An agent you & your team will actually use

0:00This agent can sell for you, and unlike Hermes and OpenClaw, it's not just meant for nerds like me, which means everyone on my team can set it up and talk with it on Slack easily. It's called HyperAgent, and you're going to watch how to set it up, how to use it, how to get sales coming at you. Alex, show me how I can get a customer with

0:17HyperAgent. Let's do it, Andrew. So, we are here in Slack, and this is where I work with my team of agents and my team of humans. I've got a couple different agents in here, as you can see. This one called Signal is my sales agent, and I can @ mention Signal directly right here. And Signal is actually a template that anyone can

0:37install on the HyperAgent platform. We'll give them links to all it all below, okay? Yeah, links and credits to to get started with your first run. Um I'm going to fire off this message message to Signal right here. It is going to be asked to go and learn about this company, Swiss Solar, which is a really cool local company. I'm here in Waterloo, Ontario, uh which has some amazing engineering talent, and this is a company that is building uh

1:01self-cleaning solar panel technology. Very, very cool stuff. And you'll see that Signal just replied to my thread. So, Signal with its little astronaut avatar right here is on it. It is going to start running that outbound sprint. Signal has a set of skills that we've set it up with to be able to do this. What I can do is fast forward to one that I've run. This outbound sprint that Signal does takes about 20 minutes, and I'll show you how Signal is set up, and then I'll show you the output of one of these outbound sprints, if that sounds

1:32good. Okay, yeah. Perfect. So, this is Signal's profile first off on the HyperAgent platform. So, out of Slack and now into the HyperAgent web app, you'll see that Signal has access to several different tools right here. So, it has ability to draft in Gmail, it has ability to use Apollo to find and enrich contacts, find buying signal. It's going to be reading and writing to Airtable as its CRM. It's going to be booking meetings in Google Calendar and of

1:59course we just saw it working on Slack. Alex, let's pause there for a minute. I I have an I have a hyper agent. It's the only agent that I have in our team Slack. Just for simplicity, all I had to do was just click on the tools that I wanted. I got the OAuth screen. I said yes, I want access. I want to grant

2:13access and then that was it. Here they are. Yep, exactly. So, The only one that was Sorry, the only one that took a little bit of time to set up was Slack. Um I had to click it and then follow some instructions to get it set up. That's right. Yeah, it worth spending a second on those because the ability to mention your agent and also have them mention you and mention each other in Slack is one of the things that our our

2:35customers are loving about Hyper Agent. So, right here in the Slack setup, uh you can give this bot its own identity. That's Slack's term of a of a bot, but this agent uh has its own app essentially on the Slack platform and that's why it has its own name and its own avatar and everything. I can also set the way that I want this agent to behave. So, certain agents I might want them to only respond when

2:59mentioned. I might also want some agents to be very proactive. And so, if my team and I are having a conversation about our next event plan or campaign plan or something and I have a marketing agent in there, uh I might want that that agent to be proactive just like a great employee would be proactive and say, "Hey, notice that you're starting to kick off planning for that next event. I went ahead and did it some initial uh research on speakers and found some venues for you." That kind of thing is where you can get super proactive with these agents. So, that's in your control

3:27by channel right here. That I didn't know. I've had it set to I think the default was only when I mention it and what I want is mention and thread replies. And now, I've got a couple of other questions about when I'll use the relevant, but let's keep going. All right, sweet. So, it's got this suite of tools. I mean, Andrew, tell me like what would your dream sales representative, dream sales employee do all day? What is like the the loop that they would just kind of keep running to

3:50help you build your business. Figure out who the ideal customers are, then find their contact information, create something custom that doesn't feel like I'm just sending them a templated message, send it out to them on uh by email from my email account with only only if I say yes to sending it. This agent is running in what we call live mode. So, this is a single thread that just runs in a loop 24/7 and you set the guardrails, you set the specs for what it can and can do. This particular agent, every time that loop

4:23runs, is going to do a series of things. It's going to uh check the inbox for any inbound leads. So, I actually have it uh in my inbox. It is allowed to respond to warm inbound leads. So, this came from my agent. It's using my inbox right here, but a warm inbound lead came in. We're going to get on that right away and book that meeting. We don't want

4:43that to ever cool off. It's also going to find us new uh outbound opportunities that we can prospect into. But, rather than just create like a plain text email cold email out to those prospects that's going to be kind of generic and flat, it has packaged up its work for us in this command center and it's done something that our our customers are finding like pretty unbelievable results with. It's actually generated a unique landing page for every buyer. So, we're still on the the solar panel cleaning example here. It has found industrial scale solar panel projects that are

5:22active, that have real buying signal. This is they would be in the market for this kind of offering right now. And for every single one of them, it has built did it know? How does it know that that this company right here is interested in buying? Yeah, I'll show you it sites it's all of its sources on this page. So, what it has done is create a page personally prepared for these individuals who would

5:41be responsible for this buying decision. It has actually created a video directly in line using images, shots of of the product software, some of its findings from the research that it did. It's got some animations on here explaining how the technology works. And then it has linked over to its sources of different announcements that have been made about this project, what's been discussed publicly, all the way down in the bottom here. As the seller, I can inspect its work, but even as the buyer, you might be curious to be like, "Where did you find out about that?" And it will actually go through and be like,

6:14"Yeah, I read the whole PDF. I read the 5-year rollout plan, and I can tell you exactly when and where our product might be a fit for you and the types of business outcomes that it might have. So, this kind of thing like you would never have a junior BDR able to do this kind of work at scale.

6:31That would be completely impractical. Okay, so now imagine I'm selling AI services. I've interviewed a few people who are helping companies get up to speed with AI. They will use this agent to find potential customers, to create whole landing pages telling them why they should sign up to work with them to to get up to speed with AI, and explain

6:52why they were picked and all that stuff. Got it. We have a marketplace where people can publish their agents and skills that have been working for them as templates for others to use. And very similar to the idea you just described, Corey Gannum is an awesome YouTuber, and he had this concept of finding small businesses that have great reviews, but either no website or very poor websites. And so, I'm just going to pull up his skill right here. This agent will go and find those businesses and build a sales command center very similar to the one I just showed, and says, "These are the businesses you

7:25should go and sell either your AI services or web design, whatever AI package you might have, and it will find the perfect businesses for that." Very, very similar concept. So, it's it's crazy the kinds of things people are doing with the platform. I record a weekly news show with him, and he said that this is Yeah, he goes, "My hyper agent is bringing me customers." Okay, I get that. Let's continue. Now you've got the landing

7:46page. Let me see how the email goes out. Or what's this? What is this screen actually, the one that you just showed me a moment ago? This here. Uh let me head back here. This was actually for a different company that I was running. Exact same motion. But this was for uh a different company. So I'm using the same template uh sales agent to build the exact same thing for another company. You see the

8:07output is a similar shape. I see. And you can see it on your website or Slack. Why would I not just stick with Slack and see all the outputs on Slack? So Slack is a little bit limiting in terms of how much uh visual it will bring into the thread. So you'll see even with my thread with Signal right here, it's giving me this nice crisp text update. I actually like in Slack keeping my agents on the concise side. I don't need a big verbose response. And then it will link me out to the command center it created. So that's that's what

8:35it just created right here. Got it. Okay. So now this is all in your command center. How does it go out to the customer and get the business? So you'll remember we saw that agent was hooked up with the Gmail integration. That is exactly where these are going to happen. So it in this setup is creating draft emails and it's going to ping me

8:52when the next batch is ready to review. I've been instructed to keep things super tight and concise here. I can have that link over to the artifact and then I can have a look at that artifact. Does everything make sense? Does it line up? Does it seem to be using that customer's language? And then I can be the one to hit send. Uh some people are actually uh very trusting of their agents on hyper agent and they're giving their agents their

9:14own inbox. So their own email on Gmail. And the agents have full latitude to be able to just send messages all day long running in that loop. And um they've seen a lot of success doing that. So you have that control based on your comfort level and how much autonomy you want to give these agents. I haven't thought to do that. And then how do you give it access to its own

9:32email account? What do you like? set that up on your Google Workspace. right? then tie in the account to HyperAgent the same way that I've tied in my my personal inbox right now. So, you just It's like standing up another employee, like onboarding another Got it. You just basically it's creating another employee and then that employee has access to this tool. Got it. That's a That's the easiest way that I found to

9:53set that up. Okay, and now how hard is it to train the agent to understand what's in your inbox and know that you got a response back from a potential customer versus just a random email? Yeah, so that's all part of the loop. And so, when we have our agent running in live mode right here, which I'll go back to, all of those instructions are laid out for it each time. So, every hour, basically this thread just gets rerun with this live mode prompt to keep that agent running in this loop. And it's got a set of rules in here around how to manage inbound. What are we looking for

10:30with inbound? What are our guardrails? What process do we want to do? Maybe if an unfamiliar email comes in, we want to run some enrichment and some checking on Apollo, do our own web research, and then decide, "Hey, this is a qualified lead. I should reply right away. This is a little bit iffy. May Maybe I should

10:44flag this for my human before I reply." It is going to run uh nurture as part of the loop as well. So, it's going to look in CRM. You can hook up something like Airtable as a lightweight CRM. And it's going to track all of its activities perfectly and say, "Hey, this lead is starting to go cold. It's been a week since we've heard from them. Let me share another helpful resource with them." And then in every loop, it's also going to run additional outbound and keep that pipeline full with new prospects. And then, it also is instructed to uh keep learning. So, the self-improvement loop is critical here,

11:16too. We don't just want to be doing the same thing over and over. As we start to earn those responses, we're going to log those into skills and memories that this agent is going to self-update on. So, within our team of agents right here, here's Signal that we've been looking at as the sales agent. It has the ability to self-update all of its skills and all of its memories based on what is actually working. And it can share those learnings with the other agents. So this is where things get really cool, which is hey, you know, a sales and marketing sync basically like a stand-up almost happens between the

11:50sales agent and the marketing agent just directly in Slack where you can review it if you want, but you don't have to. And you can just kind of let them work together. But look, so Signal, the sales agent, has tagged the marketing agent saying, "Hey, I've got some hypotheses on what's working in cold outbound. When we're running ads, consider these as things to start testing as well." And

12:11they can actually share those learnings. Multiple agents working from the same context layer. Oh, wow. I That I didn't know how to do. And here's why. Right now, I call our hyper agent hyper agent. And so we at hyper agent it. How do I get it so that I can have each one see its own personality, its own face,

12:29its own name, its own everything? Yeah, absolutely. So I'll show you the very simple setup process for creating an agent. So we have this distinction in the platform between talking to what we call the generalist, which is just sort of like at hyper agent. So you can come in here and just get started on on work and you're not speaking to what we would call a named agent. But if I dictate and say something like uh create an agent that will um run SEO and GEO checks on my website 24/7 and

12:57constantly pitch new ways to improve. Uh so I'm going to ask the uh hyper agent generalist to create that for me. It's going to start to go to work and it has tools that it can call to go and create a named agent. So as you can see on the left side here, Signal, Funnel, Redline, Inbound, these are all my named agents. We've been focused on Signal so

13:17far. So first you need the named agent. So you first you need that entity. And I'll I'll let this run a little bit so you can see how it it presents the the new agent creation and then we can go through the anatomy of some of that agent. Okay. [snorts] But what do you what are your thoughts on this concept of having multiple agents with different names for different jobs? Cuz it's something that's creating It's Yeah, it's it's becoming a bit of a hot topic. Like there's different schools of thought. Some people now are like, we just want one super intelligence that knows everything about

13:44our company for every single function. Our customers are starting to see it differently. They want the guardrails. They want the mental model of, okay, I talk to you for sales, I talk to you for marketing. You can access my email, you cannot. You can access Stripe, you cannot. And having that delineation. What are you seeing? No, I absolutely I know for my team I need that. I want some clarity for them to have them know, okay, this is the one that creates our landing page. In fact, we use HyperAgent to create landing pages for almost every single video that we post up on YouTube. It was a hassle

14:11to go into Lovable and create the landing page and all that. So, we created HyperAgent. We now say, "Look, I need a landing page. Here's the thing that it gives away." Boom, it gives us a URL. Anyone can access it. So, you want to create these different named agents. I'll let this one keep running. But uh you also want to add your agents to a team. So, we have this concept of a team, which is this shared container of multiple agents that can share skills, share memories. We also have the ability to check on all of their um their kind of vitals and their health status, see the cost breakdown

14:42per agent from what we call Command Center. Customers really love this as well about the cost transparency. You can go by agent, by thread, even by tool call, or by model per thread. One thread is is likely going to use multiple models for different jobs that the agent can orchestrate. But then once you have that agent created, and we'll we'll come back to that one once it's ready here. I think

15:01that's uh HyperAgent still cooking away. But there it goes. You can see here, creating agent. So, that's going to turn into a a one-click install when it's ready. But if I look at one that's been created, so Signal right here. You were talking about, I want to be able to at mention this agent in Slack, right? Not just at the HyperAgent platform, but at

15:17Signal. So, that is super simple. Once you are uh looking at this named agent's profile, you'll just go to Slack. I have it already set up here, but you'll just select bot identity. And then there's two clicks that you'll have to do on the Slack side, where you'll copy over a token from the Slack side. And then for the finishing touch, you can just create a avatar right here for your agent. So, I was I was good drawn to this like astronaut style one for Signal. You'll just grab that image. You can upload it on the Slack side. We're We're making all this much simpler very soon, too,

15:52with some improvements from Slack. And there you go. Now you have your Hyperagent entity, your named agent on Hyperagent, and your Slack bot that are one-to-one. And so, you can actually now at-mention the name of that specific agent, and you'll see the visual identity of that agent reflected here as well. Can I see how on Hyperagent site I can add a team member, like my chief of staff, and say she can control the creation of these agents, she can check

16:19and edit the ones that I've created? That's right. Yeah, so within this this concept of a team, you can of course have multiple agents and multiple humans. And so, I have my teammates in here. You'll see a small but mighty team of the two of us in here. Different roles. You can have an owner with full control. You can have an editor that can run things and make targeted edits. Or you can have a member who can only use the agents. And so, for example, we're seeing AI agencies that you were mentioning earlier, like the the Corigin audience or the Nate Hurk audience that we've we've been working with as well,

16:55where they want to be the owner of the workspace for their clients, and then their clients are only in here ever as members. And so, they never even need to be in Hyperagent because they can just interact with that agent through Slack or WhatsApp or Telegram, and the consultant can be the one that's responsible for building, editing, maintaining those agents. So, one team, multiple humans, multiple agents all in

17:18one circle here. I will say this, anyone on my team can just interact with Slack if they're in the channel. I don't I don't think I had to add them on here, right? Correct. Correct. Agent is in there. So, I want to talk about why this versus Hermes Agent, Open Claw, what are you seeing? So, the big thing that we're building around as a core architectural choice is this multi-agent architecture. That's why I was curious to get your thoughts on this where I don't want to have to go buy multiple Mac minis to go and have

17:48different agents with different jobs. But I I don't have to do that. I have a I have a Hermes Agent and it's on a laptop and there are multiple personalities in there. That you don't you don't need a multiple you don't need multiple laptops for that. That's true. Um so, you would have multiple agents uh all on the web and on

18:06the browser uh and running in the cloud. So, there is a there's an ease of setup. I don't know, what was your experience setting up some of those other tools? Like, is it is it as is it as straightforward as what you found with Hyper Agent or So, I found that uh setting up Hermes Agent was really easy, especially today with the desktop app. The issue I had with it was connecting it to Slack. I spent a Sunday and I couldn't get it done. It might just be me. I know other people have done it, but it's not as easy and straightforward as it should

18:32be. as um people may know, Hyper Agent is the the new standalone product from Airtable. Airtable, if you think about the kind of pedigree of our company, we took this very powerful technical capability of building relational databases, building apps, and made it really accessible through a deep investment in UI, UX, and smoothing out all the things that should just be invisible or one click. And we're aiming to do the same thing here with Agent. So, my hope is that you would feel uh a much smoother and more intuitive

19:01ease of setup and use. When it comes to Hermes Agent, it's on my computer, it's my control, I'm the one who's in charge of it. I can't say to somebody else on my team, "Can you go do this with Hermes Agent?" It's It's a problem. So, it's my own personal little thing that I I do love, but I can't say

19:17Lauren, I want you to do it. Number one. Number two, if anything goes off here and there and my laptop is down, the team can't access it. So, I can't have mission critical stuff on that. Yeah. [snorts] Totally. Yeah, really well said. Yeah, there there's this pattern that we're seeing within companies big and small where one person who's really enthusiastic about building with agents will build something incredible and then they're the only one who is

19:41able to use it. And I see you also have all the models built in here and it's it's just works, right? You can bring your own subscription for Open AI. Claude is a little bit more restrictive around that, but you can actually plug in your ChatGPT subscription, which we just introduced I not know it? Dude, I've been using

19:55this for weeks. I had no idea. Yeah, yeah, yeah. You should get this set up. So, just come over to settings, hit AI providers, and then just connect up your ChatGPT subscription right here and it will use that first before tapping into your Hyper Agent credits. Is it theoretically possible for me not to pay you anything, but keep using my

20:11subscription to Open AI? You can get pretty close. So, the one difference would be things like there's these small fees for other tools that Hyper Agent will use and we give it an initial balance to cover those things. But, for example, Exa for search, browser base for allowing that agent to actually use its own browser. You can see how how negligible these costs are, but yeah, basically 99% of the way there you can run on just your ChatGPT subscription if you have a a sufficient

20:40enough one. All right, there it is. Like I said, we'll have a link to it below and Alex, you've got a URL that we can give people that would allow them to get extra usage for free. I love it. We've been using it. Check the link below and now that this is over, I have another video for

20:56you to watch right here.