---
title: "Skild AI's robot model learns new tasks from a single video"
url: https://www.parallelquant.com/posts/skild-ai-s-robot-model-learns-new-tasks-from-a-single-video-f5ae01
source_name: "NVIDIA"
source_url: https://blogs.nvidia.com/blog/skild-ai-s1-physical-ai/
published: 2026-09-10T16:30:35.000Z
topics: ["robotics", "products"]
publisher: "Parallel Quant"
---

# Skild AI's robot model learns new tasks from a single video

*2026-09-10 · Source: [NVIDIA](https://blogs.nvidia.com/blog/skild-ai-s1-physical-ai/)*

Skild AI's S1 robot foundation model, built with Nvidia's Physical AI tools, can learn previously unseen, long-horizon tasks from a single video demonstration rather than extensive reprogramming. It targets manufacturing and warehouse settings where layouts and tasks change frequently.

**Why it matters:** One-shot task learning from video could meaningfully cut the integration cost that has kept general-purpose robots out of dynamic environments, where reprogramming for each new task has been the main barrier to adoption. It's part of a broader push toward foundation models for robotics, mirroring how LLMs generalized across language tasks.

**Topics:** robotics, products

---
Read the original: https://blogs.nvidia.com/blog/skild-ai-s1-physical-ai/
Canonical: https://www.parallelquant.com/posts/skild-ai-s-robot-model-learns-new-tasks-from-a-single-video-f5ae01
Published by Parallel Quant — https://www.parallelquant.com