$20M founder reacts to mind-blowing AI workflows | Vibe marketing challenge ft. Gumloop

0:00Max, uh, how's everything going at Gum Loop? I know I I've seen you guys are shipping a lot of, uh, a lot of new, uh, features lately. Tell us, uh, tell us what you're up to. Yeah. Um, I think we're really leaning into the, uh, the year of the agent and, uh, trying to make everything kind of effortless. Automate with natural language, iterate like you would with any other kind of co-pilot product. We have a bunch of other updates coming out in the next like couple weeks, which we're super excited about, but same sort of direction, like effortless automation

0:27across your stack. Awesome. Awesome. Yeah. Uh definitely feels like uh agents becoming like real now. You know, there's been a lot of talk about agents and everything over the past like, you know, six to nine months, but I think now you're really starting to see uh really good use cases and stuff like that and uh able to actually complete things for you that you could just like go ahead and, you know, push into like production, so to

0:52speak, which is uh which is really cool. Yeah. I've always I've been saying like you have to uh earn the right to build an agent. you have to kind of build the foundation of the things it can do before you build the agent. I I feel like it's kind of all coming together and agents are making sense now which is

1:04super exciting. Absolutely. Absolutely. Cool. So, uh we did a big challenge in the community uh and basically we we came to the community and uh you know we love Gum Loop and a lot of folks like in my audience and and within the Vibe marketers use it and uh we wanted to you know challenge everyone to see what kind of like Vibe marketing workflows uh they could build with Gum Loop. So, you know, we kind of partnered up on this and uh you know, launched this challenge and we've got the three uh winners here today. So, would love to be able to talk to them, sort of demo their workflows

1:40and uh get your reactions, Max, in terms of how they're using the product and maybe if you have any other ideas for how they could even, you know, take these to the next level. Cool. I'm excited. Awesome. So, uh, I'm going to bring each, uh, person onto the stage. And, you know, like I've I've watched all of these workflows. I was pretty blown away, uh, with what James and and Roas and Zid were able to create here. Um, so super excited to show you what we've got. And, uh, James, I'll go ahead and add you to the stage. All right. And, uh, would love to just sort of get a

2:14brief uh, background on, you know, what got you into sort of building uh, with Gum Loop? um you know what were you able to create and uh what was your experience like? Yeah, definitely not a traditional background. So I have operated a food truck and catering company for four and a half years. Always was interested in you know different aspects of the back end of the business. So, I got into like looking at AI automation back in April, May time frame and just playing around

2:41with different things that I can build. And I was looking more at like the practicality angle of small business owners like they don't have time, you know, to do different things like this. One thing that was a common pain point for many was email. And that's just instead of it's not even sending automated emails. It's you're going in, you're getting the same question back to back to back, but you're reinventing the wheel every single time and you know, starting from scratch, whereas I wanted to approach that where you just had emails drafted. So, you're 90% of the way there and you can just click send

3:14and adjust a few things. Awesome. Awesome. So, you just got into building uh workflows literally like a few months ago and now you're winning challenges like this. Yeah. Yeah. Yeah. So I was definitely always uh more systemminded where before for I guess like 8 8 to 12 months before that I was looking at like visual canvases and just like mapping business processes but all that is is map. It it is a helpful I think because then you can see your process but it doesn't do anything. Whereas once I saw this I'm like okay now this is you map and then it gets what you need done at the same

3:52time. So perfect like pairing. And I'm just curious uh throughout your your build uh and maybe you know QAing and and things like that and getting it ready. Did you use AI in the process to kind of like ask questions or you know sort of get through issues? Would love to kind of hear about uh about that

4:13aspect. Yeah, absolutely. Uh I'm definitely in Claude a lot and I uh I use Gummy and Claude. actually uploaded a bunch of docs from gum loop into claude to like as a project to go back and forth and ask that and then yeah anything time I'm getting stumped adding like I don't code so anything like that I go in ask questions there and yeah usually problems get solved after enough iteration but it's definitely planning it out first like and then going through it I find sometimes when I just start like building it helps me learn but then it's like I got to redo the entire

4:49entire thing at some point. Yeah. Yeah. And are you using this workflow like in your day-to-day uh role or or business or anything like that or are you planning to? I'm going to. So, my current business, I just started it. It's in this space. Um I don't have enough context yet as far as my email to use it for my other business which I trained it off of which is catering. You know, we have requests coming in like when are you going to be at this catering job? Are you available this date? So I trained off that. So I had plenty of context on like these are

5:22the responses that you need and um so I trained off that but again not using that anymore but was for a little bit. Yeah. Awesome. Awesome. Well uh let's take a look at the workflow. What do you what do you all say? Yeah. All right. Cool. So I'm just going to uh share my screen. I'm going to add it up here on the stage and I'm going to play the video uh and u we'll get a clear

5:45idea of what you were able to do here.

6:03This is going to be three for one. It's going to give you categorization for your emails, response templates based off of past send emails and your past inbox. And then there's also going to be a automation that automatically updates that knowledge base. And then finally the response generator. So it's going to reference those documents and it's going to create on a time trigger drafts. The reason that it's three separate automations is because they're ran on different time triggers. I went through and started and I have a note here 50 you can go 50 to 100 um emails that are good context like and then it'll just go

6:44through past send emails as well. It's going through my brand DNA document where I already have my voice and everything. So it's referencing that along with the send emails. So it has even more context to draw from this long prompt. You can check that out. That is to generate six categories that your email will go through and then response templates for each category and then

7:08it's going to write it to notion. There'll be six categories, 10 sample emails per category. You can go through edit anything you want. If you have templates that you already have, you can go through put them in here. This responder node, it starts with a unread email and this will go through and just filter out if there's zero responses. If there are responses, it's going to send it through. So, what this is going to do is going to go through your emails and it's going to reference your knowledge base and then say, okay, is this in the knowledge base? If it's not in the knowledge base, then the AI is going to

7:48take it through and it's going to write it to the page. So, that'll run on a periodic trigger, whatever you want to update that knowledge base. So when you're responding to emails, you're getting more accurate when you have these, you know, different send emails going out.

8:14My first thought when I see something like that is um how many features we're we're shipping to make that easier for you. Uh the the biggest difference between like a workflow and an agent I've been describing is like a workflow is kind of like a play in football and an agent is the quarterback. Um when I see workflows like that, you're kind of building a quarterback into the workflow, but I think we should be shipping the features that because that was super impressive. That's like an awesome awesome workflow. Um, we should be shipping the feature that gives you the quarterback by by default and then just lets you define the plays because

8:43those plays like the looking back at my template emails, uh, updating it, updating the context every like couple of weeks or whatever when when that like third time trigger was going off, those are like perfect plays, but we should make it even easier to build the quarterback into your product. That was awesome though. Really cool.

8:56Yeah. Thank you. Awesome stuff. You know, it it just amazes me and and opens up uh a whole new perspective, I think, when you see that you just started doing this like a few months ago, but you have, you know, vertical expertise, like you're you're in this business, you know, the problem that you face. Uh you know, kind of like uh what's going to save you time and and optimize. And then the barrier to going and and doing this is just getting lower and lower like by the day, which is which is super cool. Um, I also think that, you know, if you go online or

9:30whatever, you see all these like, um, you know, crazy workflows that you would probably never like implement, but this is like this solves a real problem and a real painoint. And, uh, I think that a lot of other businesses could uh could benefit from this. I think customer service is something that makes people pull their hair out and it's a perfect use case for an AI agent or an AI workflow. So nice job on like you know shipping it and uh and getting it done and uh understanding the problem deeply and uh awesome work James. This is

10:03really cool. Yeah. Yeah. Thank you. Appreciate it. Yeah. I actually look back on that one though and uh I'm like I guess that was probably like a month ago and it just feels like that's like so long ago where I would I've changed things from then and built more practical stuff. But yeah, again what what you just said, the practical angle is definitely what I go to because, you know, I kind of you show that to like a small business owner that's not in the space and they're kind of just like I don't care about that. I want the input and the output. So the simplest things that I've built like one

10:34was just like upload a resume, get a candidate report back. Um so you can upload like you know 25 resumes. Um that was like one thing that connected right away because the simplicity in it and then the ease of I put a front end on that in replet where it just sent via web hook or to AWS and then into gum loop and it runs the flow. So stuff like that is just like you know easy for me to build a backend logic in gum loop and then just the simple inputs on a front

11:01end and replet. I mean you sound like you're coding now. I mean without the coding yes I am natural language. Yeah but uh where one month feels like uh you know you can progress so so quickly. Exactly. Awesome. Exactly. Yeah. Awesome. Cool. Anything else uh Max for for James? I'm impressed. I'm excited for you to to try out the new features. Uh there's little things when I was looking at the workflow where I'm like we're going to make your life so much easier in like the very like in the next two weeks basically like extracting away a lot of that logic. Like I saw those filter

11:36nodes that you had that are kind of like big blocks to make sure that you're not responding to emails with no threads. Like that'll be just handled natively by the uh the quarterback. And it's funny you say that. Oh, sorry. But it's funny you say that because when I'm doing that and like stuff that feels cumbersome, I'm always thinking this is

11:53going to be like changed and like soon. We know the exact thing. That's all we're working on. Yeah. Yeah. Okay. Right. But yeah, so far been great. easy to use and yeah, everybody's been very helpful in the gum loop like ecosystem with all that. So, sweet. Bugs are fixed fast and there's not many

12:11of them. So, yeah. Awesome to hear. Amazing. Cool. Thanks, James. What you build next? Yeah. Yeah. Thank you. All right. Talk to you soon. See you. All right. Uh, next we have Roous. What's up? Good to see you again. Good to see you as well, James. Max. Hey, nice to meet you. So, uh I see I see you a lot in the community sharing all kinds of uh awesome insights on uh you know kind of vibe marketing and AIdriven search and and all of that. Uh so I'm I'm super excited to a kind of learn about your background. We haven't had the chance to kind of connect yet on a call. So would

12:53love to hear about uh what you're up to and uh what your workflow is all about and how you got into uh into using Gum Loop. Wonderful. So yeah, James, a pleasure to meet you once again. Um my background is actually uh I was building with AI, you know, for the past like two three years. U when all of this like trend, you know, you could say started back in end of 2022, you know, when CHB team was launched. I I was also like working in like marketing. I was doing PR uh related stuff in in an agency and yeah when CH GBT launched I quickly changed

13:29my my my skills and the knowledge that I wanted to to be in in in that current landscape. So I was try at first I was like doing a automation agency stuff. So basically trying to automate like uh small and medium businesses inside my like location. So I'm based in Europe, so Eastern Europe, my home country and and more and countries around. Uh so yeah, that was like the first step. The second step for me was like building some sort of AI education company as well. So we are one of the biggest like consultants in um in in in in the region as well by basically consulting large

14:09enterprise businesses, you know, how to apply AI and into their workflows. So I'm yeah I working in AI a lot and by working and and seeing a lot of great people you know I joined the community and that's why I also want to share the knowledge that I got you know building by you know with all the community as

14:29well. Cool. So you have a bit of a background in building workflows and and using AI and and all of that. It sounds like our our last guest was pretty new to uh this just a few months ago and uh you're kind of uh you know in the expert category. I guess we could we could say well yeah I got like a great experience you know building these workflows. I also have like my own team that we build all of these automations you know custom automations for clients as well. So yeah I have you know some some some sort of knowledge but uh once again I'm not a

15:04coder or programmer you know I don't have this background but I learned a lot. Yeah. Okay cool. So uh give us the highlevel you know uh 15 second overview of of your workflow like uh what does it do? Why did you build this workflow and uh what was your experience like uh going through the process of of doing it with

15:26Gumloop? Um yes so at the moment I'm also like running my own SAS you know it's called a Eclipse it's basically for AO optimization and the channel that I'm really interested in you know while like to do marketing in is LinkedIn. So I wanted to like look at some possibilities you know how can I optimize and automate LinkedIn. So this automation mostly like solve this problem. So I have like my own keywords that I'm interested in and whenever I want like whenever I will see like somebody post you know SEO optimization or go like mentioning this type of keywords I want to look at that post you

16:04know scrape it and have it you know in my own like database. um uh to maybe comment on this post, maybe like pick some interesting ideas and use on my own content or just you know to do competitor research. So I build this uh automation that will daily track the keywords that I'm interested in. It will fetch all the data from LinkedIn. It will parse it and it will store in in my you know air table. So that's the

16:32system. Awesome. Awesome. I'm excited to check it out. LinkedIn is uh is one of those channels that I haven't been able to crack yet. I I just can't figure it out. So, I'm excited to see what you've got. And uh maybe I'll use your workflow to help me. Cool. So, uh yeah, let's bring it up on the stage and

16:54uh bear with me for a second.

17:10What the system basically does is that whenever somebody posts they're related to my keyword, you know, I get that post profile name post full URL of article as well as number of likes, comments, the date and of course AI generation. created content that that I could use in my own profile based on the patterns and systems that I like to write content about. Play is doing all that, you know, but the pricing is really big. You know,

17:42the the starter plan starts from $150. So, basically, we start with code note. I went to uh rapid API. I searched po I use search post endpoint and I basically just uh copy pasteed the parameters you know and using car and inserted it into gum gum loop so that's uh like code note that we are using after that and using JSON reader where basically extracts the data and using a uh I qualify that data and extract it one point by one so I extract like poster name, the content of a of a text, when it when it was posted, the exact URL, the profile of LinkedIn

18:31URL of a posted, number of likes, comments, and the date. So all of this data that you are seeing right here, we need to extract it from LinkedIn API. And after that, I ask AI to generate me a post and as well as a comment that I could insert. And all of this will go back to air tables. Even though the system looks pretty simil simple, you know, but the real value that you get from this table is insane because you could track your competitors. You could be the first reply each and every time and you could get like your content calendar and just by seeing what other

19:09people are doing. That's awesome. I was uh going to ask how you got the LinkedIn content on Gum Loop. Uh and then I s I saw you integrated the API. Have you tried the custom nodes yet? Uh without not the run code step, but custom nodes. Uh regarding LinkedIn? Uh no. So uh we have this feature where Gummy will write the code for you. And I saw you had like a a run code block and then a parse JSON. You can just prompt Gummy and it'll write the code for you and build a perfect LinkedIn scraping node if you want it. And then you can always reuse that. It goes into your

19:43node library. That's the only edit but like perfect workflow. Really simple and beautiful. Uh really cool. Yes, when I was like building this solution, I haven't found this. Maybe like it was a month or so ago. I don't know if you had it already. We do a really bad job of showing off all the cool features that exist. Like you have to stumble upon them. Sometimes we have users who like have been automating for two months and they find some like power feature they've never heard of. Uh we're working on that, but I can send you a link to the custom node docs after. It's kind of

20:11like having cloud code baked into Gum Loop. You just describe what you want. It'll agentically go build it and then test it for you and make sure it's ready for production. Wonderful. For sure. I'll give it a try, you know, after the calls. Sweet. Um I I really uh when you mentioned Clay and you said, you know what, I don't want to pay for Clay. It's like $150 a month. So, you just built your own solution and you're not a developer by trade. It just, you know, really popped out to me that like what an incredible uh time it is, you know, truly the age of like kind of total

20:44democra democratization and like personal software. Like, you know, you don't have to know how to code. You can go build your own essentially SAS that that can add a lot of value uh to your business and and to your workflow, which is just so cool. We couldn't have done that, you know, a year or two ago, which

21:01is which is amazing. Yeah. One thing I really want to do, uh, I mean, we're working on it right now, but it's the marketplace of publishing flows and publishing nodes. Um, the next iteration of the marketplace will be like letting you sell your your nodes or your workflows. So, if you find some hack that emulates like a $200 subscription out there and does the same job, you could sell that for $10 on Gum Loop and if a thousand people find it, like you're making money by sharing what you've learned. Uh, so hopefully democratizing just like building and and

21:28yeah, making money online. That's an awesome idea. So, uh, awesome, Rous. Well, um, yeah, super cool workflow. Uh, again, um, I struggle with LinkedIn. So, I'm excited to check this out. I think for like, you know, not just content, but for finding like sales opportunities, people who are like commenting on posts in your niche that you can go and follow up with or DM or whatever. Like, it just takes one message or or one post to, you know, find a really awesome lead. So, uh, good stuff and, uh, excited to see what you

21:58build next. Wonderful. So yeah, I have like one extra comment in all of this like social listening is a really great trend and happening right now based on like on clay on trigify and all of these different tools you know and with like gum loop you can make it your own you know and it's really exciting because I think that's like that's the age of sales right now where you uh do a really personalized outreach at the right time with the right offer you know it's very important because right now it's not enough just to write you know AI gibberish and send it to thousands of

22:30leads. You know, you won't get any results. Awesome. Thanks a lot, Rookas. And uh we'll be uh connecting soon in the community. Wonderful. Have a great day. All right. See you soon. All right. So, uh let's see. Our third submission uh I don't think is going to show today, but uh we're going to give him an honorable mention. Uh his name is Zid Kim. and we'll play the video and we can still give it a a little bit of a reaction here. So, I'll go ahead and uh and get

23:01that going.

23:17Every marketer faces the same expensive question. Should you buy another tool or build your own? Gumblau identify the core functionality that delivers 80% of the value, giving you a practical starting point you can customize. We visit the create page where we can analyze any marketing tool. Let's try Clay this time. I'll type it in and click create blueprint. We're redirected to the dashboard where you can watch

23:40your blueprint being built in real time. Once complete, you can view the full report here and it's automatically sent to your email as well. Clay's blueprint is ready. Eight out of 10 complexity, seven days to build with 13 nodes for data enrichment, AI powered messaging, and automated outreach. Let's look at the Gum Loop workflow that powers this. Our web hook input node receives the tool name from Gumlau's interface. This connects our front end to the analysis workflow. The Perplexity web search node researches current information about the tools marketing features. We get up-to-date capabilities, not outdated data. Our verify marketing tool custom node confirms this is a legitimate

24:22marketing tool worth analyzing quality control before we invest processing time. The ask AI node extracts core marketing capabilities that matter for replication. We focus on practical features, not marketing copy. Our generate blueprint custom node identifies the required gum loop nodes in sequential order to recreate core functionality. The JSON writer node packages verification features blueprint and analytics into structured format for delivery. Gumlau transforms tool curiosity into practical build guidance and turning marketing tool research into

24:56actionable build versus buy decisions. Wow, that was so meta. Yeah. Yeah. Okay, that is uh pretty cool. What do you think of that, Max? Uh, I would have never thought to do that in a million years, which is awesome. Um, super meta build. It's like building. I mean, one of those subflows was kind of just doing what Gummy does, like uh, our new Gummy agent that builds workflows. I I imagine they like pasted in all of the nodes and explained what they're for and had AI

25:26make sense of which ones to use when. Yeah. Um, pretty uh, preient of a of a workflow. Like they kind of knew what we were going to work on before we built it. Uh, really cool. That's cool. Yeah. I was thinking uh while watching that like that could be some kind of landing page for for Gum Loop like build verse buy enter in the tool you're considering here's how you could just build the functionality and you could have like Gummy kind of spit out like uh a workflow that they could go and like activate in their account or something like that how much they could

25:56save things of that nature. Um pretty cool. Yeah, we should just have our gummy agent just like be aware of I guess have this this tool to like research other tools and if if you mention clay or something, it could look it up and find the features. But this is really sick. They like built a feature

26:12of the product in the product itself. Yeah. Yeah. Yeah. And uh Zid is u is more of a on the engineering side as well. So he had some more uh kind of advanced you know functionality in there and and things like that and clearly uh kind of understood some some solid workflow logic in terms of you know first researching and validating that you know the features were relevant to marketers before he went in and you know went further into the workflow and stuff which I thought was really smart uh as well there's certain people who are like very AI forward like he has AI narrating the

26:49video he probably had AI record the video. Uh he's using AI to make the workflow. Yeah. Yeah. Yeah. Very cool. Very very meta as you mentioned. Uh very impressive as well. Um cool. So that's our uh demos for the day. Uh overall, Max, what's your what's your main kind of takeaway here from from all these submissions and uh and kind of the winners that uh that we

27:14showed off today? I think they're awesome. Um all I can really think about is like how much easier we're we need to make everything. uh and like the updates we're working on right now, but uh seeing everyone's workflow kind of informs the product decisions we're making. So, this is awesome context. Like, uh there's a couple things I picked up in this call that I'm just going to go back into the office and tell the team about uh just to make things simpler. Um yeah, I'm

27:38excited though. Cool. Good stuff. Um all right, so we have a little bit of extra time. Uh we've got a few listeners here, sort of some uh dedicated community members. Um, anyone want to ask some questions for for Max while he's here? Um, AMA uh wants to know kind of what what's next? I know you gave us a few sort of hints here in terms of maybe monetizing workflows and and things like that. Uh, what excites you about uh Gum Loop's uh

28:07roadmap right now? Uh, sharing workflows publicly and monetizing them is a big uh kind of push on our side. like how can we have a directory of thousands and thousands of workflows there is a a deep technical challenge there that most people don't think about u so if you if you look at like a any other marketplace where you can execute code uh if you like take nen for example you could realistically make malware on a platform like that in 10 seconds so you make like a custom node where you have to put in your Gmail credentials or whatever but secretly one of the lines of code is forwarding all

28:42your emails to an S3 bucket or or like basically siphoning all your data. So before we release the marketplace feature, we have to make sure the security is like really in place. We have like AI security scans and human review for every workload that's going live. Um so that that's something we're working on. We want to rush to making that that happen while not jeopardizing anyone's data. Like it needs to be ultra secure. Um and then uh yeah, Gum Loop agents which are uh the quarterbacks basically that we're talking about. Um, yep. I think it should abstract away like 10 nodes in every workflow and just

29:12let you expose super super useful tools in a format that's usable by like your entire company, your entire team super easily. Awesome. Awesome. And I know uh I I always like to hear how you know founders are using uh their own product and and kind of eating their own dog food so to speak. So, uh, last time we chatted, you were telling me about, uh, like a recruiting workflow that you were using to to scan like, uh, you know, public GitHub repos and things that you found interesting. Uh, I was just curious if if that workflow was successful and, uh, you know, what else you're kind of uh, building with uh,

29:49with Gum Loop that you're using like dayto-day. Yeah, I can share my screen and show you. Um, I still use that workflow all the time. Uh, how do I share? Okay, share screen. Yeah, there's a bunch of workflows. This is the open source contributor hunter that I showed off last time. Like put in the link to any open source repo and then you get this like pretty gorgeous uh uh email as a result with like the hiring details. I'm just opening up um

30:15my own inbox here. Uh you get this like really beautiful email with like everyone's links and the whole background and who you should be hiring. Um I've used this to meet like many many open source contributors that uh contribute to like the frameworks that actually like support gum loop. So we kind of have like connects to the infrastructure that that powers gum loop which is pretty cool. Now um I've still I've hired a few engineers since our last call but most have been through my network like just I asked my friends who their smartest friends are and hire them

30:46right. Um the uh we've been doing a bunch of um kind of socials uh related workflows. So we have a growth intern at the company who automates like literally everything he does with Gum Loop. Um one of the main ones was we noticed uh we we focus like almost exclusively right now on enterprises like all my time goes to like our the Shopifies and Gustos and Instacarts of the world. Um but there is so much value in showing people just on socials what Gum Loop can do. So we've been reaching out to like people who make really cool automation content and asking like can we do you want to

31:19partner? Can we give you free access to like just like spread the word about how powerful Gum Loop is? Um, our intern built this workflow that takes our Chrome extension. Um, you just go to a Tik Tok profile uh and you run this Chrome extension on it. It will take the most recent video, run it through Gemini to give a give a a really nice like, "Hey, I loved your video about Zuck's uh AI call thing that that AI demo that went bad." Um, it looks at all of the views and average number of comments across all videos to figure out like what is the engagement ratio uh and what

31:51is the average number of views we could expect if we were to like support them and help them make automation content. And then it calculates uh basically like how perfect of a fit are they for Gum Loop based on like their most recent videos. Like are they talking about 10 automation tools? Are they just talking about Gum Loop? Are they just talking about another one? Um and it all logs into a Google sheet. So, he just browses Tik Tok every morning with the Chrome extension and just gets an entire lead list of uh like the best creators in the

32:13space, which is really awesome. Amazing. That's some leverage right there. Yeah. Um, good. I have another one if you want to see. Uh, absolutely. This one, this one's really cool. Uh, I built this with a customer. Um, but it's it was my idea, so I'm like comfortable sharing. Uh, it's nothing sensitive here. There's a customer we have that sells to restaurants. Uh, like a big customer that sells to restaurants. I'm sure you could imagine um the they they help restaurants deliver food. Um y if you want to unlock a new geography, normally you have to go figure out like

32:46what are the restaurants in this area? Um are they in like the target like ICP of customers we want to onboard first for a new geography. Uh what is the hook that will get them interested in talking to us? Um what language do they even speak if we're really opening up new markets? So this workflow starts with an interface. You just put in like the name of a small town. Um interfaces look like this. It's just like this UI layer on top of your workflow. Um, it then uses perplexity and just extracts what language they probably speak. Uh, and then for every cuisine that they want to

33:14target, it runs a subflow that scrapes Google Maps for everything about the restaurant, like the number of reviews, the average sentiment of the reviews, like based on what the people have been commenting. And it also gets like the uh similar searches. So, if you're looking up sushi in San Francisco, there's probably like five other sushi spots that will show up in the search. It shows where they rank in the search. And then it fills all of that into a translated uh call transcript because in a lot of small towns, they don't have

33:39the website for their small sushi shop. They have a phone number. Um, so it takes all this data, name drops the competitors down the street, mentions where they're showing up in the rankings on Google Maps, and says like, "We would increase your exposure and sales by x% if you were to onboard with our our service." And it logs all of that in a

33:55just a database. That's brilliant. I love that. Google Maps is like an amazing data source. uh as well like you know getting review sentiment. Uh I was looking at uh it as a kind of niche research tool looking at like kind of review velocity versus the number of service providers for you know certain markets to try to try to research little like boring local business ideas and stuff like that. So yeah, treasure trove of data. Um that's really cool. I think like referencing those uh competitors and where they rank and all that probably provides a killer open and response rate compared to a

34:35normal strategy. Yeah, you you really incite the oh no, I see I like they know my competitors like I they probably think about these this other sushi shop 247 because that's like their main competition. So if you name drop it on a call, it's really powerful. Right. Right. We have a like three more minutes,

34:49right? Yeah. Okay, cool. Um I wanted to show custom nodes because this is like one of the best parts of Gum Loop. Uh so you you notice in um uh the second workflow we saw that there was a run code node and then a parse JSON. Uh now you can just say like I could paste in the docs for any API here, but I can say like call a web hook and let me pass in data and then gummy will just build the node for you. So it'll just like uh it'll figure out what are the inputs, what are the outputs, and then it'll reason through

35:17is this even possible? Like first it's making sure the idea is possible because sometimes people will like write a joke as a prompt. Um, and then it'll write the script to make this happen in the background. So, uh, once it's done writing the script, you have this node that is forever usable in your library. You can name it like my, uh, personal API scraper that hits like has a bunch of different criteria in it and a bunch of logic in the code. Um, but it's published forever. So, it's it's like a Lego block you can reuse in all of your workflows and you can share with other people. Um, so right now it's

35:43it's figuring out what are the inputs and outputs. It'll start streaming that in a second and then uh the code should be written just directly in the background. So it determined like okay there's input is data then it lets me specify the web hook URL what the HTTP method is. It got even like fancy with it. It let me specify the headers and I don't even want all that but uh it's smarter than I am so it took it's liberty to do that. Um that's really cool. I I know uh just from you know interacting with like our community and stuff like that uh you know parsing and formatting data and all

36:14that that you're getting from external tools or scrapers uh is a big hangup for a lot of people um big blocker. So um you know making that easier uh with with the agent uh seems like an awesome idea. Yeah. And the same thing actually happens uh for all of our integrations as well. So, uh, we saw the Google map node before. That one's totally custom as well. I just prompted what I what I

36:39wanted or let me Yeah. Um, Google Map. You can just describe what you want Google Maps to do, like get reviews from a Google Maps link or get only the reviews with over five stars that were made in the last six months. It'll build that custom integration for you. You don't need to like like kind of scratch your head and and figure out how to do it with a no code paradigm. You just lean into the code and have AI handle it

37:00for you. Wow. Wow. Love that. Super cool, man. Anything else you want to show us uh on uh on Gum Loop? Uh the biggest thing is probably Gummy agent like uh it will build the whole workflow for you. So like create a weekly AI report to get my mentions on Reddit. Like Gummy knows how to do this and it will build it end to end just like kind of a cursor sort of agent. Um no more NAN learning curve if that's like the thing that's blocking you from building really amazing workflows. Uh just describe what you want and and let

37:29Gummy handle it for you. That's huge. I've been waiting for uh for something like this to to be ready and and working really well. So, uh super pumped to see that uh you all are the ones to to get it out there. Yeah, we're excited. Awesome. Cool. Well, uh yeah, Max, really appreciate you uh spending the time with us. I know you're a really busy guy working with uh a lot of customers, a lot of big companies, and uh really pushing the envelope in this space. So, uh, yeah, we appreciate the the partnership on the challenge and you, uh, spending time with, uh, with some of your users here to give them

38:07feedback. What an awesome, uh, opportunity for, uh, for James and Rous to get, uh, to get Max, uh, to react to their workflows. So, thank you. Yeah, it was a ton of fun. Appreciate the involvement. Awesome. Cool. Max, uh, James Roas, thanks again. And, uh, I'm going to end the stream now. Uh, thanks everybody for joining and uh, excited to to get this

38:29out to the world. Sweet. Bye everyone.