On premises agents (Kubernetes hosted)
It is possible to extend Kubiya with your agents. Using custom agents allows you to provision pre-defined environments with common tools and solutions such as scripts, repositories and binaries.
Kubiya can invoke the different tools and provide a conversational interface to them
Custom agents are running from your environment using local runners. You'll need to setup a local runner before being able to create and use custom agents.
Architecture
These Agents are based on simple Docker images. You can easily extend our base image by inheriting it as a base layer and extending it with your tools.
Once an Agent is created, users can invoke it using the
/agent
slash command or from within active conversationsKubiya will be able to provision or reuse a container to serve the profile on your own Kubernetes cluster for each Agent in a user-based way.
Self-hosted agents offer a flexible solution for organizations seeking to integrate Large Language Models (LLMs) into their infrastructure. Key points include:
Custom Hosting: Deploy agents on your own infrastructure, providing control over performance, security, and integration.
Bring Your Own LLM: Users have the option to integrate their own LLMs with these agents, allowing for personalized and specialized use cases.
Scalability and Customization: Tailor agent deployment and management according to specific organizational needs, leveraging Kubernetes for scalability and maintenance.
Enhanced Security and Compliance: Hosting agents in-house aligns with internal security protocols and compliance requirements.
This approach provides a versatile and secure platform for utilizing advanced language models within your unique ecosystem.
Last updated