How to Build an AI-Native Company: Humans + Agents in One Workspace | Kylon Walkthrough

So there's a lot of talk about what it means to be AI native as an individual or a business and there is a massive amount of utility for that. In you or your team working side by side with a team of agents that can all access shared data and context and integrations to work together and to get things done, right? That is what it means to be AI native, to be fluent, to be fluid between agents and human workers. And so in this video, I want to show you a system that unlocks and enables that for your business. And you can think of this as a company furnace. It's a UI UX play and environment that sits on top of your existing agents and human workers to let you collaborate and work together and build things, actually get work, real work done. It is a pretty cool system. And so what I want to do is show you how to get onboarded to it and what exactly it is first to start off. And then we'll walk through getting some actual builds done, things like a research brief and app, just standard white collar work. And the point I want to reinforce throughout is this shift that you can, if you take the leap, make to an AI native ecosystem. And this, I think, is a great route to do so. So the platform is called Kylon. They did partner with me to make this video happen. And again, you can think of this as a company furnace. You already have your company and your business and the way you work. This sits on top of that to let you better utilize all of those resources by working synonymously with agents. And there are three things inside the platform that enable that that you'll see in a second. The first is connections. That's basically integration. So it's how the agents can access all the apps that you do your existing work in. Second is the wiki. It's basically shared context and data so that you have this company brain that just expands over time and learns and knows how to do things. There's also a genetic memory which ties into that. That's basically the agents individually learning and getting good at things over time. And then that third enabler is rooms. That's kind of the internal structure, kind of like a Slack channel that you're already used to for messaging and for building and organizing different tasks and initiatives. So we're going to jump into my computer to kick off the build. Now, if you do want to follow along, it is free to do so. You can go to this link on the screen or in the description or comment K Y L O N under this video. You do help support the channel when you do that, but also you're going to learn better from this video if you just follow along with what I'm about to do. And again, it is free to do so. That said, let's jump into my computer. OK, so here we are at Kylon.io. And I want to walk you through the setup so you can get a feel for how you build into this harness as a company and as a business over time. Let's click get started. And so overall, what this helps you do as a company is become much more A.I. native, right? You have a harness, you have shared contacts, shared data, shared integrations in a dedicated workspace. And we're going to set up a kind of agent to help with that. Let's call the agents A.I. Wizard. We'll click continue. And now here we can start to see rooms. That's one of the principal features and aspects of the platform in a room is a environment to which you and team members can interact together on a dedicated subject with your agent. So let's say that my YouTube channel is the business. We'll click confirm. And there are three core aspects of the platform that form this kind of unified environment for your team. That's connections, wiki and memory. Now, connections is important. So let's start there. We can click right here. And these are going to be all the apps that your existing business runs on. Right. So we have email, we have projects and code marketing and growth business ops. There's over 3000 connections, which is one of the largest libraries I've ever seen available within the platform. And so what happens again, if your whole team is on here is all of the contacts from your unified business apps get pulled together into this A.I. brain for the organization. So when you're onboarding your team, the first thing you want to do is go ahead and connect all the apps that you all uniformly run your business on. We can also see members here. We have users. So that's going to be your team members and then agents. And this is where it gets really cool, right? Because this is a native agent and human workspace. You have humans, your whole business working together with many different agents. It's just like another team member, but a artificial intelligence, a silicon team member, so to speak. And then you have this unifying wiki, right? Or this shared data and context across all the agents in both your feed. So messages or rooms where you can work on projects. And that exists, for example, let's go ahead and create a new room. And let's say, I want to call it script reviews. We have our guidelines right here. We can invite different collaborators to the room. And then right here in the top, we have apps, automations and files. So that's what forms that wiki. So we have files, that's shared context. That's reference information that both you and agents or your team members can add to the channel, right? We also have automations, that's to automate whatever's being built or certain agentic work or reporting, for example, in analytics. And then finally we have apps. So these are mini apps that can be built within a dedicated room or a channel. And so let's say we really do want to write a script. What I'll do first, I'll go to connectors. Now I've connected my Google Drive and you can do this times all the apps in your business. And then here I'm going to select the room. So we want to work in the script-review project. Under connections, we have Google Drive. And now I'm going to go ahead and call my AI wizard agent. And then I've given it this prompt. So I need a research brief. I want you to pull the key themes from my Google Drive script. So existing tools that I've written about in the past. And then pick out three that I could potentially make a video out of. Let's go ahead and let that run. And all of this context and this conversation in these files are going to be saved as we go in that files wiki component of the room. Okay, here we go. We can see the brief that was created and it's pretty cool. So I've done videos in the past on Notion, Zapier and Todoist. So it pulled from those existing scripts and then it's gone ahead and created this brief for me. Let's click on that. And so it's proposing a video about do productivity tools actually save a solo creator time where I'm testing three productivity tools as mentioned. So these ones, let's just follow up and let's say, okay, what's the script format I've used before? Can you pull that from Drive and add to this script to adhere to that format? Now we can continue as we'd like to conversationally. And while that's running, the value of a system like this where multiple agents, right, we can create other agents, we can talk to them. We can do the same thing with humans, can all work together and build and improve and iterate on things using data like this, using your Google Drive scripts from a year ago that the agent can pull from or Slack messages or your CRM and HubSpot, whatever it may be. All of that becomes synced and becomes uniform. And so the more people and the more agents you add into a system like this, then the more valuable, again, it becomes. Okay, so here it pulled from a PixAI script that I wrote, same with Storyblocks and a third one there. So we can see now it has a format of the video title on mail text, concept, outline, and then the full script after that. So that looks great. That's exactly what we want. And since it knows this now, right, this is a document that is saved in the context of the room. So it'll understand and remember this format for later. And anyone else in my team or other agents can learn from that because I'm beating a dead horse here, but the data is shared. So that's something on the script side, but let's take it a step farther with an actual app. So what we've done here is worked on a script. Let's go ahead to a new room. Let's call it app builds. Let's say build apps for my YouTube channel. We're going to add our agents. Now let's head over here to members. Let's go ahead to agents. Let's click new agents, and we're going to call this coder. And we're going to say the specialty is to code apps. We're going to click create. Then let's go ahead and add skills. So we can see the build app skill has been added to the agent. So now let's go back to the room we just created of app builds. And I pasted this prompt. So build me an app called video pipeline. So a way in which I can post scripts for the channel, make it easier to manage a whole lot of scripts and projects. And we're going to reference the Slack for the business that's now being connected, and then just push it. And this is going to be hosted in that app section so everyone in the team can use it. So that's all great. Now the final big feature I want to showcase here is the ability to add external AI agents. So what that means is it's very easy to shift into a system like this over time because you don't lose the siloed data from the current AI systems that you use. So what we can do is go back into the members page where we have our human workers and our agentic AI workers. We go to create, we go to create, then hit new agent. And then what we're going to do, let's call this clod, is instead of Kylon hosted, we're going to go clod code. I'm going to click next. I'm going to copy the prompt. Now we're going to jump into terminal. Let's open up clod code. And then I'm going to just paste this prompt. And now this is going to connect the agent. And now this gives me the ability within the platform to query my external agent. So I could say, hey, clod, go pull that document from nine months ago that you and I wrote for the app that we built and use that to build this new app, right, as a reference inside of Kylon. And that's one more layer of having your team, internal agents, so agents inside the platform, and then your external agents as well, all working together. Okay, we can see the demo. Let's click right here. Let's make this a little bit bigger. Okay, and so we can see this is basically a project manager for YouTube apps. So all videos, ideas, stage, research, scripted, film, publish. And if I click on one of these, I can go ahead and edit it, and then I can change its stage over time. So that all looks good. That's exactly what I want, but I do want to iterate on it. What I want to say is make each project drag and droppable through various stages. So let's let it make that change. Okay, and here we can see that was updated. So now we can drag and drop, and that looks good. So if you like your internal tooling, right, looks good to you, we can go ahead and just click publish. And let's say everyone in the workspace can access the app. We'll click publish. And then now we have an app that everyone in our company and all of our external and internal agents can go ahead and access on our behalf. So that overall is the platform. And I want to keep this short to emphasize the fact that this is a very easy thing to get your business and your team into. A system does not need to feel extremely complex for it to be valuable. That in large part is the point of all the AI systems and this Cambrian explosion of intelligence that we're seeing across the board. It is very easy to set up many agents and many people in a shared environment for them to go and start creating together and building off one another. Now, what do I think about the drawbacks and the challenges of a platform like this? I think first, they are as good as you let them be, right? You want to go through the setup process we outlined in this video. So adding your team members, adding your external agents, setting up your different rooms and then connecting all of your apps. If you do those four things, you will get the most use out of this platform. And then as you use it, that advantage just compounds because that data, that context, that experience of you giving these agents feedback and all the agents learning from them, internal and external just compounds. Now, second, I think there is still a human element of reviewing things. You still want to be involved in that kind of a process. And I think as the model layer improves over time, you can just step back and really not have to review anything at all. For now, you still want to just keep an eye on things as the agents are doing the work on your behalf. And then third, this is a new kind of app. So what you typically see with something like this is the UIUX is going to evolve over time. What I would say is it's most important to have a shared workspace for your agents and for your business. And if you are not familiar with this world at all, just get agents set up and working for you. If you do those two things, yes, the UIUX, the setup and the models, for example, will change over time. But as long as you practice that skill of working with AI and you maintain, last time I'll say this for the video, but you maintain shared data, then all of it grows with you and you'll stay on the frontier for your business. And there we go. That is what I wanted to show you about Kylon. Now I have two thoughts generally on how to best use the platform. The first is that I think there is still a review process that should be there, that should be human driven in a lot of these workflows. You do want to just check the final work before things go live or deployed. And that's just kind of the evolution of the models themselves. As these things get more intelligent, you'll need to do less review over time until eventually it's fully hands-on. And then second and more importantly, the platform will be as good as you make it. That is the whole thesis of being AI native. If you put the effort into being AI native, you can get a lot of productivity juice from your squeeze there, right? But if you only do one workflow thing from occasionally on the platform, if half your team is on it and half isn't, it doesn't really work, right? There are positive marginal returns. The more you use it, the more of your team you have on it, then the more utility you get out of it. And so the more you give it, the more useful it will be back to you because it's intelligent. Intelligence needs to learn. It needs time, it needs experience. And so the more you give that brain, the more the brain can do for you. And there you go. That is Kylon. I think it's a pretty cool glimpse into the future of what working with agents and humans looks like. I hope you find it useful and out of mind. I appreciate you being here and I will see you in the next one.