Prototype ML with Visual Blocks
summary
Visual Blocks is a no-code ML prototyping tool (launched at I/O) that claims users can build an ML app in 20 minutes using a 60 node library of inputs, processors, models and visualizers. It wires camera/text/audio inputs to models like PaLM and MediaPipe, runs 475 face landmarks for AR stickers and body/hand segmentation, uses a Shader Renderer and Image Mixer to produce working AR and “laser hands” effects, and composes pipelines for summarization, multi‑tone email drafts, and a Werewolf game with six PaLM agents that reason and vote.
AI-written summary · about
chapters (7)
- 0:00Introduction to vblocks
- 2:58AR vision prototyping
- 9:08Laser hands and shaders
- 12:53Text summarization with LLMs
- 16:49Email generation and review
- 22:10Werewolf reasoning game
- 25:15Building custom pipelines
related talks
23mWhy MLX
22mVue + TensorFlow.js: Building Your First ML Powered App
44mArtificial Intelligence and Machine Learning in the browser
32mAdvanced on-device ML made easy
17mMallard: Turn the Web into a Contextualized Prototyping Environment for Machine Learning
47mBuilding smarter apps with Machine Learning, from magic to reality
19mML at Client Side
40mKeynote: Open-source Multimodal AI in the Wild