What Data We Collect and How We Use It

This is the short, plain-English version of how data flows through The GAiGE. It's written so admins can answer their team's questions confidently, and so you can paste relevant sections into an internal comms email when you roll the platform out.

The one-sentence summary

We collect the short structured answers your team gives in pulses, plus the basic account information needed to log in. Everything your admins and leaders then see is aggregated, with no name attached to any answer. We never read the contents of the AI tools your team uses, never share data with anyone else, and never use it to train AI models.

What we collect from your team: pulse responses

When someone answers a pulse via the Chrome extension, we receive:

• The answers they chose. Yes/no, time-saved bucket, areas helped, satisfaction rating, and a free-text comment if they leave one.
• Which tool the pulse was about (e.g, chatGPT).
• When they answered (timestamp).
• Their user ID. Stored so the person can see their own history, and so a one-off pulse is only ever asked of them once. It is never shown next to an answer anywhere in the platform (more on this below).

That's the entire pulse payload. We don't capture the page they were on, anything they typed into the AI tool, the conversation they were having, or any other browser activity.

What we collect about your account

When someone signs up or is invited:

• Email address and name
• Their role (Owner, Admin, Group Leader, Member) and which groups they're in
• Their time zone (so reports show their data correctly)
• Authentication tokens (handled by Clerk, we don't store passwords)
• Last-active timestamp (so admins can see who's been around)

We don't ask for or collect: phone numbers, addresses, dates of birth, payment details (those go directly to Stripe), or any other personal information beyond email and name.

Information you (the admin) supply about your tools

This isn't really collected (you enter it), but it lives in our database and shapes the reports:

• The AI tools your org uses, their websites, and their per-seat costs
• Your blended hourly rate and currency
• Which users are assigned to which tools

This data is used only for your org's reporting and is invisible to anyone outside it.

What the browser extension can and can't see

At install time the extension is granted:

• Local storage: keeps pulse questions, the schedule and unsent answers on the user's own machine
• Alarms: a once-a-minute timer that checks whether a pulse is due, plus a fifteen-minute re-sync
• Scripting and tabs: places the pulse card on the current page
• Access to thegaige.com and api.thegaige.com: sign-in and submitting answers

Broad site access (<all_urls>) is optional and requested separately, from the extension's own popup. Pulses are delivered on a cadence rather than in response to visiting a particular site, so when one comes due it appears on whatever page the person has open. Chrome has one permission for "any page", which is the one the extension asks for. Each user grants it themselves; it cannot be granted centrally, force-install included.

Reviewers should read the permission as broad and the behaviour as narrow. With that access, the extension draws a card in a frame it owns and reads nothing else.

It deliberately does not:

• Read or transmit the content of any page (prompts, conversations, generated output: none of it leaves the browser)
• Transmit any web address. A submitted response carries the answers, the pulse and tool they relate to, the extension version, and whether the pulse was scheduled or user-initiated. The current tab's address is read locally, once, purely to rule out pages no extension may draw on (chrome://, edge://, about:, the Chrome Web Store)
• Store credentials for AI tools
• Read browsing history or bookmarks. Neither permission is requested, so neither is available to it
• Communicate with any third party. The GAiGE API only

A long-lived API token signs the extension in to your account. The token is stored hashed in our database; only the original device can use it.

Who can see what (within your org)

Every query in the platform is scoped to your organisation, and within that, scoped by role:

A Member sees what they personally answered. A Group Leader sees results for their own teams and not other teams. Owners and Admins see the whole organisation. In every one of those cases the results are aggregates: counts, averages, trends and comments, with no name beside them.

Why a response is stored against a user, and why nobody sees that link

Every pulse response is stored with the user ID of the person who answered it. That link does two jobs:

1. People can see their own history. A member's dashboard shows their own responses, their own hours-saved estimate, and the tools they rate highest.
2. A one-off pulse is only asked once. Where a pulse runs a single time rather than on a cadence, the link is how we know someone has already answered it.

Nobody inside your organisation ever sees that link. The reports, the response list and the CSV export all leave names and email addresses out of the query, so an admin or a leader sees what was said and never who said it. There is no responses-by-person view, no name column in the export, and no per-user response count on anyone's profile.

Two things are worth being straight about.

Comments are quoted word for word. Free text is the most useful thing in a pulse, so it appears in reports exactly as it was typed, with no name attached. If a comment describes a situation only one person is in, the words themselves can identify the writer. We say so plainly in the article your team reads, Who sees my answers?.

A segment needs five people before it says anything. Results for a segment appear once at least five different people have answered it. Below that the report gives the number who answered and nothing else. That floor covers group rows, question breakdowns, one-off pulse themes and head-to-head comparisons.

Per-person signal is possible, but only if you build it on purpose: create a group with one member in it, filter a report to that group, and the numbers you get are that person's. It takes a deliberate act by an admin to set up, and it still shows numbers rather than a name against each answer. If you go that way, tell the person.

Where the data lives

• Database: PostgreSQL on AWS RDS in the Sydney region (ap-southeast-2)
• File storage: AWS S3, same region
• Authentication: Clerk (US-hosted; GDPR/Australian Privacy Act compliant)
• Payments: Stripe (encrypted at the edge; we never see card data)
• Logs and metrics: AWS CloudWatch, same region

All transit is TLS-encrypted. Data at rest is encrypted using AWS-managed keys.

What we don't do

Important things we make a point of not doing:

• We do not sell, share, or rent data to third parties
• We do not use your data to train any AI models. Ours or anyone else's
• We do not read page content from the AI tools your team uses
• We do not combine your data with other organisations' data without explicit consent
• We do not send your data to advertising networks

How to delete data

• A single user. Delete them on the Users page. Their name and email are anonymised, and their past responses stay in the reports for historical continuity, carrying the label "Deleted User". Owners and your own account can't be removed this way.
• A single user's responses as well. Full erasure, responses included, is a support request today. Email support@thegaige.com and we'll run it. There's no button for it in the platform yet.
• A copy of the data first. Also a support request, same address, whether you want one person's data or the whole organisation's.
• Your entire organisation. Cancel your subscription, then ask us to delete the organisation. After cancellation the data is kept read-only for 90 days so you can export it or come back, and may be permanently deleted after that.

Sharing this with your team

Here's copy you're welcome to paste into an internal email when you roll out The GAiGE. Please check it still matches your own privacy notice before you send it.

We're rolling out a tool called The GAiGE, so we can work out which AI tools are actually saving us time and which aren't worth what they cost.

Every so often a small panel appears in the corner of whatever page you're on, with one to three short questions about the AI tools you have access to. About 30 seconds. Answer it straight, including when a tool has wasted your time, because that's the answer most likely to change what we buy. You can skip it, or come back to it later.

Your answers are anonymous. The reports show averages, totals, trends and comments across the team, with no name attached to any of them. A group's results only appear once at least five people in it have answered. There's no per-person view, and no name column in anything we can export. There's no list anywhere of who has and hasn't answered.

The platform reads nothing you type into the AI tools themselves. Only the answers you give in the panel.

We'll use what comes back in aggregate, to decide what we keep paying for and where to put learning time. Privacy details are at [link to your internal privacy page].

Related: Who sees my answers? · Will my answers affect my performance review? · What data is collected · User roles explained