Privacy Policy

1. Who we are and what this policy covers

This policy explains how TREFUR PTE. LTD. (UEN 202615732M) (“Trefur”, “we”, “us”) processes personal data in connection with the trefur.com website, the application at app.trefur.com, our APIs, SDKs, command-line tools, collector software and connectors (the “Services”). Capitalised terms not defined here have the meaning given in our Terms of Service.

Two roles. For account, billing, website and support data, Trefur decides how the data is used and acts as a “controller”. For Customer Data (the telemetry your agents send and the data we read from apps you connect), we process personal data on behalf of our customer, who is the controller; we act as its “processor” or “service provider” and follow its instructions, including under any data processing addendum (“DPA”) we have signed with it. If your employer uses Trefur and you have questions about data it sends us, please contact your employer first.

2. What we collect

2.1 Account and contact data

Name, work email address, organisation and workspace name, role, password (stored only as a hash) or the identity returned by the sign-in provider you choose (Google, GitHub or Microsoft, where offered), multi-factor authentication settings, invitations, and the messages you send us through forms or email.

2.2 Usage, device and security data

IP address, browser and device information, timestamps, pages and features used, sign-in events and approximate location derived from the IP address, audit-log entries of actions taken in your workspace, and diagnostic logs and error reports. We use this to run, secure and support the Services.

Product analytics. While you are signed in, we record which Trefur features you use (for example, that you connected Google Workspace or opened a report), linked to your workspace and an internal user ID, to improve the product. We do this ourselves, with no third-party analytics service or advertising cookies. These records contain no IP address, page content or free text, and we delete them after 13 months.

2.3 Telemetry from SDKs, the collector and APIs

When your agents use our SDKs, our collector, an OpenTelemetry exporter or our APIs, they send telemetry such as traces, spans, tool and model calls, timings, costs, errors and related metadata. What is included is controlled by your configuration:

• Redacted on your side by default. Unless you choose otherwise, our SDKs send prompt and response content only after replacing common sensitive values with placeholders: email addresses, phone numbers, payment card numbers, bank account numbers (IBAN), some national identifiers, API keys, access tokens and other credentials. Sending unredacted content requires an explicit “full” capture setting. Our collector also redacts by default, and content passing through its model proxy is replaced by a one-way digest unless your workspace setting allows more.
• Workspace privacy setting. Each workspace has a content privacy setting, which starts at “redacted” for new workspaces and can be tightened to metadata only. The SDKs follow whichever of their local setting and the workspace setting is stricter, and we apply the workspace setting again on our servers when telemetry arrives. By default, full prompt messages and system prompts are stored only as a one-way hash.
• Redaction is pattern-based and cannot recognise every kind of personal data, such as names in free text. You decide what your agents send, and you should not send data you are not permitted to share with us.

2.4 Connected-app discovery data

If an administrator connects a third-party system (for example Google Workspace, Microsoft 365, Slack, GitHub or a cloud account), we read activity and inventory data from it to show which AI apps, agents, other third-party apps and integrations are in use and what each one can access. This can include app names, granted permissions, event types and times, and the accounts involved.

• People are pseudonymised at rest. In connected-app activity and discovery records, the people involved are replaced, before storage, by a per-workspace pseudonymous reference, with the underlying identity held encrypted. The application shows counts rather than names by default, and network addresses of the people involved are dropped. This is pseudonymisation, not anonymisation: the identity can be recovered by the controls below.
• Audited reveal. A workspace owner or admin can reveal the identities behind one app, agent or department at a time, giving a reason. Access is short-lived, API keys cannot do it, and every reveal and its reason are recorded in the workspace audit log before anything is shown.
• Departments. From your Google directory we read each user’s department or top-level organisational unit. We keep it only for people who use an AI app, attached to their pseudonymised record, and remove it when they stop or are erased; for everyone else we only count how many people each department has. Departments of fewer than 10 people are never stored, a department appears only for groups of at least 10, and no department figure about fewer than 5 people is ever shown.
• Some features that you turn on separately, such as access reviews of your directory, may store directory attributes such as email addresses so that you can act on the results.

2.5 Billing data

Paid plans are billed through Stripe. Stripe collects and holds your payment card details; we do not receive or store full card numbers. We receive and keep the billing email, plan, subscription status, invoices and the identifiers Stripe gives us.

2.6 Cookies and similar technologies

The application uses only cookies needed to run it: a secure, HTTP-only session cookie that keeps you signed in, a cookie that remembers a trusted device after multi-factor sign-in, preference cookies for your language and time zone, and two non-credential cookies that route you to the right page after sign-in, each lasting up to 7 days. The application also stores interface preferences in your browser's local storage. We do not use advertising cookies.

On trefur.com we measure visits with Plausible Analytics, which does not use cookies, and we set a language preference cookie. Both trefur.com and the application load fonts from Google Fonts and Fontshare, which receive your IP address when fonts are fetched. Sign-up and some sign-in steps use Cloudflare Turnstile to prevent abuse, which processes your IP address and browser signals.

3. How we use personal data and our legal bases

What we do not do. We do not sell personal data. We do not share it for cross-context behavioural advertising. We do not use Customer Data to train generalised AI models.

Aggregated data contribution. Aggregated, de-identified statistics about how tools and models behave may be contributed to shared product intelligence. Prompts, outputs, credentials, personal data and workspace identifiers are never contributed. The default for each plan and the one-switch opt-out are described on the Data Contribution Terms page.

AI-powered features. Some features answer questions you type about your workspace data or evaluate agent outputs, using a large language model. Depending on the feature, the model is either one you connect with your own provider key (your agreement with that provider then applies) or one we engage as a sub-processor, listed in Section 5. When you use such a feature, your question and the relevant records from your workspace, after your workspace privacy setting has been applied, are sent to the model to produce the answer, and for no other purpose. Optional evaluation features are off until an administrator turns them on.

4. Google user data

Trefur's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

4.1 What we read and why

When a Google Workspace administrator connects Google Workspace to Trefur, we ask Google for the following read-only access. Google shows the exact list on its consent screen before access is granted.

If an administrator also connects Google from the discovery page to extend discovery, we additionally request read-only access to Google Chat spaces and memberships (chat.spaces.readonly, chat.memberships.readonly) and Apps Script process activity (script.processes), and access to the third-party apps each user has granted access to (admin.directory.user.security). Google describes that last permission as able to manage those grants; Trefur uses it only to list them and never changes or revokes anything. We use this data to find AI tools and other third-party apps connected to the workspace, and to show what each one can access.

Where Trefur offers it, an administrator may also connect Google so that your own agents can use Google Drive, Docs, Sheets, Gmail, Calendar or Contacts as tools. Google then shows the additional scopes on its consent screen. Trefur uses that access only to carry out the actions your agents request, on your instruction.

If you sign in to Trefur with Google, we receive only your basic profile and email address (openid email profile) to sign you in.

4.2 How we use and protect it

• We use Google user data only to provide and improve the user-facing discovery and observability features described above for the workspace that connected it.
• We do not use Google user data for advertising, we do not sell it, and we do not use it to train or improve generalised or non-personalised AI or machine-learning models.
• We do not transfer Google user data to others except as needed to provide those features (including to the service providers listed in Section 5, under confidentiality obligations), to comply with law, for security, or as part of a merger or acquisition with notice to you.
• Our staff do not read Google user data unless you ask us to (for example for support), it is needed for security or to comply with law, or it has been aggregated and de-identified.
• People named in Google activity and discovery data are stored as pseudonymous references, as described in Section 2.4. OAuth tokens are encrypted with AES-256-GCM before they are stored.
• An administrator can disconnect Google Workspace in Trefur at any time, and can also remove Trefur's access in the Google Admin console. Data already collected is kept and deleted under Section 6, or sooner on request.

5. Who we share personal data with

We share personal data only with the service providers (“sub-processors”) that help us run the Services, under contracts that limit their use of it; with your organisation's administrators, as part of the Services; with others you direct us to (for example an export destination you configure); where the law requires; to protect rights and safety; and with a successor in a merger, acquisition or sale of assets, with notice to you.

We will update this list before we add or replace a sub-processor that processes Customer Data. Customers with a signed DPA receive the notice described in it.

6. How long we keep data

• Telemetry and connected-app activity is kept for your plan's retention period, measured from when it was recorded: Free 14 days, Pro 30 days, Team 90 days, Mid-Market 180 days, Enterprise 365 days, or the period in your order form. A daily job removes data once its period ends; removed data is hidden immediately and permanently deleted after a further grace period of up to 7 days. If you move to a plan with a shorter period, data older than the new period is deleted.
• Connected-app inventory (the current list of apps, agents and integrations found) is kept while your workspace exists and is deleted with it, or sooner on request.
• Security, policy and audit records about actions in your workspace may be kept for longer, up to the period needed for security, compliance and dispute purposes, and in de-identified form where possible.
• Account data is kept while your account is open. When a workspace is deleted, its data is deleted from our production database, except for a minimal record that the workspace existed and was deleted, audit records we must keep, and de-identified aggregate records.
• Billing records are kept for as long as tax and accounting law requires. Stripe keeps its own records under its privacy policy.
• Backups of the database are kept for a limited period and then expire on a rolling basis, so deleted data may remain in a backup until that backup expires.

A workspace owner can request deletion of the workspace in the application, and can cancel the request during a short grace period (currently 7 days on self-serve plans). You can also ask us to delete your workspace or your personal data by emailing privacy@trefur.com.

7. International transfers

Our application and database are hosted in Singapore. Some of our service providers process data in other countries, including the United States. Where the law requires a transfer safeguard, we rely on appropriate safeguards required by applicable law, such as the European Commission’s standard contractual clauses and, for transfers from Singapore, contractual protections that meet the Personal Data Protection Act 2012. You can ask us for more information about the safeguards we use.

8. Security

• The application is served over HTTPS, and our SDKs and collector send telemetry over TLS by default.
• Connections to our database require TLS.
• Credentials and OAuth tokens you give us are encrypted in our database with AES-256-GCM, using per-workspace keys.
• Workspace data is separated by row-level security enforced in the database, in addition to application checks.
• Access is role-based, sensitive actions such as identity reveals are written to an audit log, and sign-in supports multi-factor authentication.

No method of transmission or storage is completely secure, and we cannot guarantee absolute security. You are responsible for keeping your passwords and keys safe. If we become aware of a breach of security affecting your personal data, we will notify you and the relevant authorities as the law requires.

9. Your rights

Depending on where you live, you may have the right to: access the personal data we hold about you; correct it; delete it; receive it in a portable format; restrict or object to our processing; withdraw consent where we rely on it; and not be discriminated against for exercising these rights. Residents of California and other US states have the right to know what personal information we collect, use and disclose, and to request its deletion or correction. We do not sell personal information or share it for cross-context behavioural advertising.

To exercise a right, email privacy@trefur.com from the address on your account, or describe your request if you do not have one. We may need to verify your identity, and an authorised agent may make a request on your behalf with proof of authority. We respond within the time the law requires (for example one month under the GDPR, which can be extended in some cases). Where we process your data as a processor for one of our customers, we will pass your request to that customer and help it respond.

You also have the right to complain to your data protection authority. We would appreciate the chance to address your concern first.

10. Children

The Services are for businesses and are not directed to children. We do not knowingly collect personal data from anyone under 18. If you believe a child has given us personal data, contact us and we will delete it.

11. Changes to this policy

We may update this policy. We will post the new version here with a new “Last updated” date and, if the changes are material, tell account owners by email or in the application before they take effect.

12. Contact