Home All Episodes About Official Page Subscribe on YouTube
Episode 64 Aug 05, 2025 49:51 65.3K views

How to Manage Your Gmail with AI Agents (Email Draft Agents)

About This Episode

Struggling to keep up with an overflowing inbox?

In this episode of The AI Agents Podcast, hosts Demetri Panici and Aytekin Tank explore how AI-powered Gmail Agents can streamline your email workflow by drafting customized responses based on your communication style.

Learn how tools like Jotform AI Agents and Relevance AI can intelligently generate email drafts, organize messages, and even help schedule meetings—freeing up your time to focus on what matters most.

Discover real-world examples, including how small service businesses and content creators are automating email replies using AI agents trained on their previously sent emails and website content.

Whether you're handling sponsorship requests, customer support, or podcast guest scheduling, you'll learn how to build powerful automations that boost productivity while maintaining a personal touch in your email communications.
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
⏰ TIMESTAMPS:
0:00 - How AI Can Process Emails
1:12 - Meet The Hosts And Episode Goal
2:28 - Saving Hours With Gmail Automation
5:25 - Intro To AI Draft Responses
6:08 - Inside Jotform Gmail Agents Demo
8:02 - Smart Labels And Email Categorization
13:00 - Complete Agent Setup Walkthrough
20:30 - Advanced Agent Features And Calendar Integration
27:20 - Real-World Email Demo Responses
35:56 - Measuring Success And User Feedback
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Sign up for free ➡️ https://link.jotform.com/ngx1n53aZC
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Follow us on:
Twitter ➡️ https://x.com/aiagentspodcast
Instagram ➡️ https://www.instagram.com/aiagentspodcast
TikTok ➡️ https://www.tiktok.com/@aiagentspodcast
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬

Transcript

And uh today we are going to talk about how AI agents can actually help you process your emails. Not in terms of how you like categorize or filter or like label them, but more in terms of how you can actually respond them faster. And uh one of the ways you can do that is if the AI agents could actually respond to your emails like if you could actually train an AI agent that can create a response. uh how to do that using Jatform's new we just released this this week just like literally like two days ago Jatform Gmail agents hi my name is Dmitri Bonichi and I'm a content creator agency owner and AI enthusiast you're listening to the AI agents podcast brought to you by Jot Form and featuring our very own CEO and founder Idkin Tank this is the show where artificial intelligence

meets innovation productivity and the tools shaping the future of work enjoy to the show. Hey there and welcome back to another episode of the AI Agents podcast. We're back together today. The duo is back at it again. We haven't done a podcast together in a little while. Idkin's been enjoying some some time uh out and about. So, we finally are getting back together. And today we are going to be talking about how to manage your email inbox and send drafts using AI agents. How you doing, Idkin? >> Yeah, I'm doing great. Uh we were we were we were traveling with the family. Uh we were in uh in in the UK and in London. So we just had some fun and you know um with the trick kids it's actually out of work but uh I'm back at uh back at uh work and you

know uh doing episodes uh for our podcast is exciting. Yeah. >> Yeah. I'm really pumped, man. It's been a pretty good sort of, you know, obviously not as great as when we were working on it together, but no, we've had a lot of really cool guests in the last um couple months. I'd say that recently we be getting more and more cool guests. Our streams are going up every every week, and we just hit 50,000 total um downloads, uh which is pretty awesome. I really appreciate all the the listens, the likes, the reviews. It's it's been awesome. So, please keep doing that. Uh it's very helpful every time you leave a review or a comment on YouTube. Um obviously reviews on podcast platforms, that's the key. Really appreciate those. Um so thanks so much for for everything that you guys have done for us in

the last little while. And what we're going to talk about today more specifically, I know I mentioned it at the start of the episode, but Jot Form as well as, you know, some of the tools I'm using are really useful for saving you time on possibly the most annoying thing in work, which is just like replying to emails. You know, uh, a lot of times it's the same sort of thing, mannering on the topic. you know who you are, kind of what links you need to send, what decks you have to send, and um AI agents can really help you reduce your time for, you know, customer support obviously as we've seen on websites, but we still get emails and those emails need to be responded to and that's what these agents are going to be able to help you with. You know when I

when I started Jot Form in the early days, uh one of the biggest uh problems for me that I was spending like so much time processing my emails and I will just uh sometimes like spend like six hours, eight hours a day just uh responding to customers, uh processing my emails and one of the first automations I did was to uh use like kind of use Gmail labels and filters to create an automation. So that you know I can actually uh fil filter all these emails into like different uh priority levels of inboxes so that I can actually work on the highest priority emails first and it made a big difference instead of spending like six hours a day. I would just spend like an hour a day and uh and I don't I wouldn't miss like the most important emails and then you know

the less important emails they could they could I would like check them out later on. But when you don't know what's important it's uh usually when you don't know which emails are important when your inbox is full of emails it's just like uh so like so time consuming to process the email. So that was one of my automations and and uh today we are going to talk about like AI agent how AI agents can actually help you process your emails not in terms of how you like categorize or filter or like label them but more in terms of how you can actually respond them faster. And uh one of the ways you can do that is if the AI agent uh could actually respond to your emails like if you could actually train an AI agent that can create a response but you don't want

like email is kind of sensitive because you don't want like >> uh you don't want to send an automated email most of the time right you want you want it to create a draft so that you can just check out the draft check out the email and then if it looks good just click on send >> or if you want to change something add something you can edit. So, um this is something AI agents are great and uh you're going to show how to do this using relevance AI agent. I'm going to show that uh how to do that using ago we just released the Jatform AI agents uh called Jatform Gmail agents which is an AI agent that can actually be uh trained on your on your inbox and then it can actually create drafts for you automatically and I'm going to demo that

and and then you're going to actually you're going to showcase how you're actually doing something similar using relevance and you're actually you're going to actually show you the real use case uh real demo like like the real uh business your your own business demo um so I'm just going to go ahead and start with mine um so we just released this this week Jatform uh Gmail agent uh it's it connects your AI connect your AI agent to your Gmail account and instantly get draft responses for every incoming customer email so basically uh it's trained on your existing so it's so easy to use because all you have to do is like click on a single button and it does everything for you just it will just go and read your previous I think like last hundreds emails that you responded and then from that it's

going to know like which emails to respond and how to respond and all that stuff and as soon as we released this like that was a huge interest like this was one of our biggest launches because I feel like this is a big problem for everyone. Um so but before I do the demo um you know I'm going to show the page a little bit more. Uh so this is just like you know um there are some you know different uh demos here healthcare you know product uh you know education IT different um use cases. Basically, you receive this email and it previews this draft and you can just review the draft and send it. And how it works is it's very simple. You give the authentication and then it just reviews your emails and then it will start uh just writing drafts for every

email you receive. After that, you're going to it's going to start creating drafts. And yeah, this is basically it's very simple. Uh but what what's really great about this is like uh instead of spending so much time rewriting the same things again and again for every customer you you you will you will always start like ahead by just having these and then it's just very fast as soon as you receive the email it's just creates the drafts. Uh so that's also very fast and it's trained on your send emails. Um smart labels. It creates these labels. It's decide which labels to create and then because it's because it's train trained on your previous emails. Uh it knows like what kind of uh emails you receive and then it designs these labels. So it's easier for you to see like what kind of so it's not

just helping you write the rest but it's also categorizing your emails. uh depending on the request. Um >> yeah. Um so basically we and it it also has all the features of Jatform AI just basically this means like if you want to integrate it with with your Google calendar if you wanted to take appointments, make appointments, it can just have a conversation with other party and create an appointment for you uh for for the other party as well. um or if you need to take a payment and you know all the other stuff all the integrations are coming with that um and basically you can actually give feedback uh using Jot forms on conversations in back inbox as well to the to the AI agent so they can provide better um answers. So you know Jatform is like almost two decades old company with uh

30 million users. So, you know, you know, we have all these like sucks to GDPR, HIPPA compliance and all that stuff. So, >> you know that you're not just trusting your Gmail account with with a company that just came out yesterday. >> Um, and yeah, some testimonials here as well. So, yeah, we can go ahead and start with the >> Yeah, let's check it out. >> Yeah. So you if you remember like I you know I I do I I actually like doing like all these like uh small business uh use cases. So again like you know I you know went to web flow created this from the from a template. So basically let's say you have a cleaning service right this is a expert cleaning service you provide deep cleaning services and you know office cleaning services and you know this is uh let's

say this is your um you know and you can actually train your agents uh not just uh in your from your previous emails but just providing your website and then it it can train from your website and give answers by just looking at your website as well. So, uh we created this uh demo account. It's just uh it's just a demo email account and we sent these emails just like you know uh and ask some some questions that actually these questions were actually generated by CHP as well. Uh but basically uh these questions you know so so that we have some um we have some data to work with. So yeah, we have all these questions. Let's um let's go ahead and then connect with this account. Um so I'm just going to go ahead um connect with this Gmail account. So maybe you're not

seeing it because it's different window but basically I'm giving it access to to the Gmail and um so now it's basically creating an agent for me and uh basically it's now started looking at the previous emails um just reading those emails reading the answers and learning from that uh that's the first step uh you know this sometimes think can take long time but you don't have to wait here. If you have like you know uh many emails you can you can go and come back to this and it will continue but since we we don't have a lot of emails this should this shouldn't take too long and it already finished. Um and it's also like creating all these labels and categorizing depend on different kind of answers and so it's doing some more processing here. I guess what's really interesting um here is and

you know I what I'll show you later it did take a little bit more manual work but I wanted something like this about two months ago when I was setting this up that essentially would train on my inbox you know and I think this is this is pretty incredible because this is simple right it's simple you set it up it does what it does here and nobody wants I mean not saying that both don't have their merits but this is very simple to set up as my point and you know that's what a lot of people are looking for. >> Yeah, it's done. It's ready to answer your emails. So I say next and so basically it's um it's learned from the emails and it hasn't created any drafts. So if I look at my email inbox there's no new email but it created these

um labels for me by looking at these previous answers. So here I created this um this question for for this uh move out deep cleaning scheduling. I would like to know more about scheduling options available to for move out deep cleaning services. Could you please provide me with the details again? I you know we got help from CHPT and sending this and then um let's see um so first I will get I will receive this email and then that's once I receive that email that's actually going to trigger uh that's going to be um so okay yeah I have the email now so in the back end right in the back end this is my demo account um back end Uh we are now waiting for it to for a draft to be created because uh it's connected to my account. Um let's wait for

a while. Oh yeah, I already got it. I think uh you know I I while I was preparing for I connected it to multiple agents so I might have multiple drafts. Uh so yeah uh so thank you for your interest in move out clean and this was prepared for the AI agent and we offer and you can actually disable this part. Uh you don't have to have this one of the reason we we we decide to have that is because like people connect and they might forget that who is creating the draft. Uh so just to make sure that you know we have something here which is this can be removed. Um, we offer flexible schedule options and compressive cleansing kitchens. Uh, so yeah, it looks like it's ready to book an appointment. I didn't do a tool integration. Uh, I could do that. Uh,

I could go to like the train action tools if I want to integrate that with my appointments, right? Um, so let's find the appointments. Um, I could integrate this with with uh with my Google calendar so that um it knows the avail times when I'm available to do an appointment. Uh let's take a look at that. Um Google, let's see. Um yeah, set appointment with Google calendar is here. So basically, I can set that up here. I'm not going to do that, but I'm just going to show how to do that. uh basically but um you can work on the persona how you want it to answer questions. you can build your knowledge base if you want to if you wanted to like learn from your website basically and it created all these knowledge bases uh basically uh yeah uh it created some knowledge base

uh from the AI agent but it's not shown here but uh basically I can enter this link and it can crawl and learn further about you know different services and it can crawl the website and um yeah learn from it. >> Nice. And can you add um I mean with this knowledge right like you can add things to train it on for context but could you like in regards like what the company does but could you add >> uh could you go back is there the same kind of rules logic that you have with the other agents as well right? So like tools, rules. So So you could maybe add a link like say if somebody says they want to do or go somewhere or do something, you could be you could respond with a specific link or respond with a like a a booking

link or like a form or something. >> Yeah, I mean those are actually >> you could do that using the actions >> basically. Um, if the user wants to make an appointment, you could just say, you know, uh, you you could just give them the link, give a give a link to them. Uh, maybe just, uh, send them, uh, this calendarly link and then you can just enter the URL here. And so just like how you can select things from drop down, you can also put like prompts right here like that and it it will also work. >> Um yeah, so definitely uh you could do that. Uh accidentally deleted that but send them a calendarly link for example, right? Um and basically you can save this. I didn't provide the link but uh basically I could provide the link as well. So you could

put those kind of uh prompts and uh actions like that as well. Or if you have forms uh you can you can send those forms uh form links to people as well. And you can also train your agent by just talking to it. Um if you need to uh talk to your agent and give instructions um just you know uh start your let's say start your emails emails by saying uh hey how are hey hey how are you doing? I mean, we wouldn't probably do that, but you could you could provide instruction like that and um just uh train your agent uh by talking to it. Yeah, it basically updated the persona with that uh information. And then if I proceed, hey, how are responsive with grading? Let's proceed. Let's see if it's going to work. Uh let me send another email. Uh let's say

um uh let's see um so this is not the one uh I will just do it here basically. Uh I will just say yeah that email address I need that email address. One second. Yes. Um how are you doing? Um how is everything? And let's see. And we haven't received the email yet. Okay, I received the email. Now, let's wait for the uh response draft. Let me just refreshing. It looks like there's draft, but that's the old one. Yeah. Okay, I got the draft. Um, uh, it doesn't fall off. It doesn't start by saying that. Uh, but yeah, it's some something I'm going to report to the team. Uh, it didn't get that. Uh but yeah, I mean that's how uh sometimes it's just it might actually like one of the good things about the agency is it might actually decide that hey already asked

like someone actually already asked like how are you doing you don't want to say you want to say back how are you doing right uh so maybe you know I'm right here maybe I'm wrong here but we'll check that out >> okay >> yeah uh so this is really exciting like the possibilities are like limitless this there's just like so many things that can be done with AI agents and and just uh the Gmail agent is another way people can uh you know get AI agents to do things for them so they can save time and uh spend their time on more important things. Yeah. I know that you you also have a similar workflow and you're going to show us your your the AI agent you created with relevance. So I'm leaving it to you. Yeah. So, my um mine definitely is a little

bit similar or well, not a little bit, it's a lot of bit similar. So, when I go through this, I'll just kind of call out that like I get it. Like, it's it's decently similar. So, I don't um I don't blame you at all for if you're watching this and going, "Oh, it seems like it's Well, I I think sometimes it's good to point out there are different solutions for the same type of process and also some are easier to set up and some have different abilities and stuff." So um just to start out with mine, I will go with just going to the first part here and showing um this is relevance AI and I set up >> what is called like a director agent. So I named this guy Dan the director for my company. I do a couple little different things. Uh

obviously do we do this podcast? You can see my screen right? You're good? >> Yeah, I can see it. >> Okay. We do this podcast. We do uh content services. So I make like uh content actually for Jaw Form's YouTube channel and stuff like that. I make content for other companies and then I also have a YouTube channel and I get sponsored requests. So what this director does is anytime there is a email similar to what happens in the jaw form email agents. It kind of parses out what's the question about and figures out what to respond with. But rather than having it just be a general responder, I kind of made it a little bit more specialized where, as you can see here, these agents are made to respond based off of what it determines to be. So it's like, okay, if this is

about a sponsorship post, let's have it respond with this agent and have that agent figure it out. So for the podcast for example, >> Uhhuh. >> you start with this guy and what'll happen here is I I essentially worked through training this agent by using this inventor tool where you can essentially write out stepby-step SOPs. And this is where being organized helps. And I and something I just want to call out to everybody is whether it's through the jaw form agents or through something like this, mark out your processes, guys. Because in the next two years, if you have them marked out, you just hand it to an AI agent, it's going to be pretty much able to understand what you want to do and even give you suggestions above what you may know. So, I gave it a bunch of explanation as to, you

know, what prospects look like for a podcast guest. So, for example, if we're receiving a podcast guest request, I have it utilize a knowledge base I built of like what is a good guest and then I send it to a research agent, right, to check out the website of the person and that kind of stuff. and and uh more so than just draft responses sometimes. Oh, sometimes that happens. It gets weird. Um did you see my text on my screen got weird? Uh whether it's making the response or not is a big part of it. As you can see, there's like a writing agent, but I had a research agent before because say for example, it's not a good guest fit. I just don't want to deal with it. you know, like and we're gonna it's gonna make a draft that represents, hey, appreciate you

for reaching out. Probably not the best idea that we have the podcast together. Maybe some of this is a good idea. So, that's what happens for the podcast. That's a great example of it. Um, and we get a lot of different email requests uh throughout the So, what I can do is just to go to the podcast um on Let me grab my label for the AI agents podcast and see if there's any good examples of this real quick. >> Yeah. >> And if you think about it, even if like if you have an email address, uh if you make it available publicly, you're probably receiving a lot of spam as well. Like nothing. >> Exactly. Yes. You want to be able to sort between real people who are interested in like joining the podcast and just spam. Yeah. >> Yeah. No, absolutely. So, for

example, let's go here and and this is not only in the is it a request for the podcast or is this a follow-up message, right? So, we already had a podcast guest on. This was Evan. Um was a great interview. Um, and this back and forth was mainly led by AI discussions. >> Mhm. >> Thanks. You guys do good work. I'm looking forward to the market team to pump it out. He asked questions on, you know, are there any hashtags or rat codes? And this is a draft that was made for my client success manager to to handle. Hi Evan, thanks for reaching out about social media promotions. We really appreciate you taking the initiative to help spread the word about your episode. Here are the hashtags and handles we'd love to use. Feel free to use any combinations of these that work for your

post. I didn't need to write that. >> Was this generated by AI? Completely generated by AI. >> Yeah. So, um, we have a specific, and this is where sub agents kind of go into it. as I as I mentioned, you know, there's the research part and then I have this thing called Emma the email writer, which is a specifically trained agent on how to write emails. >> And then I have a sub agent sort of in there >> that is the one I've made before called Person Pete where it basically turns AI writing into non AI sounding writing. >> And then last but not least, I have a QA agent. So I I I really want to make sure it doesn't sound like crap to be honest, you know, like and that's >> I just think about it like if this was a team, you

know, make agents that are a team and um make a workflow that that reflects that. >> So yeah, this isn't the only example of this. Um it but it is a a great example of just like okay, I don't we can just store the hashtags and the handles and stuff like that and >> it'll give a response. >> That's cool. Um, this saves us a lot of time cuz we get a lot of spam. And it's not only for this, you know, we we have I think sponsorships is a big one for people who are like YouTubers or social media content creators in general. The amount of work we get through sponsored requests is pretty funny. And this amount of bad requests we get is pretty, you know, there's a lot of dumb like like it's like, okay, I'm not doing this. like it's some

weird scammy small app or some weird scammy uh physical product or they're clearly not like real. Um what I'll do is uh I'll basically go through and check out what what I used to do. Sorry. What I used to do before this agent existed is I used to go through and I'd say okay this is the product. let me look at some basic indicators as to whether it's a good opportunity or not, right? And oftent times it was not a good opportunity, right? Like it just it wasn't. So rather than dealing with vetting that out, I made early stages of agents and then I moved into this situation which was figuring out whether it's a good fit or not, if it is a good fit, respond with an actual showcase of, hey, this is my sponsorship package. this is how we operate. If you'd like

to move forward, let's do it. Um, and then if we received more things, we would negotiate, right? Or at different steps in the process when we'd finish the video and they'd say, "What are the next steps?" It would answer those questions. It's just a a very complex process to some people's initial mind, in my opinion, is not that complicated. Like it's just a really long really long >> and varying long mattering on what the process is step of if then logic, >> right? >> If >> Yeah. I mean, if you think about it, it's it's just like >> if you had a like a real human assistant, right? >> You would just, you know, tell them that okay, you know, if this happens like do this for me, right? And you're basically doing that but you're doing it on relevance uh by with your AI

agents and exactly it's just like that and then you know you're basically kind of kind of writing code but in just like uh regular in English you're writing code in English uh because LMS can understand your uh your you know code your English code um so it's it's amazing it's it's it's yeah we live in an age that's incredible. Yeah. >> Yeah, you exactly get it right. It's it's like training the code in such a way that >> um like if I were to I trained people on this and I actually funny enough and this is where I'm I'm I have a very strong stance about meetings um and that is obviously you don't want too many people on there. >> Mhm. if they're not needed, obviously don't need them to be there. But the other thing that's very strong is unless it's a personal

conversation, there's no reason that any projectoriented meeting shouldn't be recorded. Um, now with the new advent of these meeting generators, you can actually make agents like this literally by going through all of the recordings where you would have had a meeting and trained somebody and then taking the transcripts and figuring out the the logic of decision-m based off of the training. So, you don't even have to write it out. So, I all these agents were set up by um somebody that I'm paying on the side that I I could have said this up myself. I'm very advanced with this stuff, but I literally gave them here are all the trainings for sponsorship management. Can you make knowledge bases and can you make agents based off of that? And that was it. Now, that's something that I think is very valuable to you if if you're

listening to this. Like don't limit your, you know, frame of mind to, oh, I have to write it all out. Oh, that's going to be so much work. You could talk it now. Like that's something that's such a such a hack from previous situations. If you trained if you've trained anybody once like just record the call and then it Yeah. What some people who are seeing this like they might be scared like this is so much work >> but um I think what you can do is like you can just start small like just just >> like pick just one thing that you do constantly like maybe people are asking about let's say if you're a small business maybe people are asking about your you know opening hours right you could just have this agent and you just say okay my if you if someone

is asking about opening hours Right? Uh you just just uh send them in information otherwise don't send any mail right you could just automate that part and you could just start with that and that will be your like small AI agent that's actually handling that thing right once you do that once you feel the confidence that this can accomplish more you can just gradually add more and more automations that way and this would this is actually very healthy like if you try to do this all at once That's not very healthy. >> It's just going to, you know, if you miss some things, then you maybe you will miss emails, maybe you will, you know, it's it's just going to get too complex very quickly. But if you just add things slowly and gradually and watch it working well and perfect it, you know, just

make it right and then add add another one and make it right another one. It's an iterative process. If you can actually do it that way, maybe you will reach that this kind of like automation like this this kind of prompt a year from now uh and then it's not going to look uh you know too much like too complex to you at that point because you you gradually edit all these prompts. >> Yeah. No, that's a very good point. Um, I I and and I want to call out what you just said again and kind of maybe also touch back on what I said a minute ago in that context. These calls that I recorded have happened naturally over like years, right? Like I've been doing this for a couple years now. So it wasn't like I said, "Oh, I'm going to do all

this at once." Well, the nice thing is um I actually went through and set up all of these processes as I went along to save myself time. It's always like it's and the great thing about agents is this kind of speeds it up. But previously what you would have to do if you wanted to like optimize I know you had a book called you know buying uh I know getting you back your time was it what the specific >> automate your busy work >> automate your business and then the subline was like buying your time back or something >> right there's always sublines to the I forget but >> yeah so automating your busy work that's something that I'm sure we can agree on is so important and this is one of those things where all you need to do is start small you start

with templating it for yourself if you're doing it yourself. You like you write out the workflow, you template the parts that can be templated and then maybe if you get good at automation, you can automate stuff. And then now the last step is okay, we'll just AI agent the portions that are harder to quote directly automate. And you know, if we go through my email inbox now, I'll just show you the list of different things that it's it's doing now. you know, we have different categories like let's go to the sponsorship, right? Um, we'd love to feature your YouTube channel to have help you expand your reach. Um, this is a company outskill where they originally reached out world's first AI focused educational platform. Been following our YouTube channel, done 500 partnerships with people over the last few years. Um, long story short, the agent

figured out, all right, looks good. here's my deck, right? And they said, uh, after reviewing it, we're excited about the potential for a barter collaboration. Um, and then we had another response ready, right? Because there was kind of a weird curveball they sent with the whole like newsletter thing. We don't really mention newsletters on our YouTube channel. So, we asked some questions, right? And the agent came up with these questions. Um, let me make sure I unread these so the guy doesn't lose them. Um, question about the podcast. Let's see another one for business consulting. Oh, here's another great one. Um, this is basically something where someone asked us a question. >> And it's simply it doesn't fit under our prerogative. There's no sponsorship. There's no content services. There's no podcast. Kind of feels like spam. >> Well, get and also we don't do >>

info. We don't do >> No, we are definitely not doing those stuff. >> Yeah. Exactly. Yes. And we appreciate it and we go Yeah. After careful consideration, by >> you know, like no, no need to Yeah. >> By the AI agent. >> Exactly. Exactly. >> Carefully. Yeah. He He goes, "Uh, I don't know about that." Um, now let's see some other situations here that are good. So, um, here there's another draft here. Hi, thanks for following up on the video draft. I wanted to update that. The first video draft is currently in progress. The recording it has been completed and our editing team is actively working on it. So, this is a draft that was written essentially based off of looking at our project workflow and answering the question based off their question. Uh the nice thing is with the sponsors agent is we got

it to work with our notion. So what does that mean? Um our notion manages our content and based off of status names you can get answers to questions. So if it says like editing right and they ask a question about a video we say okay what is the spons you get you obviously have the person's email which is attached to their sponsorship you know card in notion and then that would connect to a list of videos and you know you can get an answer like this so it can be pretty advanced it can be very basic like ID saying which is literally just like a spam remover which helps everybody like that that fits in every category. of business or it can be as advanced as say somebody's requesting the status of a widget being produced. If you have a database and you can connect

that database to the agent, it could get the answer especially if they send you their order number or something, right? And the order number is tracked in a system and you know other systems you have internally are updating statuses properly rather than you having to go and check the agent can go and check and then respond. So yeah, this is this is the power of of what this stuff's this stuff does, you know. >> Cool. Very cool. >> Any other uh questions about like what I'm doing with this? >> No. Uh it looks great. Uh very great. Uh just like you're just saving so much time uh by letting your AI agent process your emails, preparing drafts, and um as I said, like it's an iterative process. Um, every time your AI agent doesn't perform something that uh you want it to perform, you can

always go back to your prompts and add something else and and keep keep reviewing all the work it's doing and then keep improving it at some point. Like it's just you don't even have to improve it. Like just it just works perfectly and just you just save time. it's h it doesn't you know take breaks, it doesn't go vacations and you know it's just constantly available all the time uh during holidays as well. So yeah it's it's amazing uh to have AI agents that can actually do uh so much these days. Yeah. >> Yeah. So my you know my recommendation for all of you is check out what we're doing at Jot Formi Agents. Make your own agents. get answers to the questions without you needing to do the dang work. You know, it's it's so hard nowadays to manage all of your different distractions,

you know? And my recommendation is just simply stop being distracted by your inbox. Uh it's the worst feeling in the world when you feel like you're overwhelmed by something and your inbox should not be one of them, especially with these. And whether that's, you know, the inbox of customer support ticket requests on your website or now your email, um, Jot Formi Agents is going to be that solution. So, just keep on the lookout. We're going to keep updating and doing more cool stuff. Um, the product is has been doing really well. How have you how how's the performance of like growth been recently? Has have we gotten more more and more users with agents recently or what? >> Before I get to that, >> oh yeah, by the way, yeah, buyin's book. It's the best book. you you were you were talking about this uh

automated do less >> achieve more and save your brain for the big stuff. So >> exactly exactly exactly >> that's that's exactly how uh how AI just can help you. You do less uh but you achieve more because the AI agents do it and it's and you save your brain for the important stuff because uh the all the stuff that you repeat are actually handled for you. Yeah. Um, so you were asking about the uh what were you asking about? Sorry. >> I was just saying like hey how's the I mean how have we found like users and stuff uh have been what have anything they've been saying recently about like >> growth? It's been top at uh like the Gmail agent was uh one of the top ones. I think it's was third or something on product count and like we had thousands of

people start using it right away and just it it turned out great just started in such a great way and um and people were actually using it like they were creating all you know the uh giving giving it um um giving it access and then being trained on and then preparing drafts and just like our measurement of success is like um how much do people edit? Like when we create a draft, do people before they send that reply, do they edit 50% or less of the email? If they send it if they edit it, it's fine. But if they if they send it more than 50% of the message that we assume it wasn't as good as it should be. or if they cancel it and just write a response themselves, we assume that we didn't do a good job. And that's the measurement uh

we are using to kind of like grade our success. And so far it's been good. It's been like uh half of the drafts were actually sent more with uh less than 50% changes. Half of the drafts we created. >> How are you tracking that way? Out of genuine curiosity, how you how do you wait you how did you get that answer? I have no How does that work? uh basically because you're giving us access to Gmail the uh it's able to like look at uh look at the drafts and uh if you send a response and if you if you didn't if you changed it little bit we assume that the draft was successful and if you changed it a lot or if you just cancelled that draft we assume that we weren't successful and I think the reason it's it's it's it has over

50% rate of that successful uh like uh drafts is because we have been in beta for like three months and we have we had a great team and they have been working on this product for a very long time and they had they had around hundreds uh beta users they were constantly working with those beta users the imported products so that's uh that's why you know right out of the gate uh this start the product started really well and did really well and of course our goal is going to be to increase that rate like how can we make show that drafts get get better and better and improve our responses. Yeah. >> Okay. Yeah, that's a Okay, that's interesting. Um yeah, 50%'s actually pretty pretty good. Um, I'd say that, you know, the something I do want to call out about this and, you

know, before we close it out is don't do not expect like from any agent absolute complete 100% accuracy in response in regards to complex things. What I was showing you was like a ridiculously overengineered like example of it. And still Luca, my guy on my team, does have to tweak those emails, right? >> Because it's not a person. >> And we're in like we're in like the first year of this being a thing. So 50% is amazing because what I like about it and what was really big for me just from what I heard from my team was like, "Hey man, this is saving me a lot of time because at a minimum I already have an outline of what the email should be and then I just got to like tweak a couple things, right? Whereas even if you didn't even if you don't

have that, you have to go copy that email, maybe put it into an agent to be like, could you write me an email response? All those little tweaks here and there. He's like, we used to actually send it to our Slack and have a Slack agent that would basically be like, is this a good response? Press approve or deny. And then you could reply in the thread in order to make it change. And what we realize is it's actually just easier to just put it in his email drafts >> and then when it's in his email drafts he can like just easily tweak it. >> So I like the way that you went about doing it. It's very funny that I didn't even know you guys were doing this and then I was like ah it's probably easier to just put it in the drafts.

Um yeah so very very cool stuff. >> We we didn't want to just send the emails because it's just uh emails are very sensitive. Uh but creating drafts is like uh you know it's even if you know even if it gets it right one out of two emails I think it's good because like it's just very easy to discard the draft and write your own response. Uh but if it's right you just click on send or you just add some you know additional sentence or two and click on send and you're done. You just saved five minutes or 10 minutes or half an hour writing an email from scratch. Yeah. You know, I I agree with you, by the way. I I do think it makes sense for for all of purposes. Like, there is no reason to, in my opinion, kind of I don't

I don't really think you should send emails like without um verification, right? And I I think we're not at that point. So, I I like that you're doing drafts. Um maybe some people are comfortable with it but >> I've had experiences in the past that I asked the question like a customer to a customer service for a you know for a product and I would get this copy and paste answer that's that has nothing to do with what I asked right someone actually like they didn't they didn't even like properly read my email and then just copied and pasted past the long answer but here we have this AI agent that actually really understand what you're asking and then because it knows like the all the history of your answers it can prepare a really good answer and that's just blows away like that you

know copy copy and pasting answers um you know uh and you know uh if you are very busy you don't have time to like spend so much time writing emails so I think it's it's just um as as the book title says uh save your brain for the good stuff, right? Uh let the automation handle the repetitive tasks and uh and you know the automation can write the email and then you can just add some something to it and just you know uh make improve it like you take your time to improve it and then you will provide a much better service and everyone will be happy. Yeah. >> Yeah. No, I think I think you're right on man. It's it it's I think it's the better approach you're doing to be honest. All right, man. So, with with that being said, please everybody go

check out Jot Form email agents. Uh we're really well Gmail agents, I should be clear. Just are do you have any plans on uh expanding no to the suite of email or is it just going to be Gmail you think for the p foreseeable future? >> Uh yeah, I mean uh we have uh new agents coming. Um we we will have another agent coming next but you know as you know we don't announce them like just we just keep keep it >> oh okay keep waiting very big uh like the presentation agent was huge uh I don't think we did an episode about that >> we should definitely do that we should definitely do an episode as well >> and the Gmail agents has been a huge launch and the next month we are we have another agent that's coming out that's that's that's also

going to be a huge line because I know the huge number of usage we see with our beta users. So that's also going to be huge. Yeah, we we we should probably cover those agents uh in the next episodes. Yeah. >> Okay. All right, man. Well, I appreciate it. With that being said, thank you so much for listening to this episode of the Agents podcast, and we'll see you in the next one. >> We'll see you the next time.