Why AI Needs Human Oversight | #SoftwareDevelopment #DevOps #AICoding #AgenticAI #Developer #Shorts

Key Takeaways

The way people consume our platform is through a library of AI agents. We call it harness AI. It's like these are all interconnected agent. These are about like you know a dozen or so agents that are at the top layer like agents for devops tasks for s sur for testing for phobs for security for you know all these tasks and then there we have we have purpose-built agents that are for smaller task under them like say if you go to our AI harness AI and say create me a you know a deployment pipeline for my application and you know it starts with okay what is your application what does deployment pipeline for your company means what are your security standards what are your compliance standards, what does your production infrastructure look like? So we have to have all of that you know first of all we have to have like you know the small agents for each of those tasks like you know the to create a deployment pipeline we need to do 30 different tasks to do that. So we have like small agents for each one of those that we will farm out the task to. So these are purpose-built for those but they also use your knowledge and your context what we create like a SDLC knowledge graph that we create for the customer which is like you know what does your infrastructure look like what services depend on what other services what environments you run on what tools you have for security testing what's your you know SLAs around different things. So we have all of that information. So now these agents are using that information for you know a smaller context for that particular task because you can't give like a big context to like one big agent like you know you have to break it into smaller agents and give them the right context at the right time. So that's how kind of our AI works. A lot of agents who are purpose-built for these delivery tasks and then they use the right knowledge graph and the right context from a engineering organization's information to do the right things.