Privacy Policy
Last updated: June 2026
This policy explains what data InternalNexa processes and the third-party providers involved. It is provided to help you make informed choices; please review it alongside your organization's agreements.
Meaning-based (semantic) search
When meaning-based search is enabled, InternalNexa sends your search queries and eligible extracted text from your uploaded documents to our search provider Cohere to generate embeddings (numeric representations used to find relevant results) and to perform reranking of candidate results. This processing happens on Cohere's external cloud service.
We never send secrets, passwords, API keys, credentials, hidden system prompts, or content excluded by our security filters. Scanned documents without readable text are not indexed. Access is always restricted to your own account and projects — retrieval is authorized server-side.
You can choose Exact search to stay keyword-only, and administrators can disable meaning-based search and indexing at any time. See Cohere's privacy practices at cohere.com/privacy.
Data we process
Account details, content you create or upload (conversations, documents, projects), usage and billing records, and operational logs used to run and secure the service.
Third-party processors
To provide AI features we use model and infrastructure providers, including OpenAI, Anthropic and Google (text), Perplexity (opt-in web research), Cohere (meaning-based search embeddings and reranking), and Stripe (payments). Each receives only the data needed for its function.
Retention and deletion
You can delete your files and conversations. Deleting a document also removes its search index entries, including any semantic vectors. Backups are rotated on a limited schedule.
Security
Data is access-controlled per user and project. Provider credentials are stored server-side and are never exposed to the browser.
Contact
Questions about this policy: support@internalnexa.com.
