Privacy Policy
How CatalystLab collects, processes, and protects agricultural research hypotheses, key integrations, and literature synthesis datasets.
1. Research Data Collection Boundaries
CatalystLab is built exclusively for sustainable agricultural and food engineering research. Unlike typical general-purpose applications, we enforce strict containment parameters over what we harvest:
- Hypothesis Variables: Custom components, slider ranges, constants, and variables modeled inside the Thought Collider are stored in secure session caches and your synchronized database layout to withstand connection loss.
- Indexing Literature: Metadata, research abstracts, and citations harvested recursively from our 17 parallel academic index providers (including OpenAlex, PubMed, and Semantic Scholar) are stored with query filters, enabling immediate deduplication.
- Personal Credentials: When using custom or Bring-Your-Own (BYO) API keys for LLM synthesis, these secrets remain in the isolated secure backend. We never transfer client credentials directly to third-party endpoints.
2. Zero LLM Training Guarantee
We understand the sensitive proprietary nature of innovative chemical compositions, genetic crop modifications, or food-pipeline engineering hypotheses.
Zero Leak Policy: All telemetry logs and synthesis requests processed by Google Gemini (1.5 Pro and Flash) bypass general foundational training caches. Your variables, agricultural inputs, and synthesized reports are isolated and will never be used to train global generative AI structures.
3. Parallel Index Fan-Out Rules
To fetch references fast, the Orchestrator runs parallel API fetches to 17 concurrent servers. During this reaction:
- No personal information (e.g., your email, account name, identifier details) is exposed to external citation servers.
- Only sanitized scientific search strings derived through AI automated keywords mapping are transmitted.
- External APIs (such as Europe PMC, DOAJ, arXiv) only detect the secure server proxy address of CatalystLab.
4. Storage, Retention, and Deletion
All stored drafts, academic blogs, and research templates are persisted utilizing cloud architecture managed with Firebase Firestore.
You can permanently delete any saved research results, drafts, or account profiles at any time through the Settings Panel within the application workspace. Deleted datasets are immediately pruned from current active nodes and permanently overwritten within 30 days.