Before you enable file capture for AI memory
Check attachments, sharing permissions, destination access, and retention before allowing files to be copied into an AI memory service.
Choose what gets shared 2 min read
Where does AI memory data go?
The path depends on the product. In Dray, chosen conversations are kept in a local browser archive and sent for server-side memory processing. Connected assistants may then retrieve context. Local storage does not mean all processing stays on your device.
Follow the content through each service
Start with the original AI provider, then identify the capture tool, memory backend, processing providers, and any destination assistant. Ask what each one receives and why. A single label like “private memory” does not explain that path.
Dray’s data notes say external model services through OpenRouter process conversation content to extract durable facts and prepare memory for search. The server stores memory records and supporting references. File capture, when enabled, adds files to Dray’s storage.
Ask about the deployed policy
A product may support several processing policies. What matters for your data is the active deployment and endpoint, not a capability that exists somewhere in the code. Confirm retention and access terms before connecting sensitive work.
Keep website analytics separate from conversation processing in your review. They involve different data and controls. An analytics opt-out does not mean the product stops processing the conversation content you choose to capture.
TAKE THIS WITH YOU
Data-flow questions
An illustrative template. Replace the details with material you are allowed to share.
What is stored locally? What conversation content reaches the server? Which model services process it? What is the active retention policy? Are files included, and can I leave them out? What can a connected destination retrieve? How do removal and access revocation work? Who can confirm these answers for this deployment?
USING THIS WITH DRAY
Keep the context. Check the source.
Dray does not claim that all processing is on-device, always zero-retention, or end-to-end encrypted. Check the data notes or contact support before connecting sensitive conversations.
Dray is in beta. Available sources and connections depend on your setup. Read the data notes before granting access.
Sources & next reads
Product details are grounded in the guides below. Example projects and prompts are illustrative, not customer results.