Privacy Notice

This Privacy Notice explains how the Trinetix FLOW browser extension and embed runtime (together, the “FLOW runtime”) process information when they operate on the web applications used within your organization. It applies to the software components that run in your browser and to the associated Trinetix services that support them.

1. Overview of the FLOW runtime

The FLOW runtime displays in-application guidance, including guided tours, tooltips, pop-ups, a launcher, a resource center and, where enabled, an artificial-intelligence assistant, on top of the web applications you use. All such content is configured by our organization through the Trinetix FLOW platform. Your organization determines which guidance is published and whether artificial-intelligence features described below are enabled.

2. Information the FLOW runtime does not collect

The FLOW runtime is designed to minimize the information it collects. In particular:

• By default, it does not collect your name, email address or login credentials. Your organization is identified on the server from a signed token, and your individual identity is not transmitted.
• It does not store your IP address or User-Agent string within analytics. Your IP address is used only transiently on the server for rate-limiting purposes; it is not stored together with events and is not returned to the page.
• It does not read or transmit cookies, and it does not log keystrokes. Rules that respond to form fields rely on the browser’s “change” event, which is triggered only when a value has been committed, and never on individual keystrokes.
• Page content used for targeting purposes, including on-page text, form values, browser storage and page variables, is read solely within your browser in order to determine which guidance to display, and is not transmitted. Where your organization sets up role-based targeting, this may include a value it designates, such as a user name or email; like all targeting data, that value is evaluated in your browser and is not transmitted, and only the resulting segment membership is used. The artificial-intelligence assistant, where enabled, constitutes the sole exception, as described in Section 4.

3. Information the FLOW runtime collects, and when

Anonymous identifiers. The FLOW runtime generates a random visitor identifier, stored in your browser’s localStorage and scoped to each website, and a session identifier, stored in sessionStorage on a per-tab basis and rotated following thirty minutes of inactivity. These identifiers are not linked to your identity and are not shared across devices. They are transmitted together with analytics data.

Product analytics. Analytics is enabled by default for your organization. For each event, the following information is transmitted: the event name, drawn from a fixed list (for example “tour started”); a timestamp; the page URL reduced to its domain and path, excluding any query string or fragment; and structural properties such as content identifiers and labels, durations, device type, delivery channel, and the manner in which a tour was initiated. The events cover the guidance surfaces (tours and their individual steps, tooltips, pop-ups, launchers and the resource center) and a small set of health and diagnostic signals, for example when a session starts or when a tooltip could not locate its target element; the latter may include the element selector involved. No page content is transmitted. The only free-text value transmitted through analytics is the text you enter into the Resource Center search box, which is limited to two hundred characters, is not further masked, and is transmitted only if you use the search function; this text also helps identify searches that returned no results. “Device type” is recorded as a category (desktop, tablet or mobile) derived from your window width; exact screen dimensions are not transmitted.

4. The artificial-intelligence assistant

The artificial-intelligence assistant operates only where your organization has enabled it and has configured an artificial-intelligence key. It is activated in one of two ways:

• Upon a direct request, in which case it processes the question you enter, subject to a limit of two thousand characters.
• Through proactive assistance, where enabled, in which case it monitors locally for on-screen errors and for indications that you may require assistance, such as repeated clicks, prolonged inactivity or failed form validation. Such “assistance” signals are evaluated on your device and are not transmitted. Only when the assistant offers help is the captured on-screen error text transmitted, subject to a limit of four hundred characters.

In both cases, the assistant also transmits the page URL and the CSS selector of the relevant element, so that appropriate guidance can be identified.

Masking. Sensitive values are masked before any request reaches an artificial-intelligence model or is stored. On your device, email addresses, long numeric sequences (of seven digits or more), and recognizable secret, payment-card and social-security-number patterns are masked first. On the server, which is authoritative, phone numbers, an organization-defined denylist of keywords, and organization-defined patterns are additionally masked. Where such server-side masking cannot be applied, no call to an artificial-intelligence model is made and no data is stored (a “fail-closed” approach). The page URL and the element selector are deliberately not masked; please refer to Section 7.

5. Service requests

In order to load content and verify eligibility, the FLOW runtime transmits a public site key or a signed session token. The signed session token encodes the relevant system and environment, is never transmitted as separate fields, and is used to derive your organization on the server. For eligibility checks, the runtime additionally transmits a stable internal identifier derived from your system and organization; this identifier is not personal. No personal data is transmitted.

The browser extension additionally transmits the website’s domain (the origin only, and not the full URL) in order to verify whether your organization has published guidance for that website, and requests the list of registered domains upon starting. To perform these functions, the extension operates on the websites you visit; on websites for which your organization has published guidance, it removes that website’s content-security-policy response headers (including any report-only policy) so that the guidance can load.

6. Information stored solely on your device

Tour progress (enabling the “resume where you left off” function) and “already shown” and frequency-cap flags are stored locally within your browser and are not transmitted. The anonymous identifiers described in Section 3 are likewise stored locally but, as noted, are transmitted together with analytics data.

7. How information is used and shared

• Information is used solely to deliver guidance and, where enabled, to measure adoption and to operate the artificial-intelligence assistant. Trinetix does not sell personal data and does not use it for advertising purposes.
• Artificial-intelligence processing. Where artificial-intelligence features are used, the masked question or error text is transmitted to an artificial-intelligence model provider (currently Anthropic (Claude), whose processing takes place in the United States; the model and the key are configured on a per-organization basis) in order to generate a response. To identify relevant guidance, the masked question is also converted into a numeric vector by the platform’s embeddings provider (Microsoft Azure OpenAI), which processes it within the European Union (Sweden Central). Masking is applied prior to transmission in both cases. The model provider processes data outside the European Union; the sub-processors and their locations are listed in Section 10.
• Information stored on the server for artificial-intelligence purposes (where enabled). This comprises the masked question and its response and citations; where the assistant is unable to provide a response, a “content gap” record containing the masked query text and the page URL (see Section 8); anonymized masking counts (that is, the number of email addresses, phone numbers and similar values masked, expressed as counts only and never as the underlying text); and token-usage counts for budgeting purposes.
• Organizational control. Your organization determines what content is published and whether the artificial-intelligence features are enabled. Data is processed within your organization’s FLOW tenant.

8. Page URLs used for artificial-intelligence features

For the purpose of generating assistance, the full page URL, including any query string and fragment, is transmitted and stored within the artificial-intelligence “content gap” record. Analytics data, by contrast, retains only the domain and path. As a query string may contain sensitive values, you should avoid including personal data in URLs on pages where the artificial-intelligence assistant is enabled.

9. Retention

FLOW keeps analytics events and artificial-intelligence records for as long as your organization’s FLOW tenant and the corresponding system remain active. These records are deleted when your organization’s tenant or that system is deleted, and Trinetix will delete or return them earlier at your organization’s request, ordinarily within one month. Session recordings captured to help author artificial-intelligence knowledge are deleted automatically after sixty days. Because the runtime identifies end-users only by random identifiers, deletion requests are handled at the level of your organization’s data rather than an individual end-user (see Section 10).

10. Legal and data-protection information

Controller and processor roles. Your organization is the data controller for the information processed through the FLOW runtime, because your organization decides what guidance is published and whether the artificial-intelligence features are enabled. Trinetix acts as the data processor, processing that information on your organization’s instructions and within your organization’s FLOW tenant. Trinetix acts as a controller only for the limited operational data it generates to run and bill the service, such as aggregate token-usage counts and masking counts; this operational data contains no page content and no free text.

Your rights. Where the General Data Protection Regulation applies, you have the right to request access to your personal data and its rectification or erasure, to request restriction of or to object to its processing, to data portability, to withdraw consent where processing relies on consent, and to lodge a complaint with a supervisory authority. You may exercise these rights as follows:

• Because your organization is the controller, requests should be directed to your organization in the first instance. Trinetix will assist your organization in responding and will forward to it any request it receives directly.
• Requests may be sent to the contact in Section 11, and a response will ordinarily be provided within one month.
• The FLOW runtime identifies you only by random visitor and session identifiers that are not linked to your name or account. As permitted by Article 11 of the Regulation, if Trinetix cannot identify you from the data it holds, it may be unable to act on a request unless you provide additional information that enables your records to be located.

Sub-processors. Trinetix uses the following sub-processors to provide the service:

• Microsoft Azure, which hosts the FLOW databases, cache and application services in the European Union (West Europe region). This applies to all FLOW customers.
• Anthropic, which provides the Claude artificial-intelligence model that generates assistant responses, with processing in the United States. This applies only where your organization has enabled the artificial-intelligence assistant.
• Microsoft Azure OpenAI, which converts the masked question into a numeric vector so that relevant guidance can be found, hosted in Microsoft Azure’s Sweden Central region within the European Union. This applies only where your organization has enabled the artificial-intelligence assistant.

11. Legal entity

For any questions regarding this Privacy Notice or the processing described above, please contact:

Name: Trinetix Inc.
Postal address: 30 N Gould St STE 4722, Sheridan, WY 82801, USA
Email: requests.legal@trinetix.com
Data Protection Officer: Taras Lytovchenko, email: requests.legal@trinetix.com