Cortex: Physical AI Agent Skills

WARNING

Cortex is not yet released.

This module is under active development and will be available in a future Telekinesis SDK release.

INFO

Cortex is a module in the Telekinesis SDK for building Physical AI agents.

It allows developers to create action graphs that combine primitive skills from other modules, enabling robots to perform complex behaviors. Cortex integrates large language models (LLMs) and vision-language models (VLMs) to generate context-aware motions and decision-making pipelines.

When to Use Cortex?

Use Cortex for robotics applications that require coordinated, high-level behaviors, such as:

• Automating pick-and-place sequences by combining perception, planning, and control
• Programming assembly tasks using action graphs
• Orchestrating multi-step operations for Physical AI agents
• Leveraging LLMs or VLMs to generate context-aware robot motions
• Experimenting with autonomous agents in simulation or real-world environments

What Does Cortex Provide?

Cortex includes a collection of modular skills for:

• Defining and executing action graphs
• Composing primitives from Telekinesis modules (Vitreous, Retina, Cornea, Pupil, Neuroplan, Medulla)
• Integrating LLMs and VLMs for decision-making and task generation
• Managing high-level agent behaviors and execution pipelines
• Simulation and real-world deployment support