Application privacy policy
Last updated 27 October 2026
This policy explains what Temiro does with personal data in the application (app.temiro.ai) and in the connection through which outside AI apps read from Temiro. We wrote it against what the product does rather than from a template, so it names the data we hold and the companies that process it on our behalf.
The operator of the service is TEMIRO LABS S.R.L. (Trade Register no. J2026056434006, fiscal code 55658691). You can reach us at hello@temiro.ai about anything in this document.
Who decides what
Temiro is used by companies. The company that opens a workspace decides which documents and sources it connects and who sees them: it is the controller of that content, and we process it on its behalf under the data processing agreement (DPA) we sign with it. For your account (email address, name, settings) and for how we run and secure the service, the controller is us.
In practice, your company’s owner and admins can see what is shared with the whole company or with teams, can remove members and can delete company data. Ask your administrator about internal rules; this document describes what we do.
What we process
Your account and the company workspace
- Your email address and sign-in credentials, held by our authentication provider. We never see or store your password ourselves.
- Your name, profile picture and interface preferences.
- Your company, your role in it and your teams.
- Invitations sent and received, with the invited address.
Content in the application
- Documents. We extract each document’s text, split it into passages and keep the passages with the numeric representation used for search. We do not keep the original file, with two exceptions: uploaded spreadsheets, when your company has switched on querying them, so the assistant can calculate on the cells; and the files the assistant generates for you.
- Chat attachments. We keep the extracted text and the file name, not the file.
- Conversations. Your messages, the assistant’s replies, which tools it used, the sources it cited and any feedback you leave on an answer.
- Assistant memory. Facts the assistant keeps to be useful from one conversation to the next. You can see them, delete them or switch memory off in Settings.
- Wiki pages, projects, scheduled tasks and automations created by you, your colleagues or the assistant.
- Content from connected services: see Connected services below.
Operational records
- Usage: the model used, token counts and the cost of each request, so an administrator can see spend.
- The audit log: who did what and when, for actions that matter for security (deletions, permission changes, connections), with the IP address the action came from. If your company switches on the extended log, each sign-in is recorded too, with the method, the browser and the IP address; the extended log is your company’s decision, and your company tells you about it.
- Safety events: the fact that a request was stopped by the protection rules, without its text.
- Email sends: to whom, when and what kind of message, without its content.
- Server logs, which can contain the IP address and request details.
Credentials
Access to the services you connect, and any API keys your company supplies, are stored encrypted with a key held outside the database. They are decrypted only for the call you asked for, never reach the browser and go nowhere but to the provider they belong to.
Why, and on what basis
| Purpose | Legal basis (GDPR Art. 6) |
|---|---|
| Providing the service to you and your company | Performance of a contract |
| Answering from your company’s knowledge | Performance of a contract (for content, on the company’s behalf, as processor) |
| Security, abuse prevention, the audit log | Legitimate interests: network and information security (GDPR Recital 49), that is detecting unauthorised access and abuse and reconstructing an incident |
| Usage metering and running the service | Legitimate interests: showing your company what it uses and billing that use correctly |
| Account and service emails (invitations, resets, alerts) | Performance of a contract |
| Legal obligations (accounting, requests from authorities) | Legal obligation |
Your email address is required for an account: without it you cannot sign in. The rest of the data you provide by using the service.
We do not sell personal data, we do not use it for advertising and we do not train our own models on your content. We make no decisions based solely on automated processing that produce legal effects concerning you or similarly significantly affect you.
Who else processes it for us
Each provider below receives only what its part needs. The AI model providers work, by default, on Temiro’s accounts. The last column says, for each, whether data leaves the European Economic Area and on what basis.
| Provider | What it does | Where | Transfer and basis |
|---|---|---|---|
| Supabase | Database, sign-in and file storage | Ireland (EU) | Data stays in the EU; the provider is based in Singapore: Standard Contractual Clauses (Art. 46(2)(c)) |
| Railway | Hosts the application’s server | Netherlands (EU) | Data stays in the EU; the provider is based in the US: EU–US Data Privacy Framework (Art. 45) |
| Vercel | Hosts the application’s interface | Ireland (EU) | The provider is based in the US: EU–US Data Privacy Framework (Art. 45) |
| Anthropic | The Claude models, which write the answers. They receive the question, the passages found and the conversation history. They also write conversation titles, safety checks and wiki pages, read scanned documents, search the web when asked, and provide the environment in which the assistant runs code | United States | Standard Contractual Clauses (Art. 46(2)(c)) |
| MongoDB (Voyage AI) | Turns text into numeric representations for search and reorders the results | EU | Content is processed in the EU; Temiro’s account data with the provider: EU–US Data Privacy Framework (Art. 45) |
| OpenAI | The GPT models, only if your company has switched them on | United States | Standard Contractual Clauses (Art. 46(2)(c)) |
| Resend | Sends account and service emails | United States | EU–US Data Privacy Framework (Art. 45) and Standard Contractual Clauses |
The EU–US Data Privacy Framework is a European Commission adequacy decision for US companies that have certified under it; they are listed at dataprivacyframework.gov. The Standard Contractual Clauses are those of Commission Decision (EU) 2021/914, published on eur-lex.europa.eu. You can ask us for a copy of the clauses signed with each provider at hello@temiro.ai.
If your company uses its own API keys, requests go to that provider under your company’s agreement with it, not ours.
Model providers process a request to answer it. Anthropic and OpenAI do not train models on data sent through their API, and the setting that would let data be used for training is off in our MongoDB account. Anthropic and OpenAI keep requests for a limited time, usually up to 30 days, to prevent abuse, under their own policies.
Connected services
All of them are optional. Nothing is read until someone in your company connects the service, and each one reads only what it is pointed at.
- Google. Drive (the folders chosen), Gmail (reading and drafts) and Calendar. For folders that follow Drive’s sharing, we read who has access to the folder.
- Microsoft 365. OneDrive and SharePoint (the folders chosen), Outlook (email: reading and drafts; calendar) and, if your company switches it on, the Teams channels an administrator connects. For folders that follow Microsoft 365’s sharing, we read who has access to the folder, group members included.
- GitHub, Notion, Slack, Jira, Confluence, HubSpot, Salesforce and public websites. The content pointed at enters your company’s knowledge base like an uploaded document. From HubSpot and Salesforce we do not read contacts’ email addresses or phone numbers.
- Tools in a conversation (Canva, Atlassian, HubSpot, Salesforce): each person connects their own account, and what the assistant asks for goes to that service.
- Tools, MCP servers and webhooks your company defines send data to addresses your company chooses; your company is responsible for them.
- Public registers. When it checks a company, the assistant sends the tax code to ANAF and the company name to the Romanian courts portal (portal.just.ro). We remove the names of private individuals from the results before they reach the model.
- Outside AI apps. If your company allows it, a member can connect Temiro to Claude, ChatGPT or Cursor; what that app reads from Temiro then reaches its provider (Anthropic, OpenAI or Anysphere, in the US), under the member’s account and that account’s terms. On a company account the provider usually works for the company; on a personal account it processes the data on its own responsibility, under its own rules. That is why we recommend company accounts. The connection only reads.
The assistant never sends an email. Any file, draft or event it creates, changes or deletes in a connected service happens only after a person presses a button or, in automations, only where your company explicitly allowed it.
You can disconnect a service at any time; syncing stops. What was already indexed stays until you delete it.
Where the data is
The company’s data is kept in the European Union: the database in Ireland, the server in the Netherlands, search in the EU. Processing with the Claude and GPT models and sending email happen in the United States, on the bases in the table above.
How long we keep it
- Content (documents, conversations, memory, generated files): for as long as the company has a contract with us, or until someone in the company deletes it. It does not expire on its own. Deleting a document deletes its passages. When the contract ends, the company’s data is deleted within 30 days, or returned to it if it asks.
- Your account: until you delete it or your company removes you.
- The audit log, with IP addresses: 365 days, then deleted automatically. Sign-in history, where your company switched it on: 30 days. If you delete your account earlier, the records stay without your name.
- Usage records: for as long as the company exists, anonymised if you delete your account.
Your rights
Under the GDPR you can ask for access to your data, its correction, its erasure, a portable copy or restricted processing, or object to processing based on legitimate interests. Write to hello@temiro.ai; we answer within one month. For your company’s content the request goes first to your company, which is its controller; we help it answer.
Some of this you can do yourself in the application:
- Erasure. Settings has two actions: delete everything you created and keep your account, or delete your data and your account. Both are immediate and final. Folders and sources you connected for the whole company or for a team are not deleted: they pass to the company’s owner, because they are its knowledge. If you are the only owner of a company with other members, transfer ownership first. On erasure we also revoke the access given to your Google account; access to Microsoft you can revoke at myapps.microsoft.com.
- A copy of your data. Settings downloads your data as a JSON file, and each conversation can be exported as Markdown.
- Memory and notifications. You can see and switch them off in Settings.
The right to object. You can object at any time, on grounds relating to your situation, to the processing we do on the basis of legitimate interests (security, the audit log, usage metering). Write to hello@temiro.ai; we stop unless we have compelling legitimate grounds that override yours or need the data to establish, exercise or defend legal claims.
If your company removes you from its workspace, your Temiro account is deleted, because an account belongs to a single company.
If you believe we have handled your data wrongly, you can complain to the National Supervisory Authority for Personal Data Processing (ANSPDCP) in Romania.
Security
- Every request is authenticated and limited to one company; access to another company’s data is refused, and the database carries its own row-level rules as a second barrier.
- Access to connected services and API keys are encrypted, as above.
- Traffic is encrypted in transit.
- Where it is switched on for your company, two-step sign-in, which your company can make mandatory.
- Important actions are recorded in your company’s audit log.
No system is perfectly secure. If you find a vulnerability, write to hello@temiro.ai before disclosing it and we will fix it together.
Cookies and browser storage
The application uses only what it needs to work, so it does not ask you to accept cookies:
- session cookies, which keep you signed in;
- your language, theme and sidebar state, kept for a year so you don’t choose them again;
- two cookies lasting a few minutes, when you sign in with Google or Microsoft and when you connect a service, which tie the return to your request;
- in the browser’s local storage: the active company, the assistant you chose and how you grouped the knowledge graph.
There is no traffic analytics, advertising or third-party tracking in the application.
Children
Temiro is a product for companies and is not directed at anyone under 16. We do not knowingly collect their data.
Changes
When we change this policy we update the date at the top. Before a material change takes effect, we tell those affected by email.
Contact
Questions, requests or complaints: hello@temiro.ai.