Skild AI has released S1, a robotics foundation model capable of learning novel tasks from a single video demonstration without requiring fine-tuning. The model can execute commands with 10-minute planning horizons.
S1 represents a shift in robotic learning architecture, applying principles from language models to physical task execution. The system learns from minimal input—a single video prompt—and generalizes to unseen tasks without post-training adjustments.
The model's capabilities address a long-standing challenge in robotics: reducing the data and computational overhead required for task adaptation. Traditional approaches demand extensive datasets and retraining cycles. S1's architecture eliminates this friction.
The 10-minute planning horizon indicates the model can reason about multi-step sequences, a requirement for complex manipulation tasks. By leveraging foundation model principles proven in NLP, Skild AI extends the paradigm to robotic control and perception.
The development signals growing convergence between language modeling and embodied AI, where pre-trained representations enable rapid task transfer. This approach could accelerate deployment of general-purpose robots across varied environments and applications.
Anthropic has enabled shared memory across Claude chat and Cowork, allowing the AI to retain context about projects and preferences without repeated briefing.
A new tool called Claudette lets users remove casual, clickbait-style language from Claude's responses. The open-source project addresses complaints about AI-generated text mimicking viral content formats.
Apple's latest desktop computers are built to support local artificial intelligence work, addressing the growing trend of developers using multiple Macs for AI tasks.