5/13/2025 – BuiltOnAir Live Podcast Full Show – S22-E06
Duration: 0 minutes
Be Sure to Subscribe to the podcast!
To get all the latest videos and demonstrations from the BuiltOnAir Podcast, subscribe and get notified on our Youtube channel here and our newsletter/community here.
FULL EPISODE VIDEO
Watch the full video of the show. See below for segment details.
The BuiltOnAir Podcast is Sponsored by On2Air – Integrations and App extensions to run your business operations in Airtable.
In This Episode
Welcome to the BuiltOnAir Podcast, the live show. The BuiltOnAir Podcast is a live weekly show highlighting everything happening in the Airtable world.
Check us out at BuiltOnAir.com. Join our community, join our Slack Channel, and meet your fellow Airtable fans.
Alli Alosa – Hi there! I’m Alli 🙂 I’m a fine artist turned “techie” with a passion for organization and automation. I’m also proud to be a Community Leader in the Airtable forum, and a co-host of the BuiltOnAir podcast. My favorite part about being an Airtable consultant and developer is that I get to talk with people from all sorts of industries, and each project is an opportunity to learn how a business works.
Kamille Parks – I am an Airtable Community Forums Leader and the developer behind the custom Airtable app “Scheduler”, one of the winning projects in the Airtable Custom Blocks Contest now widely available on the Marketplace. I focus on building simple scripts, automations, and custom apps for Airtable that streamline data entry and everyday workflows.
Dan Fellars – I am the Founder of Openside, On2Air, and BuiltOnAir. I love automation and software. When not coding the next feature of On2Air, I love spending time with my wife and kids and golfing.
Show Segments
Round The Bases – 00:01:40 –
Meet the Creators – 00:01:41 –
Meet Scott Bergman.
Previously a designer at Apple, I left to build a visual language to describe mental states with more objectivity. In the process I created and sold a semi-ironic tarot deck called Millennial Tarot (designed in Airtable, partially with AI). By automating that business, Tablemate, an Airtable extension company was born.
An App a Day – 00:01:42 –
Watch as we install, explore, and showcase the Tablemate App from the Airtable Marketplace. The app is described as “The Airtable extension to get higher-quality AI responses at scale by affording more context, generating multi-field outputs at once, and getting actionable prompt suggestions tailored to leverage your data—all while paying per row instead of per token like Airtable AI.”.
A Case for Interface – 00:01:43 –
Explore Interfaces with “Interface Legend & Conditional Visibility”.
Alli will share 2 interface hacks – a fun way to display a legend of sorts and another around conditional visibility.
Full Segment Details
Segment: Round The Bases
Start Time: 00:01:40
Roundup of what’s happening in the Airtable communities – Airtable, BuiltOnAir, Reddit, Facebook, YouTube, and Twitter.
Segment: Meet the Creators
Start Time: 00:01:41
Scott Bergman –
Meet Scott Bergman.
Previously a designer at Apple, I left to build a visual language to describe mental states with more objectivity. In the process I created and sold a semi-ironic tarot deck called Millennial Tarot (designed in Airtable, partially with AI). By automating that business, Tablemate, an Airtable extension company was born.
Segment: An App a Day
Start Time: 00:01:42
Airtable App Showcase – Tablemate – The Airtable extension to get higher-quality AI responses at scale by affording more context, generating multi-field outputs at once, and getting actionable prompt suggestions tailored to leverage your data—all while paying per row instead of per token like Airtable AI.
Watch as we install, explore, and showcase the Tablemate App from the Airtable Marketplace. The app is described as “The Airtable extension to get higher-quality AI responses at scale by affording more context, generating multi-field outputs at once, and getting actionable prompt suggestions tailored to leverage your data—all while paying per row instead of per token like Airtable AI.”.
Segment: A Case for Interface
Start Time: 00:01:43
Interface Legend & Conditional Visibility
Explore Interfaces with “Interface Legend & Conditional Visibility”.
Alli will share 2 interface hacks – a fun way to display a legend of sorts and another around conditional visibility.
Full Transcription
The full transcription for the show can be found here:
Intro: [00:00:00] Welcome to the Built On Air Podcast, the variety show for all things Airtable. In each episode, we cover four different segments. It's always fresh and different, and lots of fun. While you get the insider info on all things Airtable, our hosts and guests are some of the most senior experts in the Airtable community. Join us live each week on our YouTube channel every Tuesday at 11:00 AM Eastern, and join our active [email protected]. Before we begin, a word from our sponsor onto air backups. Onto air backups. Provides automated Airtable backups to your cloud storage for secure and reliable data protection, prevent data loss, and set up a secure Airtable backup system with onto air [email protected]. As one customer Sarah said, having automated Airtable backups has freed up hours of my time every other week. And the fear of losing anything longtime [00:01:00] customer. David States OnAir backups might be the most critical piece of the puzzle to guard against unforeseeable disaster. It's easy to set up and it just works. Join Sarah, David, and hundreds more Airtable users like you to protect your Airtable data with OnAir backups. Sign up today with promo code built on air for a 10% discount. Check them [email protected]. And now let's check out today's episode and see what we built on air. Dan Fellars: Welcome back to the Built On Air Podcast. We are in episode six of season 22. Good to be with you myself, Dan Fellers, Camille Parks, and Elliot Elosa With us this week and special guest, Scott Bergman. Welcome, Scott. Thank you. Good to have you on. We're gonna hear more from Scott later in the show. I'll walk us through what we're gonna be talking [00:02:00] about today. As always, we'll start with our round the bases, get you up to date on everything going on in the world of Airtable, all the latest news and updates and discussions in the communities. Then talk about how you can start backing up your data using ware backups. And then after that we'll learn more about Scott, his background and story and everything he's got going on. Then he's gonna walk us through a product that he built, tablemates and text ai, I believe is what he's gonna share with us and how that works, that extension, and then a shout out to join our community. And then finally, we're gonna end with some tricks and tips in the interface layer with Ali. So with that around the bases, few big updates. First spotted in the built on air community. Actually I'm gonna move this one. Let's jump here. Okay. Dark Mode is [00:03:00] finally here. So Dark Mode came out. We have seen hints of it in the previous weeks. It is now available and I've been using it. Camille and I were talking before the show about it. I know Scott Durelli, if you're using it yet. Thoughts on Dark Mode. Alli Alosa: I get more excited about Dark Mode on my phone, like for, I don't know, literally for, for nighttime, but like I. I on my computer, I, I don't usually like, get super excited about it, but yeah, Alli Alosa: I can see why people are. Yep. Kamille Parks: I personally used dark mode on my on my computers and I was trying out the dark mode for Airtable and some parts look really good and some parts to me still need a lot of adjustment. Particularly in interfaces, color just doesn't seem to translate well. Like [00:04:00] if you have if you color your base to any one of their colors, the sidebar in interfaces is supposed to maintain that color and they look prac so dark that they're practically imperceivable from black. So there's some balancing. I think there needs to be done some legibility for high contrast areas and stuff like that. Dan Fellars: Yeah. Yeah, like Martin mentions here, the, the icons a little bit difficult. Yeah, kind of similar experience. I've been using it. I don't know if I'll keep using it on my computer. It is difficult sometimes, like bug or errors in automations still show up is why you gotta highlight the, the text to actually be able to read it. So there's definitely some room for improvement. It is in beta, so they, they made that clear. It's, it's beta, public beta, so understandable. There's still some, some work to be ironed out, but they also made the community [00:05:00] dark as well. So now the community is in dark mode. In honor of that, I don't know if it'll stay or switch. I don't know if the community, if you can toggle that or is that for everyone no matter what. Kamille Parks: It might be following your system setting. But Dan, if you could scroll all the way to the bottom of the page. This, the footer that this, and then the footer. This is what I bean, it's like some of these should have been plainly evident when they were ready to you know, make dark mode publicly available beta or otherwise. This feels like a glaring omission. This is just the community page, but the Airtable, the, you know, application has several areas that are similarly, I think undercooked. Dan Fellars: Yeah, [00:06:00] yeah. So, yeah, I think people giving kind of similar feedback there. Yeah. Still areas. Okay, let's see. Next one. So yeah, let us know your thoughts on, on dark mode. Okay. Another big announcement, custom interfaces. They announced this sneak peek of a feature coming soon. Mm-hmm. So custom interfaces, so the ability to, to move or write extensions inside of your interfaces. What I thought was interesting is it, it it's full screen. So it's, it's basically an essentially a, an interface page where the entire page is, is your custom extension. Although it looks like the filter in the sort is outside of the extension, I don't think that's the extension. [00:07:00] Kamille Parks: It looks like it's taking up the same area that a full page list or a full page calendar, et cetera, would take up. Mm-hmm. And that makes sense to me because ideally you would have, you, you would be able to use native parts of air tables functionality to provide filtering, so you don't have to bake that into every extension you make. And that just gets passed down as like a data source property or something to the extension you add. I would be curious to see if you would be able to make a, not just custom full pages, but custom elements that could go on a detail page or on a dashboard or something like that. I think that'd be very useful. Especially the example they're showing is, is a a map and maps and dashboards go hand in hand. Dan Fellars: Yeah. Yeah, I agree. I hope that that is coming. Yeah. And kind of some context [00:08:00] awareness of, you know, you're on a single record view. Yeah, it would be interesting how that would play out. So I'm definitely interested. I've signed up for the, for the beta. I have not gotten access to it yet, so excited to try it out. Kamille Parks: Same. There's some investigation, I think someone in our Slack community had done where Airtable, GitHub, they have a couple examples for interface based extensions already on there. Mm-hmm. And the application that they're using is the same. Blocks API that is currently in place today for regular extensions that go within the, the data view. So that's pretty promising from my perspective because it seems like there might not be as heavy as a lift as maybe I was fearing for. I have several extensions already. Do I have to recode them from scratch for them to be compatible with [00:09:00] interfaces? I'm sure there's some stuff that I would need to edit, but because it's the same SDK, it's probably not. I don't, I probably don't have to change everything. Dan Fellars: Yeah, yeah. I agree. Yeah, so definitely excited to, to see this one evolve. This was always top request I've had for them for a long time. Also being a extension that makes me think I'm like, maybe I need to bring back some of my old extensions. That that would play nice side an interface. Let's see. All right, next one. Lemme make sure didn't miss any. Okay, here's one. If you are asking, so a question about chatbots, if how to build a chatbot for your website where the data that the chat bot uses comes from Airtable. So there's some responses here. Scott actually did a demo of this on our podcast using [00:10:00] chatbot.com. Ben also says it's fairly easy to set up in Zapier. You can set that up. Eight NI think you can also set that up as well. So there are ways to do that if you're looking to, to create your own chatbot on that front. There's one more here. Okay. Air, cable and customiz. Okay. Hyper DB is now, is now available to anyone in the enterprise level. Jen saw this who's actually notice that in the API documentation, it now has the documentation for how to interact with the Hyper db via the API. So you can now, at first it was only read, only. Now it looks like you can update I think delete was there as well, but, so now you can update, so you can insert new records into your hyper [00:11:00] DB and whatnot. So I believe they also announced that oh yeah. And it looks like they're now asking for it on all their OAuth. For the permissions to interact. So even if you don't have enterprise, IT asked for it. So this led to some confusions. Scott Rose mentioned some of his clients are like, wait, do I have access to Hyper db? But they're not on enterprise, so that's kind of confusing. Mm-hmm. Kamille Parks: Yeah. Yeah. One of the pieces of feedback I gave for Hyper DB prior to its general launch was if you're an individual user, you could, and you had access to the hyper DB beta, you could specify an API token as being for a particular base or set of bases for everything in your enterprise org unit or a specific, or several specific hyper dbs. But if you were editing a service account, service accounts didn't have a [00:12:00] separate hyper DB option, and that kind of spun into, how Airtable, I guess, defines resources for what a token or I guess OAuth as well can have access to. And in some places, hyper DB was lumped into everything else, and sometimes it was specifically called out as something separate. And maybe Scott's perspective, it might've been more clear to users if hybrid E was kept separate because it is technically a, a separate like objects type it, it's similar to a a table within a base, but it's also vastly different in some regards. So I can see them making dec a decision either way. But it is interesting that originally it was a little messy of how Hyper DB was categorized with regard to granting external applications, the ability to edit them in the first place. Dan Fellars: Yeah. Yeah. And with Hyper db, [00:13:00] we're gonna have to do a deep dive on this now that it's, it's out there. Let's see. So as far as creating or updating, is there a limit on the number so you can do multiple records at once? Kamille Parks: Yes. So from what I've been able to test prior, I don't know if this has changed. There was no hard limit for the amount of records that you could modify or add at a given time. It was more based on the payload size. So I was regularly doing tests for a hyper DB table that had about 20 fields on it, 300 records at a time. If that gives anyone any kind of baseline. I think it really depends on. The amount of data you're sending per record, rather than the number of records itself, which is the advantage of storing certain information in a hyper db because with the regular web API, you're limited at 10 records maximum per call, which is terrible. Yeah, Alli Alosa: [00:14:00] not very many. Kamille Parks: No. Dan Fellars: Yeah, not at all. Yeah, and it looks like as far as like field types, it only supports strings and numbers. There's Kamille Parks: dates are supported as well. I. Unless I'm misremembering. I feel like there was one other field type that is available. But yeah, there's no arrays. There's definitely no array type. You can have single selects as a the way that strings are interpreted. So you can make a field a single select type, and it works similar to how you're familiar with them, but there's no multi-select, there's no linked record. There's yeah. Dan Fellars: Yeah. Interesting. Okay. Yeah, definitely something we'll keep an eye on. Love to hear feedback if people are using Hyper db, what your thoughts are with it. Yeah, Alli Alosa: I've yet to see anybody using it. I'd be really Dan Fellars: interested. Yeah. Yep, yep. [00:15:00] Alli Alosa: Absolutely. Dan Fellars: Yeah. All right. Next one. This I thought was interesting. So CEO of Airtable, Howie Lou is starting his own podcast. He's Dan Fellars: starting a podcast called Builders and Breakthroughs, a new podcast series hosted by Airtable as co-founder and CEO. So he just had his first episode with a chief product officer at Cisco. And so, yeah, I think this will be interesting if you want to see where Airtable, CEO Howie is thinking and talking to people. This might be a podcast worth checking out out. Did you give it a listen? Curious how it was? I no, I just listened to just this ten second clip. Got it. So I haven't listened to the whole one now. Yeah, Alli Alosa: I probably will. [00:16:00] Dan Fellars: Yeah, Alli Alosa: definitely. Dan Fellars: Yeah. Thanks. Yeah. And this ten second or 22nd clip doesn't have, how are we talking? It's just a response from G two. Okay. One more. This is from the Reddit, Airtable subreddit. How do you clean up your base? I thought this was a good question. If you get, if you've been using Airtable for any amount of time, you know, it can get a ton of unused fields and linked record fields that get disconnected but still remain as orphaned. All sorts of stuff. Formula fields you no longer need or don't think you need. So the process of cleaning up your base, that's something that everybody has to deal with. So there's, there's some good, good approaches here. You know, start with a fresh view, hide all the fields, then just add 'em back slowly to evaluate 'em, any of that look. So that's kind of a manual process, [00:17:00] I think. Was this I think maybe Scott you might have responded to this. I did, yeah. Yeah, it's, I noticed your logo there. So, using ai, you wanna talk about what you do there? Scott Bergman: Yeah, so less so for cleaning up the fields, though, it can be used for that too. So often the data you have is just like messy and a formula won't be the thing to clean it. It could be that all of the data is formatted, you know, with different casing or whatever it is. And so a lot of times I just rely on AI to be the thing, to clean it up. But also if you have several fields that are like similar data, but not exactly the same, and you wanna just create like a master version and get rid of all those other fields, you can use AI to synthesize the information and combine it in some format you want. And have that become your master field and then just delete the other ones, basically. So that's just a couple of the ways you can use AI to do some of that cleanup. Not all of it, but some of it. Yeah. Dan Fellars: Okay. Yeah, and it looks like we might get a demo of a tool that could [00:18:00] help in that use case. I know Ali Camille, what are your tips and tricks on grooming? I have so, Alli Alosa: so, so many little things that I've done over the years. I could go, I could literally dedicate an entire episode just to talking about this, but I, one of the, my favorite things I've done recently is I have, you know, a metadata table of all of my fields and I use a script to update that. I have my own script, but you could also co Kon I believe still has one in the marketplace to be able to do this to. Create a table of your fields using a script. But then I took another script and like one table at a time like that I was focusing on, I had it look at all the fields on the, the table. So my fields are now records, and I had to iterate over each field record and find out how many records had a value in that field on that table. So then I could populate and show how many of those fields were actually empty, like in a [00:19:00] table. Which made it really easy to be like, oh, none of these fields have values. So I can look at those first and decide if I can delete them, but that's just one little trick. Scott Bergman: Yep, yep. Yeah. One other like incredibly simple thing that took me too long to discover was that if you go into the tools part next to extensions, you can click on manage fields and actually just seal the fields there and and delete them, and edit them. I've also done it where I've taken a screenshot of those, of that list. Put into chat and said, can you give this back to me as text one per line, copy that and then paste it into, as you said, a table where I can then have all my fields listed there. And sometimes if I'm doing automations with something like NAN or just programming, I can use those fields as sort of a reference for it to know what's going on rather than, I guess, clearing all the tables and figuring out what the fields are. It can break, but it's a little hacky way to kinda get all the fields. Alli Alosa: Yeah, absolutely. I find that [00:20:00] super, super useful. 'cause then you can like document too and be like, these things are used in this automation or Right. Whatever. There's a million things you can do from that point. It's really fun. Yeah. Kamille Parks: One thing that one of my clients actually ended up doing on their own was they had this legacy base that had been in place for forever, and it was set up with a schema that they thought they needed, but they realized later down the line that they needed something different. So they had a lot of extra fields. They had not enough tables actually, and several other things. And it was made at the time before interfaces existed. So what they ended up doing was they created an interface that had all of the fields that they know that they need, and then they came back to our team and said. If it's not on this interface, we don't need it. It's not necessarily saying that the base doesn't need it. Like there might be helper fields that [00:21:00] might help calculate certain stuff that they see, but they did the due diligence of saying, well, I know how I do my job and in my day to day, these are the handful of fields that my role is responsible for. So that was very helpful for our team in saying, okay, if it's, if we look at, if we click delete on all of these fields, if one of the dependencies is not that interface, then I know that my client doesn't need it. And, you know, I don't recommend just bulk deleting without checking yourself. But that is a, I think a decent approach is to design an interface. It doesn't have to be pretty, but. You create a grid view that's like, here's all the stuff that I know I need, and then use that as like a, a negative dependency of if it has this, leave it. And if it doesn't have it, then maybe it's okay to get rid of. Alli Alosa: Yeah. I actually really love that. 'cause it kind of makes it super obvious no matter who's the one trying to delete the field. It's like, [00:22:00] even if you made the interface name, like, don't delete this field. Critical. Yeah. Yeah. I like that. Oh, we lost Dan. Kamille Parks: Well, he was the one sharing his screen, and he's the only one who can allow us to share screen, so Alli Alosa: hopefully he gets back. But yeah, I could talk about cleaning up fields forever. I, I did another thing recently where I, I wrote a script to find any fields that ended in the word copy or had a number. As the end. So that way, 'cause those are the ones that, you know, if people don't know what they're doing and they're adding extra links and then you end up with like links to other tables and they all end with 2, 3, 4, whatever. Yeah. And then, so yeah, I just wrote a script to be able to quickly find where all of those fields were and that was really, really helpful. Scott Bergman: Yeah. And are you, are you both good at when you make like a helper field that is more [00:23:00] specific? Moments in time where you're doing a specific task to get rid of it. I often, because I'm thinking like, I'm probably gonna need this in the future, and then I never end up needing it again. Yeah. Kamille Parks: What I try to do is preface it with the the hourglass emoji to just say to myself, this is temporary. And if I come back to a base a month later and an hourglass thing is still there, I know I can get rid of it. And similar to Allie, I had a, a script that I use, I might actually end up publishing it. It'll go through and ask you. Which table has a bunch of linked records that you want to rename linked record fields. And it'll allow you to select a naming convention for those fields. So name it after whatever its companion field is named, give it just the same name or name it after the table that it's linking to, and a couple other options. So you can click one button and have them all named. Right, because [00:24:00] like Allie, I had so many bases that just had like a bunch of contact records linked to a project, but on the context table, it's projects one, projects two, projects three, and that's not useful. You don't know what each field is for. Alli Alosa: Exactly. That was, that's, that's really helpful to be able to do that, especially before the dependency checker trying to figure out, it's like, well, which one is this? I don't actually know which field this links to. It's so frustrating. I, one final thing I will say I did too is I have a script and I think I've shared it on the podcast before. I'm happy to share it again. It's super, super simple. It literally just copy and paste it and click run and it shows you all the fields that are broken in your base. So all the fields that have the little triangle next to it, like there's something happened that it's no longer working. 'Cause that's something you can access via scripting. So it's super easy. It just shows you, boom, here's [00:25:00] everything that's not working. Yeah. Great. Dan Fellars: Sorry about that. My internet likes to crash only during this podcast hour. Yes. All right, well, let's move on. Okay. This is general AI news. Big acquisition. OpenAI bought Windsurf for $3 billion. I mentioned this in Slack. Windsurf has only been around for six months, less than six months, something like that. They went live, I think, in November. And so that's quite the acquisition after six months in existence. Pretty crazy. Mm-hmm. Quite so Windsurf is basically, if, if you're in the, if you're in the AI development realm, it's, it's an editor similar to Cursor, competitive to Cursor that came out recently to help you build with ai. Scott Bergman: That is wild is that's a huge deal. Yeah, [00:26:00] because I use Cursor heavily and the open ai, like I go through the Reddits and very few people use any of the open ai APIs for that. People tend to use Claude Gemini. And it could be that this is why they haven't provided that much support or like really paid attention to making it good for Cursor is because they had this in mind. So it's, it's a very interesting thing. Maybe, maybe. Dan Fellars: Okay. Last one or two more upcoming events. It's always good to see what's going on. I didn't realize Boston in two days or in a week. There's an event an in-person one, so if you're in that area, Alli Alosa: I'll, that's me. Dan Fellars: Yeah. I'm going to that Alli Alosa: page right now. Dan Fellars: Yeah. Yeah. This one. Is there then if you're going to any of these, look like this one's going on right now. UX DX, this one I'd love to go to in Sydney. There's one in Germany and then Paris, Melbourne and [00:27:00] then LA It looks like those are the upcoming live events if you wanna go out there. I have a comment here going back to the Windsurf. It's fairly new, but the company has been building for a while. Kudos to them. That's true. Yeah, they were, they have been around before that. Okay, last one, A fun one to end on. This guy is the Marco Angelo of Microsoft Excel. So I was watching this video. Comes from great big story, which Patel pretty cool short snippets. He basically draws using like the line figure in Excel, so you can draw lines in Excel. And he does all of this in in Excel and then prints it off. He's, he's Japanese. He is in Japan. And that does some pretty cool stuff. I don't know if I can show this. You kinda see what he's done all in Excel. Alli Alosa: That's so cool. I I Kamille Parks: I wonder how you start [00:28:00] with making this your medium. Dan Fellars: Yeah. He talks about it. He, he said he was too cheap to buy paint and he's like, so I'm gonna use, I've got a computer, I'll just use this. And he didn't even wanna buy like Scott Bergman: Adobe or something else. That's so funny. Paint by numbers to a a new level. Yeah. Yeah. I, Alli Alosa: this is actually like, I relate to this like way more than one would think. It's kind of funny. I. I really have a, I hate Microsoft Word with a fiery passion and I refuse to use it. And for the longest time, part of my job was creating these really intricate PDFs and I had to make these like really like, you know, tiny little check boxes and all these different places and I was like, I'm not touching word for this. So I took Microsoft Excel and I just created a template for myself where I made the whole sheet was just little quarter inch squares. Like I made all the [00:29:00] cells be a quarter inch by a quarter inch, and then I just merged them where I need them merged and like make these really intricate Excel pages like, and that's exactly kind of what this reminds me of. Like I could imagine coloring in each one of those little squares. And I Dan Fellars: love that. That's awesome, man. He's put some pretty, you got, you still gotta be even maybe more artistic to figure this out. Kamille Parks: Mm-hmm. I thought this was gonna be pointillism, like he made all of the you know, rows and columns, a, a similar size, very small, and then coloring each individual cell to form a larger picture. But this is, I, the line tool in Excel is just simply not one that I reach for personally. Mm-hmm. Dan Fellars: Yeah. Very cool stuff. Let's see if you could do that in Airtable, that would be very impressive. I'm not aware of [00:30:00] Now with the custom extensions in interfaces, Alli Alosa: there's something W Van Hall did, I believe. Way back in the day, he like made this crazy heat map in page designer and I still have no idea how he did it, but wild. Dan Fellars: That's true. If anybody could do it, maybe that's where he's gone off and why we haven't heard from him. He is in a basement somewhere. He'll come out with his artwork. Kamille Parks: I, we need one episode where it's just looking back at W's old posts and trying to do them in modern Airtable. They were already difficult to do in 2018 or so. Dan Fellars: Yeah. I Kamille Parks: just wonder what you could do today. Dan Fellars: Cool. That wrap up wraps up around the bases. If you are running your business in Airtable, it's best practice to make sure your data is stored outside of Airtable. That's where [00:31:00] ware backups comes in. It stores your data, your schema, your attachments, all outside of Airtable, into box, Dropbox, Google Drive, or OneDrive, and also has restore functionality. If you need to restore your data back into Airtable, check it out at onto air, use promo code built on air for a discount and start backing up your data today. With that, let's learn more about Scott and your story. Camille's gonna ask some questions, put you cool spotlights, Kamille Parks: and well, welcome Scott. Thanks for joining us. Scott Bergman: Thanks for having me. Kamille Parks: So, classic question, how long have you been using Airtable? Scott Bergman: It has been probably about six years of heavy use, maybe seven years of ex of experimentation. Kamille Parks: Do you remember the first use case you had where you were, whatever you were using before, just wasn't cutting it? [00:32:00] Scott Bergman: Yeah, I do, and it, it'll be a bit of a story, but I used to work as a designer for Apple for about six years. Mm-hmm. And my task I was working for a team trying to figure out what's the future of collaboration and collaboration software especially for enterprises and for Apple specifically. Like, can we make something internally? Would we ever consider making it something that's external also? And so I had to research tons and tons of tools and see what was out there. And the biggest problem that they were kind of plagued with inside the company, and honestly probably still are, but is that there are so many different teams using so many different tools and they're all siloed from each other. And they all have data stored in a slightly different format, and they don't talk to each other. And of course you can use tools like Zap here to try to connect them all, but when you're talking about the entire company and all the different tools and all the different instances of it, there wasn't a great way to do that. And to me, Airtable doesn't solve all those issues, but it was sort of the [00:33:00] gold standard of having any, almost any type of information to be stored in a very simple to use format. But whether you were tracking tasks, you were tracking you know, like KPIs contacts, CRM, like any type of information you're tracking, it was, it could work for that. And to me that was a very powerful thing and the fact that you could connect it between things. So I found it very I found it to be awesome and I started using it for my own personal life. There are instances of it being used at the company but they were looking to create. Kind of what we came up with was creating a solution that wasn't the source of truth for any of the information. You still might use Asana or Airtable or whatever else, but it could bring in all the information so there'd be one source of truth for each project that had all the information you needed. It was a very cool project that that I haven't really seen come out any companies make sense, which was like five or six years ago. So it's possible they have, but Airtable was sort of at the time, the best [00:34:00] alternative I could find to something like that. So that's how I got started. Kamille Parks: That's interesting 'cause I, I often find myself, trying to think back to when I first started using Airtable, which is around the same time, I think 2018 for me. And at the time, obviously a useful platform. 'cause I ended up using it and here I am a thousand years later still talking about it. But there were a lot of features that I'm used to today that aren't, that weren't there at the time. So there was no blocks or extensions or apps or whatever they're called now. There was no interfaces. There was automations, but there were pretty simple, there was no syncing, et cetera. But to your point, it still provided an easy place to go to, to put in data that could be in, not any structure, but a lot of different structures. So importing stuff from other applications. To have that data in one place as straightforward [00:35:00] as it could have been at the time. Scott Bergman: Yeah, and I think something else to that point of like, not any structure but many, is that I do think a lot of times we use like a note of some sort, a shared note for something that could be cells in rows. And I, I think that there's just a lot of value in being able to even just like single select, multi-select to be able to turn it into like a discreet unit because so much gets written down and then lost because it isn't usable, because it is this it's like this amorphous shape, it's a paragraph. And being able to turn things into different fields and different columns that are discrete units make them like usable make 'em usable for formulas. And I guess something that is interesting, we'll get into the AI stuff is that the paragraphs start to become more usable because it can interpret it into side. Categorize it into something that then is, can be used in a if then statement, you know, how to react to this. But I just think there's such power in like discreet concrete [00:36:00] objects which Airtable kind of forces you to, to do. Mm-hmm. Kamille Parks: My sort of day-to-day is building applications for other people and I find myself saying internally you're gonna get a strongly typed system if it kills me. Yeah. And I, it's one of the, the benefits of using a system like Airtable in that you can be pretty strict about what. The form in which you're collecting data. So you can take a status field in a number of ways. But once you make a decision that field is one type and all of the other data in that whole column or field, it's gonna look very similarly. So it does kind of force you to make some decisions which can be changed down the line. You can change your base, but it is helpful to make a decision about your data so it's not so freeform that no one understands it. [00:37:00] Scott Bergman: Yeah, totally. Yeah. Kamille Parks: So after you started getting comfortable and using Airtable in your own sort of day-to-day life, what was your sort of next big thing that you put Airtable for use to? Scott Bergman: Yeah, so I after being there for like six years, I knew I wanted to leave. I had this idea of something that I wanted to create. My long-term goal is a way to visualize thoughts and emotions in a more concrete way. Like almost the way we're forcing types here, like to do that with emotions and thoughts and create this model. It's a long thing. It's like my long-term project, but in order to work on that, I wanted a way to build, to make some money as I started to work on that. And I just had this funny idea. I used to live in San Francisco and had this idea for a millennial themed tarot card deck. I wasn't into tarot at the time, but I just, this is probably too much information, but I went on a date with someone who was like religiously into SoulCycle, and I was just like, this is such [00:38:00] a millennial version of like, religion just, you know, like spirituality. And I just was like, it's like a tarot card. And so I started making them and I started selling them. And it, it was going decently. And so I started finding. Retail stores that could potentially sell it. And I wanted to reach out to them. And that's when I started using Airtable heavily. So I use Airtable as like the content management system to create all 78 cards of the and I can, I dunno if I'm sharing information sharing screen, but I can show a little bit of what that looks like. Am am I sharing? Should I pull it up? I can, Dan Fellars: I can make it right. Yeah, right now it's the, it's good. Kamille Parks: And if you haven't heard of Chris Dance, you'd love him. Scott Bergman: I, yeah, we we messaged a little bit and I'm actually getting married in two weeks, so I've been very busy. Thank you. So I really should follow up with them more. But anyway, so I have, you know, all 78 [00:39:00] cards listed here. I have the original card artwork. I have my new card artwork, and then I have all the texts that ended up being in the guidebook. And so I used this as basically the content management system, and it's kind of I think it was my, my very first project and it's been the sorts of truth, like all the way through. Until this past year I actually got a publishing deal for it and also to make a Gen Z version, which you can see a sneak peek here. And I had to hire a Gen Z writer to work with me because I don't know all the jokes. So we use the ai, you can see there's 238 hidden fields. There's a lot of people like giving feedback who got access to it temporarily like voting on which cards they thought, you know, were their favorite. And making. And then I used my ai, which I can get into to make different versions of like jokes and coming up with the iconography for the different. Images based on the original descriptions and basically being able to use AI in mass, in bulk all at once. To do things like generate jokes, to generate hashtags that are [00:40:00] relevant. And that's where you can see all of these are, these were all AI generated. So ultimately I ended up writing the whole guidebook, but I used a lot of the stuff that the AI created to, to do that. So I don't remember what the original question was, but this is kind of what my foray was into Airtable for my personal use. And then I also started using it to do outreach and to track like influencers I wanted to talk to and retail stores. So it's like a CRM. And during this time I am extremely lazy in some ways and hardworking in others. And I was frustrated having to copy and paste back and forth to chat GBT and I was like, there's just gotta be. An easier way, and that's where I built this extension to let me do that. I can, Dan Fellars: yeah, let me, lemme do a transition real quick on that. So yeah, so you're gonna talk about product you built. So the company is Table mate, [00:41:00] but the product is, is text ai, is that the name of it? AI text, yeah. Okay. All right. Yeah, why don't you walk us through that. Lemme put it back on the screen. Okay, cool. Your screen is showing. Scott Bergman: Yep. So what did I do with the table? I just had open? So I'll go to a, a more universal example, but at the time, Airtable didn't have AI out yet. And yeah, so as I was just saying, it was a pain to have to copy and paste back and forth to chat GBT and like I had 78 cards that I needed to create these writings for. And then I had like 2000 retail stores I wanted to create custom messages for. And I just was not gonna do that by hand. So I built this extension and I've since improved it. But essentially you have a text, sorry, you have a prompt that you can write. And then you have the fields. You can either type a field in here that you want to add, [00:42:00] which would be you know, shows up in Ping so you know that it's a field. And you can also just type it like this and then you can preview it and you can see, oh, I see I have this as one of the outputs, so it's not letting me have it as an input. Anyway you can type in prompts here. You can also append additional fields that you wanna send over to the ai. The AI will process whatever it's that you're asking for and take in all the information from that row that you want, and then it will output it to the columns that you want that are generally empty. And it started out as just one output. You know, you could just do one at a time. Now you can do multiple fields, all for one one prompt. And Airtable ended up coming out with a similar thing about three or four months later. I don't think they [00:43:00] copied me or anything. I think it was like an inevitable idea. But what I noticed when they came out with it, and if you guys have tried it, you've probably seen, is that it is extremely expensive, in my opinion. So they charge per token or per credits. And I think to get a good response using ai, like the, one of the top things I hear is that people get kind of unuseful responses, especially if you're trying to create something like social media content that's so subjective. It isn't just like a quick data transformation. You need to give like really rich long prompts and they charge based on the length of the prompt. So to get like a rich prompt is gonna cost you like 30 credits or so, and it ends up being that you can only use like 30 to a hundred rows and you're out of credits. So I, I just want people to like, I'm very passionate about people like creating their own businesses and just [00:44:00] being able to do more work with fewer people and less time. And so this, you pay. For tokens directly with open ai which means that you can, you get charged pennies on it. And so you can have really long prompts and get much better responses. So you can see like this prompt I have in here is extremely long and it was a lot of back and forth with chat GBT to create it. And it's defining the format I want and you know, like different parts to embody and then I can give spec more specific instructions for each field in here. And, and then I can, you can run it. So in this example I have this list of ideas for posts for marketing for marketing my millennial tarot deck. And so if I wanna create like a year's worth of social media posts, I might have 365 rows in here. I just have like a brief description. [00:45:00] I created all these in chat, GBT and I said, just gimme one per line. I copy and pasted it in here. And then I wanted to start actually creating like an Instagram video script an Instagram caption, what the first comment could be on that post, A post for x, a email subject line from my newsletter and the email body. And I want all of those to relate to each other. You know, it should feel like one cohesive post, not separate prompts for every single one. And so I was able to make each of those an output. Most of 'em were pretty self-explanatory, so it knew kind of what it needs to write. Plus I had the description here, and then I put some extra description for like the email body. And then you can just, I like to run a test first, which is just on three rows. So it'll look for the first empty rows essentially. And it'll take a second if you do a ton at once. Like it takes a couple seconds first, but then they all just [00:46:00] start popping in. So let's pray that in this live demo. It works. It almost, I like there we Dan Fellars: go. One thing that's cool about this is I like yeah, being able to update multiple records all off the same prompt. 'Cause that's not how Airtable AI works. So that, that's really cool feature. Yeah. Kamille Parks: Does Airtable AI let you update multiple fields with the same prompt, or do they each have their own prompt? Dan Fellars: Each has their own, Kamille Parks: yeah. Yeah. So it's a lot of consolidation of, unless Dan Fellars: you did through an automation. Kamille Parks: Yeah, Dan Fellars: sure. You could, you could output, you could get multiple As like A-J-S-O-N or something and then update. Alli Alosa: Okay. That's what I'm doing. We have to be really careful to. Sic so that nothing's ever empty. Scott Bergman: Yeah. Mm-hmm. Yeah. So I'll say, I mean, there are benefits to like, in terms of using this versus using Air table ai, [00:47:00] air table ai, AI has all of those like additional features that aren't just this, which are like recommending which record might be relevant or using a token when you generate a formula, even though people always complain that they're not great I still find those to be pretty handy and us, and I find that I don't need to pay for an AI subscription to have enough credits to do those tasks. I find that also like I don't have it built such that you can that it will just automatically run this you do manually press the button, but you can at the moment choose a formula field or a checkbox field. And so it will see everything that passes a certain condition based on other columns. It will decide to run on just those rows. So you could say that once text has been filled in then you run it on that row when you press run. So that's another area where AI text I think is beneficial and I'm sure that they will [00:48:00] catch up on some of the stuff I have here, like multiple outputs. But the cost, then it, it comes largely down to cost. So I have found another good thing is it's really useful if you are iterating on something and you want to just like keep trying your prompt and changing it quickly. You can just select a row and then press, use the selected method and I can just run it on that one row or do a couple of rows, which is really handy. Which, I like to do a couple of tests, see where it's wrong, like where it's producing results. I don't want, tweak my prompt and then I'll, once I'm getting consistently good responses, I'll press all and I'll run it on the entire thing. And yeah, and like in, you know, a minute, let's, this is a dangerous move, but let's see what happens. You can get, you know, just a ton of content. So this is for more like [00:49:00] subjective kind of content. Yeah. Like creating a social media post. But I'll also use it for things in addition to like iterating on something. I use NAN heavily in my workflow. And a lot of times I'll just forget to have a. Put something some part of the data that I need in my N eight N flow. And so to retroactively kind of fix that, I'll use this tool also. So like I work with this lawyer to collect a bunch of information for like tax foreclosures and cleaning ladies making vacuum noises in the background. Hopefully you don't hear that. But I forgot to include like the title. So I just wrote a brief prompt that can you know, like look at all the, look at the first name of each person and just figure out what the gender is so that when I write letters to them, it can be the right format. And that's like a perfect task for ai that a formula, no formula that I'm aware of can do. So I find if it's Kamille Parks: an [00:50:00] extraordinarily long formula with every possible first name. Yeah, Scott Bergman: yeah, that's true. Yeah. So that, that's your other option. You could have extraordinarily long formula. Yeah. So that's just like another little place, places. Let's also check on this. Yeah. So this looks like it's still going, but it's filled in a bunch. Very cool. Places that I've seen people use this, like customers that I've spoken with, some there's like a few financial groups and one of them loads the text from financial reports in, I think I probably hit a rate limit, but they'll put in financial reports and then if they have like multiple KPIs that they wanna pull out from that report and they wanna do it for like. A hundred different companies, they can load one into each you know, into a cell or each company, and then pull out all of those different like, key metrics that were reported in that [00:51:00] you know, financial report into the different fields. Another big use case is people who are building out, like content platforms like you saw I, I'm doing that for my tarot deck. But there's a a couple startups that use it and one has this platform where they have like a massive database of what do you call it, like doctor's offices and medical supply information. And they have an online platform where you Alli Alosa: can Scott Bergman: view all that. And so they're taking like the bits of information they have and generating like longer paragraphs to fill out all of the information that's. Then powering your, their platform basically. And they're doing like 10,000, they've done something like 10,000 records. So far people have processed, I think it's like 800 million tokens or something through it, which is pretty cool. Yeah. Yeah. So that was a lot. I just rambled on about yeah. Happy to answer questions and stuff. Dan Fellars: Cool. So yeah, tell us so they can find it in [00:52:00] the marketplace. AI text and very cool. This is awesome. Yeah. Scott Bergman: And you can go to table make.io. Also just connect your Airtable token and a chat, BT token. Then you're ready, Dan Fellars: set to go. Scott Bergman: Awesome. So Dan Fellars: cool. Very cool. Thank you Scott, for sharing that. Yeah. Okay. We got five minutes, Ali, we got enough time? Alli Alosa: Yeah, yeah, these are, these are really quick. Dan Fellars: Okay. Let's do it Alli Alosa: Scott. Yeah, thanks. Kamille Parks: Yeah, there we go. Just gonna say we can hear the vacuum. Dan Fellars: Okay. So Ali's gonna share before that, if you haven't joined our community, join us built on air.com/join. Get you in with thousands of other Airtable enthusiasts and users talking every day about Airtable. So join us, built on air.com/join will get you in for free. And let's learn some tricks of the trade from Allie in interfaces. Alli Alosa: Awesome. [00:53:00] All right. So these are just two little hacks that I sometimes use that I think are fun. One is around legends in. Interfaces. I am not a huge fan of the built-in Airtable legend on the, the charts. I think the dots are too small. I wanna be able to really see what those colors mean. Sometimes the colors are way too similar. There's just a lot of things that would change. And it also takes up a lot of real estate of the actual chart. So like, I just, I, I usually would prefer to keep the legend hidden, but sometimes it's nice to be able to see it. So as a little hack that I've done a few times, and it's actually gotten a lot easier now because of Airts new feature around translating or formatting rather what the output of a formula is to look like a single select. So that's what I'm doing [00:54:00] for this hack is I'm using, I. These formula fields. So let me just show you what it looks like here. So it's literally just three formulas and each one just has the value I want displayed in quotes. And then on the formatting tab, I'm using that same value. And very simply just setting it up so that it literally always says up. Same thing for down and the na, and I'm just matching those colors to what my actual formula is that I have this chart grouped by, which is I go do, where is it Group by? Under y it's grouped by a field called up slash down, which is showing me if I added the emoji for fun. But. This is literally just gonna either say [00:55:00] Na, if my percent change between this year and the year before is zero, then otherwise up if it's up, down, if it's down, so above or below zero. And those colors are defined here. So essentially I'm just kind of taking each one of those options and creating a formula field for it so that every single record on this table has those three values in a field in three separate fields. Then you can display those on your page. And this will only work on like a detail page or a record review as unfortunately, you can't display the big, pretty single select things on a dashboard, but if you change the appearance to large and then just right click and say hide labels for entire row. That's how I'm doing these three fields here. So it kind of just turns it into your own little custom legend. And just keep in mind, you know, if you change the colors on what you're [00:56:00] grouping your chart by, then you'd wanna co come back and update your colors to match that as well. And these three fields. But I find it to be a pretty nice little hack because now you've got more space in your chart. It's not taken up by those the, the real estate's not taken up by the legend. And I just find these to be much prettier and I attention grabbing. Kamille Parks: That's a cool hack. I had to build a legend for a list view that was colored a certain way. Each of the rows were colored based on certain conditions, and I had to use like emojis to say if it's green, it means. Whatever. And it was very annoying to have to update because it was like a text element on the page. So if you changed it, you had to go back and edit the interface. This is much nicer. Alli Alosa: Yeah. I find this to be fun. It's like, it's easy and simple to set up. I used to have to do all these crazy things before the single select could [00:57:00] be formatted like this, or the formula could be formatted like this. Yeah, that's a fun little quick one you can do for sure. And then I know we're like at time, but the other thing really quickly is I, I have the same data here displayed as quarters and then again as months. And I don't love how much real estate this takes up. And sometimes people get confused. I've noticed as I'm building many dashboards, if I have these charts next to each other, they'll be like, what am I looking at? What's the difference between these? And it's like, it's same data. It's just quarter. Versus month. So I'd love to be able to like, give the user the control to switch between those two visualizations. And with conditional visibility, which is available on the business plan and up you can do that in a simulated way here. So Camille's done a demo before on like an edit button here. So you turn, you click the button to edit, and then it turns certain [00:58:00] fields editable and hides others. And this is very similar to that. So if I click quarters, it's gonna flip this. So now I'm looking at the quarters and if I click months, it's gonna flip it back to the months. You could, you know, expand on this in a million ways, but this is a really fun little quick thing you could try. And essentially all I'm doing is I'm setting this view option field, which is a single select with months and quarters. And that's what those buttons are doing. Here is the months button, is an update record setting that view option two months. And finally this group for the months is conditionally displayed when that view option is months. Same thing. Conversely for the quarters and altogether, you've got this pretty little switching visualization. Dan Fellars: That's cool. Yeah, very nice tricks of the trade. [00:59:00] Alli Alosa: Two quick hacks. Dan Fellars: That's awesome. Awesome. Thank you for sharing that. And Scott, thank you for joining us. Really appreciate it. Excited to see what else you've got going on and congrats on the wedding as well coming up. Thank you. All right, and that concludes this week's show. We will see you all next week with our next episode. Have a good week. Thank. Intro: Thank you for joining today's episode. We hope you enjoyed it. Be sure to check out our sponsor onto our backups, automated backups for Air table. We'll see you next time on the Built On Air Podcast.