Privacy-safe AI: what your data does (and doesn't do) inside VED Saarthi

When you type a question into an AI tool, it's reasonable to want to know where that question goes, who reads it, and how long it sticks around. Most policies bury this in 4,000 words of legalese. We won't.

What we do with your messages

Your chats are stored in our database in Mumbai (India region). They are attached to your account and only visible to you. We send the message plus the relevant context (the chat history) to the model you picked (GPT-5.5 or Claude Sonnet 4.6) over an encrypted connection. The model generates a reply and we stream it back to you.

What we don't do

• We don't train models on your data. Neither do our underlying providers — OpenAI's and Anthropic's enterprise terms (which is what we use) explicitly forbid training on customer messages.
• We don't sell your data. Ever. To anyone.
• We don't read your chats. The only humans who can see your messages are you. Engineers occasionally see anonymised error logs (e.g. “a 500 happened at this endpoint”) — never the content.

Voice and image uploads

Voice clips are transcribed once, the text is stored, the audio file is deleted within 24 hours. Images you upload are stored in encrypted object storage for as long as the chat exists. Delete the chat and the image goes too.

Delete everything any time

Settings → Account → Delete account wipes every message, every project, every pinned diagram, every voice transcript and every image — permanently, within 7 days. We don't keep a “just in case” copy.

India DPDP compliance

We're an Indian operator, our servers are in India, and we operate under the Digital Personal Data Protection Act, 2023 (DPDP). If you ever have a privacy concern, our grievance officer is named in our Legal page and replies within 7 days as required by law.

Read the full policy at /privacy — or email us at info@vediautodevops.com with anything we haven't answered.