​

Core Concepts

​

Functions

Functions are the core building blocks of Inferable applications. They can be written in any supported programming language. They run on machines managed by users, and are registered with Inferable via the Client SDK

More details on Inferable Functions

​

Run

A Run represents the execution which Inferable undertakes to complete a task. It may include one or more calls to a large language model and the functions that have been registered.

More details on Inferable Runs

​

Cluster

A Cluster is a logical grouping of functions and the machines running them, as maintained by the Control Plane. These machines are managed on-premises by users, while Inferable maintains only the contextual information. Functions and machines can only belong to one Cluster.

More details on Inferable Clusters

​

Secondary Concepts

​

Control Plane

The Control Plane is the central management system of Inferable. It maintains the context of all registered functions and the machines running them. The Control Plane orchestrates function execution based on user prompts, LLM reasoning, and manages the overall health and performance.

More details on the Inferable Control Plane

​

Client SDK

The Client SDK is a software development library that allows you to register your functions with Inferable. It acts as a bridge between your codebase and the Inferable control plane, making your functions available for use within the system.

More details on the Inferable SDK

​

Machine

A machine is a physical or virtual machine managed by users, on which functions are executed. Inferable does not manage these machines directly but works with them through the Control Plane to execute functions. For example, if you’re running in a kubernetes environment, a machine is a kubernetes pod.

​

Prompt

A prompt is the input you provide to the Control Plane when interacting with Inferable. It is a text-based input that triggers the execution of a function or a sequence of functions within your Cluster. Prompts can be simple queries, instructions, or more complex requests that return structured data for complex workflows.