SciencePal 2.0

SciencePal 2.0 is the next-generation AI research platform developed by HKAI-Sci and designed for scientific research workflows. Access it through your browser with no environment setup or coding required. Use natural language to complete end-to-end research tasks, from literature management to computational simulation. With cross-session long-term memory, a self-evolving skill system, domain-specific agent toolchains, and multi-agent orchestration, SciencePal 2.0 gives researchers a personalized AI research team that continuously evolves. Six Core Capabilities 1. Long-Term Memory A cross-session semantic memory system automatically stores research context, experimental parameters, and key findings, so you do not need to repeat background information. Semantic search creates a personal knowledge base that keeps evolving with your work. 2. Self-Evolving Skills Define experimental workflows, analytical methods, and domain knowledge using natural language or Markdown files, then turn them into persistent agent skills. Built-in modules include Zotero literature management, materials computation, and bioinformatics, with support for custom extensions. 3. Scientific Agent Team Use preconfigured agents for molecular biology, materials science, protein design, medical literature, and more. Each agent comes with a professional toolchain, including BLAST, AlphaFold, scanpy, Materials Project, AutoDock, and others. A single instruction can launch a complete analysis workflow. 4. Persistent Sandbox Every user has a persistent computing environment. Code, data, dependencies, and intermediate results remain available across sessions, allowing research to continue seamlessly. 5. Built-In Academic Tools Literature management through the Zotero API, citation graph navigation, and document parsing are integrated natively, with no external plugins or manual exports required. 6. Intelligent Model Routing SciencePal automatically selects the best model and reasoning depth for each query, significantly reducing cost while maintaining high-quality results.