Introducing OpenScience. A better, open-source Claude Science. • Any model: GLM, Kimi, DeepSeek, Claude, GPT, your own fine-tune. Switching is one flag. • 250+ research skills across ML, comp bio, ch
The Hook
“Claude Science just launched. One problem.”
Problem Agitation
“Anthropic decides who gets to do the science, and what science is safe.”
Problem Agitation
“(No spoken dialogue — frantic digital sound effects and bass swell)”
Product Reveal
“So we built OpenScience. A better, open source Claude Science.”
Feature Teaser
“Bring any model you want. No lock-in.”
Feature Teaser
“It just does the work. No forced degradations for 'safety.'”
Feature Teaser
“It reads the literature, queries 42 scientific databases, runs the code on real compute, and renders molecules, structures, and genomes in line.”
Feature Teaser
“Natively integrated with Atlas. Multiplayer. Autonomous. Fully reproducible.”
Feature Teaser
“Hundreds of research skills, built in. Specialist agents for biology, chemistry, physics, and machine learning.”
Call to Action
“Open source. It runs on your infrastructure.”
Call to Action
“Star it, fork it, build on it. npx synsci.”
Synthesizable in Remotion · key components of the challenger-workbench-reveal template.
Numeric value rapidly counts up from 43 to 292 while category tags expand.
Use Remotion interpolate with spring physics easing, floor the output, and display in an animated mono container.
Simultaneous live update of graphs, molecular renderers, and code terminal across a grid.
Layout with CSS Grid, driving frame-dependent SVG path morphs for loss curves and rotating Three.js canvas for 3D protein ribbon.
Bold red 'BLOCKED' badge scales down abruptly with screen shake and settles on target modal.
Spring scale from 2.5 to 1.0 with high damping, accompanied by 2-frame translated camera jitter.
Purpose. Identify frustration with existing AI safety restrictions.
Execution. Rejection dialog and social media outrage wall.
Purpose. Establish open-source, permissionless positioning.
Execution. Neon branding unveil and multi-model selector.
Purpose. Demonstrate tangible domain depth beyond generic LLMs.
Execution. Multi-pane dashboard showcasing bioinformatics, molecules, and compute.
Purpose. Drive immediate GitHub cloning and terminal installs.
Execution. CLI typing prompt and repository metrics.