Known Public Alpha Limitations

Review the current product, platform, AI, provider, offline, and MCP boundaries.

Public Alpha is free while NeuralScio is tested and stabilized. Some surfaces are limited or experimental.

Current boundaries

  • Managed cloud AI and paid billing are not active for public alpha access.
  • Local models and BYOK depend on the selected backend, model, credentials, and machine.
  • Provider availability, metadata quality, PDF access, caching, and AI policy differ by source.
  • Offline continuity is strongest for saved metadata, cached content, local annotations, and configured local AI. A first provider search still needs network access.
  • macOS has a documented build path but is not presented as a parity-tested Public Alpha download artifact.
  • MCP is a supported integration surface, but available tools and resources depend on runtime configuration and permissions.

What this means for evaluation

Check the feature status badge on each article. Experimental features need review against the source paper and the current build. Do not infer availability from a class name or an old engineering report.

See Troubleshooting, Backup, Restore & Data Locations, and MCP Security for operational limits.