Actual AI Architecture Agent Demo

Speaker 0: I'm founder. I'm John Kennedy, founder of Speaker 1: Actual CLI. We do have a business event, but we also have a company. And that company builds an architecture agent. And we've had that architecture agent, essentially behind a wait list, for, since January 27 now. But for the first time and I know some of you here Solution the wait list, we're gonna email you out today. Speaker 1: But for the first time today, we're opening up sign up to our architecture agent that helps you if you're building with Broadcode or Cursor or One or any of the coding agents. It helps you manage all of your architecture rules. And very briefly, we're gonna show you a demo of it just for fun. So thank you for enduring our sponsored demo. Austin here, who's our lead AI engineer, is gonna show you very Real, what you can do, with our architecture agent. Speaker 0: Hello. Hi, everyone. I'm Austin, lead AI engineer here at Acto Impact. In just a few moments, we're gonna be showing you a quick demo of how the Acto AI product works. So the first thing we're just going to show, before we get into that, I'm just going to explain a bit about what's, problems we solve. Speaker 0: So we build guardrails for AI agents, and we are currently focused on architecture and curricular because of the scaling problem that we encountered with agents trying to solve, large scale features for for soft regions. What we have found is that, with this concept that we call architecture design records or decision records. It's not a new concept, but it's something that we are bringing to the HNI space as a core principle for HNI development for software products. And so we'll show you how you can use ADRs within our actual AIAN application. So the first thing that you do is just sign up, with your, account, either GitHub or Google or you sign up with email. Speaker 0: It's a very quick sign in process. It takes less than a minute. You just create your organization, choose whatever you would like. As you go through the steps, you will see that we connect to GitHub so that, you can connect your ADRs directly to GitHub PRs, so that when you create an ADR, you can see APIIs. To our Solution, and what we will do behind the scenes is analyze your codebase. Speaker 0: So we have a history that we analyze, we analyze every commit, every author, every, comments, and PR that goes into your codebase. And with that, we analyze it and come up with a set of what we call ADRs that are connected to your codebase. And so after a certain amount of time of analyzing and focus, you will see a PR that shows up in your repo called rule file sync. And what this will include is a list of, records that are associated with your code base and help you, keep those guardrails around your agents. So these will be records that explain specific architectural decisions such as using a particular framework or using a particular format or API endpoints, security compliance measures they have to keep in mind. Speaker 0: And all these things that your agents, may may have to refer to your code base if you don't have this documentation explicitly laid out. So here's an example of an consequences, and risks of not following this rule, and then finally, end of course. And so with all this together, as policy become more capable and handle harder contrast, it becomes pertinent to yield and provide these records so that they can produce, you know, produce the most, relevance results for the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the cost of the confirm that every protein that goes into your protein is consistent with that EDR set. So we can see we have an example of a bad graph here. It has a few, variations that that cross some of the AR response repo. Speaker 0: You can see we have an analyzer that will analyze it, produce consistency issues for you, and get your agency to actually review it and improve it. And so here you see just a few specifics. And then, once you have resolved those issues, we have a follow-up with consistency check that shows that all those ADRs passed. And here I provided a quick summary of what this agent did to make these ADRs, passed. And so first round, it has 2 that failed, hits and fixes. Speaker 0: At second round, there's some others failed, which were more severe. Fix those. And then finally, you have the third round where your ADRs plain and consistent with your architecture records. And so putting all this together, our platform gives your agents the set of needs to produce effective architecture for your software systems, and it gives you the opportunity to provide a structured input for the agents to be able to manage, your software more effectively. Ask you to chat more about our application, AppFlow US, but please, if you're interested, the application is open for signage. Speaker 0: So you're welcome to go to appflow.ai, and you can sign up right here, register, or get started. Thank you. Thanks, Austin. Speaker 1: Alright. We're gonna now have our first demo.