, I find things that, um, have been going on on the server for probably months or years and nobody even knew about because it wasn't leading to a performance problem.
Um, but, uh, in terms of AI limitations, one of the things I've been running into, um, I think ties not, not so much into the AI's context window and its inability to deal with the large, the really large size of the f- of the log files, but more about, I think, limitations in, in this case, I've been using Claude and Gemini, but, uh, limitations in what Claude knows about the log files i- in- internally, um, versus what I'm having to teach it.
Um, and you know, an example of that is, um, it, it really misinterpret- interprets the way FileMaker, um, in the... especially like in the top calls log where it, um, top calls are recorded in, um, sections in 30, I think it's 30 second windows.
Um, and so if you have a, if you have an operation that's, uh, a two-minute long, three-minute long operation or a, a situation where, um, you know, the server's bogged down and, and you've got a long elapsed wait time or, or a wait time per call and, and so the operation extends for a lengthy period of time, um, the FileMaker Server logs that as multiple rows in the log file.
And, um, Claude completely misinterpreted that, um, and, and messed up some a- a- average calculations I was having it do about how, how long, uh, how long a delay people were waiting in some cases.
And because it was misinterpreting that, that, the way FileMaker is doing the logging, the results it gave were completely wrong.
Um, and as I've done log analysis with Claude, um, it's fairly common [laughs] that I'll be, I'll be questioning it.
I'll tell it, "That doesn't sound quite right based on my understanding of what FileMaker Server's doing.
Um, take a sec- second look at that." And, and, and you know, Claude comes right back and does its normal thing and says, "Oh, you know, that's insightful," or whatever-
[laughs]
[laughs] Mr.
Um, and, uh, and, and it completely, completely reverses itself and comes to a [laughs] completely different conclusion.
It can be pretty challenging to, especially historically, to look at the FileMaker Server logs and, and, and s- pinpoint the smoking gun of, of what, why a server is having a problems, whether it's performance problems or, or other kinds of issues.
I, I, I find things that, um, have been going on on the server for probably months or years and nobody even knew about because it wasn't leading to a performance problem.
Um, but, uh, in terms of AI limitations, one of the things I've been running into, um, I think ties not, not so much into the AI's context window and its inability to deal with the large, the really large size of the f- of the log files, but more about, I think, limitations in, in this case, I've been using Claude and Gemini, but, uh, limitations in what Claude knows about the log files i- in- internally, um, versus what I'm having to teach it.
Um, and you know, an example of that is, um, it, it really misinterpret- interprets the way FileMaker, um, in the... especially like in the top calls log where it, um, top calls are recorded in, um, sections in 30, I think it's 30 second windows.
Um, and so if you have a, if you have an operation that's, uh, a two-minute long, three-minute long operation or a, a situation where, um, you know, the server's bogged down and, and you've got a long elapsed wait time or, or a wait time per call and, and so the operation extends for a lengthy period of time, um, the FileMaker Server logs that as multiple rows in the log file.
And, um, Claude completely misinterpreted that, um, and, and messed up some a- a- average calculations I was having it do about how, how long, uh, how long a delay people were waiting in some cases.
And because it was misinterpreting that, that, the way FileMaker is doing the logging, the results it gave were completely wrong.
Um, and as I've done log analysis with Claude, um, it's fairly common [laughs] that I'll be, I'll be questioning it.
I'll tell it, "That doesn't sound quite right based on my understanding of what FileMaker Server's doing.
Um, take a sec- second look at that." And, and, and you know, Claude comes right back and does its normal thing and says, "Oh, you know, that's insightful," or whatever-
[laughs]
[laughs] Mr.
Um, and, uh, and, and it completely, completely reverses itself and comes to a [laughs] completely different conclusion.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
