Per-Vault AI
Configure AI behavior per vault. Choose which model powers chat, which enrichment features run, and how aggressive auto-tagging should be. Technical vaults can use a code-focused model while personal vaults use a conversational one.
How it works
Each vault has an AI configuration panel where you select the chat model, enrichment pipeline stages, auto-tagging behavior, and persona. Settings are inherited from parent vaults by default but can be customized at any level. You can also disable AI entirely for sensitive vaults where you want zero data processing.
Per-vault AI configuration panel with model and pipeline settings
Why it matters
One AI configuration does not fit all. A code repository vault benefits from a model strong at code understanding. A family memories vault benefits from a warm conversational model. Per-vault AI lets you optimize the AI experience for each context, getting better results across your entire knowledge base by matching the model to the content.
Two vault cards showing different AI model configurations