Averra AI Privacy Policy
Effective: 11 October 2026
Last updated: 11 October 2026
Language: this Privacy Policy is also available in Indonesian. Both versions say the same thing; if they differ, the Indonesian version prevails.
1. Summary
Averra AI is a platform that answers questions about the data you connect and runs the flows you build.
To do that, the platform stores your account, your conversations, the files you upload, the files produced for you, and records of what the system runs on your behalf.
The platform does not sell your data, does not use it for advertising, and does not use the content of your conversations to train models.
2. Who we are
The service is operated by Developer Averra AI, located in South Jakarta, Indonesia.
Questions about this policy can be sent to contact@averra-ai.com.
3. Data we process
3.1 Account data
| Data | Source |
|---|---|
| Email address | You, when the account is created |
| Full name | You |
| Profile photo | You, if uploaded |
| A short description of you | You, if filled in |
| Account status and role | Set by us |
| Password | You, stored only as a BCrypt hash and never in its original form |
3.2 Sign-in records
Every sign-in attempt is recorded with the IP address, the browser identity, the outcome of the attempt, and the reason if it failed. These records are used for account security, not for behavioural analytics.
3.3 Conversations
The content of your questions, the assistant's answers, the ratings you give an answer, and the reasoning trace that accompanies an answer.
3.4 Files
Files you upload to a conversation are kept in object storage together with their original name, type, size, and a fingerprint of their content. Files are private by default and can only be retrieved by their owner or by an account that holds the media permission.
Files produced for you, such as spreadsheets and documents, are stored the same way, with a record of who owns them and which process produced them.
3.5 Execution records
Every agent run and every tool call is recorded: when it started, how long it took, which tool was called, which model was used, and how many tokens were used. These records are what make an answer traceable to its origin.
3.6 Usage analytics
Platform usage events and their attributes, used to understand workload and cost.
3.7 Data you connect
When a database is registered as a data source, the service reads its schema and runs read queries against it. The content of that database remains the property and responsibility of whoever registered it. We do not copy it, except to the extent that query results appear in answers, produced files, or execution records.
3.8 Your browser
If you use the browser in the service, its profile keeps what an ordinary browser keeps: cookies, sign-ins, site data, and history, on our browser server, separate for each profile and not shared with other users. Pages the assistant reads are sent to the model provider as part of the conversation, and screenshots only when the assistant asks for them. Passwords and codes you type while taking over are not read by the assistant. Every action you approve is recorded with its type, the page address without parameters, and the result, without the page content. Downloads stay in the browser until you save them or the browser is stopped. Pages that are read do not become records about you. Browser profiles are not included in backups.
3.9 Imported knowledge
Documents imported into the knowledge base are stored as text, together with their chunks and their vector representations, so that they can be searched.
4. What the data is used for
- running the service you ask for;
- keeping accounts secure and preventing abuse;
- tracing the origin of an answer;
- understanding the workload and cost of the service.
Beyond that, your data is not used for any other purpose.
5. Third parties
5.1 Language model providers
To answer questions, the content of your conversation is sent to the language model provider that is active in this service. The provider is DeepSeek or Google, both running outside our infrastructure, and the choice may change at any time. What is sent is your question, the relevant conversation history, and the pieces of data or documents used to compose the answer.
The embedding model that makes documents searchable runs on our own infrastructure, unless the embedding provider is set to Google.
5.2 Web search
When the assistant searches the internet, only the search keywords are sent to the search engine, not the content of your conversation.
5.3 Voice
Turning answers into speech runs on our own machines. The text of an answer is not sent to a third-party voice service.
5.4 Telegram
Connecting Telegram is optional. We store your Telegram user ID and private chat ID, your username if available, and the time of connection, to link messages to your account. Messages, voice messages, answers, notifications, and result files sent through the bot are also processed and stored by Telegram. Voice messages are turned into text by a transcription model; we do not keep the recording, only the text, which becomes a message in the conversation. The bot chat does not use end-to-end encryption. Answers may include notes or files of yours that are relevant to the conversation, according to your account's permissions.
You can disconnect in Settings or by sending /stop to the bot. Deleting your account also removes the connection. This does not delete copies already stored by Telegram. Records of received updates that have finished processing are deleted after one day; a connection code is valid for ten minutes and can be used only once.
5.5 What we do not do
We do not sell personal data, do not share it with advertisers, and do not use it to train anyone's models.
6. Security
- Passwords are stored as BCrypt hashes.
- Access to every interface is limited by per-role permissions.
- Private files can only be retrieved by their owner or a holder of the media permission, and every request is checked, including when the browser already holds a copy.
- Sessions can be revoked, and revoked tokens are refused until they expire.
No system is completely secure, so the security of data cannot be guaranteed absolutely.
7. How long data is kept
| Data | Current retention |
|---|---|
| Account | While the account is active |
| Conversations and messages | Until you delete them |
| Uploaded files | Until deleted |
| Voice clips | 14 days |
| Agent workspaces | 30 days |
| Flow run records | 90 days |
| Browser profile | Until you, or the deletion of your account, delete it |
| Browser action records | 180 days |
| Flow versions | 180 days |
| Sign-in records and execution records | As long as needed for security and audit |
Deleting an account, conversation, message, or file marks it as deleted so that it can no longer be accessed through the service. Backup copies that already contain it disappear on their own when that backup expires.
8. Control over your data
You can:
- view and change your profile data from the profile page;
- delete conversations, messages, and files you uploaded;
- ask for your account to be deleted through contact@averra-ai.com.
9. Children
The service is not intended for children under 13, and we do not knowingly collect their data.
10. Changes to this policy
This policy may change at any time. The date at the top of the document always shows the version in effect.
11. Contact us
Developer Averra AI
South Jakarta, Indonesia
Email: contact@averra-ai.com
Website: averra-ai.com