CLAP: Direct VLM-to-VLA Adaptation via Language-Action Grounding | Startup Networx | Startup Networx