Privacy policy

The short version

Everything below is the longer, more careful version of that paragraph.

What we collect

The selection you make. When you run a visual search or ask about an area, Lensable captures the region you outlined or boxed, plus the URL of the page it came from. It is captured at that moment, for that action.

Your question. If you ask about a selection, the text of your question travels with it.

Page context, only when Content Aware is on. With Content Aware enabled, a small amount of surrounding text — typically the page title and nearby paragraphs — is included so the answer fits what you are reading. You can turn this off.

Account and billing basics. If you create an account, we store your email address and your subscription status. Accounts, and the server-held data described on this page, are stored with Supabase, our database and sign-in provider. Card details are handled entirely by Stripe, our payment processor, and never reach our servers.

Search History. When a visual search returns results we keep a short record of it: the object we identified, its category, a thumbnail of it, how many matches came back, and the top match. It does not include the page you were on or the image you selected. You can pause it, or clear it entirely, from your account.

Saved Highlights. When a highlight syncs to your account, we store the passage you chose, its page title, source URL and saved date. The account labels this collection as server-synced because a browser-local extension copy may not have reached our server. Saved Highlights are stored on their own: they are not read to work out what you might be interested in.

Topics you follow. Following is something you start. Where it is available on your account, you add a topic yourself under Account → Interests. Lensable never creates a follow for you, and saving a highlight does not start one. For each follow we store the name you gave it, how often you asked us to check the topic — weekly or daily — and whether you have muted it. The canonical name of the topic itself is stored once and shared with everyone following the same topic, so it is not private to you. Checking is shared in the same way: a topic is checked once for everyone following it, and the shortest choice among people still following it may decide how often that happens. Checking is not delivery. A brief only reaches you when a check finds a meaningful update it can cite, so a shorter checking interval does not mean more briefs.

Briefs delivered to you. When a topic you follow produces a brief, we record that it was delivered to you, when it was delivered, and whether and when you opened it. The record also carries a field for marking a brief as notified. Lensable does not currently send emails, push messages or extension alerts, so nothing is sent on the back of that record.

Usage and cost records. We count how many visual searches you have used in the current period so quotas work. We also keep one row per billable action recording what kind it was, which model answered, and what it cost us to run. Those rows carry no page address, no page text, and no part of your selection.

What we never collect

• Continuous or background screenshots of your screen.
• Your browsing history, open tabs, or pages you visit without acting on them.
• Keystrokes, passwords, or the contents of form fields.
• Data from DRM-protected video, which is technically inaccessible to us and returns blank by design.
• Interests worked out from what you read, save or search. A topic is followed only because you added it.
• Anything used to build an advertising profile. We do not sell or rent personal data.

Where your selection goes

The selected image and your question are sent to the AI provider you picked for that question. Each choice in the picker is operated by a named company: ChatGPT by OpenAI, Claude by Anthropic, Gemini by Google, Grok by xAI, and Perplexity by Perplexity AI. Lensable Auto is not a separate provider — it means our server picks one of those five for you. Your selection also passes through our own servers, only insofar as we route the request and hold the provider credentials.

For product lookups, the selection is sent to visual search and shopping data providers so matches can be returned. Which provider serves a given search depends on what is configured and available at that moment; the full set we use is SerpApi (Google Lens and Google Shopping results), Serper, Scrapingdog and DataForSEO (search and shopping data), eBay’s Browse API (eBay listings), and Google Cloud Vision (finding where else an image appears). Those results include links to third-party retailers; nothing is sent to a retailer unless you click one of those links.

If you highlight a link and ask for the page behind it to be summarised, our server fetches that page itself. It is fetched from our server rather than your browser, so none of your cookies are sent and nothing on the page is executed. We send the address you highlighted; we do not send anything else about where you were.

Your account can generate Summary of selected highlights. When you ask for it, Lensable retrieves the selected server-synced highlights by their saved IDs and sends their passage text, page titles and source URLs to Google Gemini. In extension version 1.1.0, the extension instead sends the selected browser-local passage text, page titles and source URLs to Lensable for the same Gemini summary. Both paths run only when you ask, use Gemini rather than the provider selector used for questions, and do not browse or monitor sources.

Each provider handles data under its own policy. For actions that offer a provider choice, we do not send your selection to a provider you did not choose. The Saved Highlights summary uses Gemini as described above, as do the three follow-related uses described in the next section: suggesting a topic name, judging which retrieved items are relevant, and writing a brief.

Every party your data can reach. For the avoidance of doubt, this is the complete list:

No party beyond this table receives your data, and each of these receives it as a processor for an action you took — never for advertising, profiling or resale.

Following topics

This section describes following, and applies only where that feature is available on your account. Where it is not, nothing in it happens.

Naming a topic. When you ask for a suggested name while adding a topic, the text you typed into that box is sent to Google Gemini so it can propose a name and a one-line description of the subject area. It never runs unless you ask for it. Nothing about you accompanies it, nothing is stored by that request, and you can ignore the suggestion and follow the name you wrote.

What we retrieve. For a followed topic our server fetches publicly available material about the topic itself, never about you. Today that means arXiv, PubMed, RSS or Atom feeds set up for the topic, and, where it is switched on, a third-party web-search provider. The requests are made by our server, carry no account identifier, and say nothing about who follows the topic or what they have read.

What we keep from it. For each retrieved item we store its title, an excerpt, its address, the name of the source, any listed authors, the publication date where one is given, and a score and short reason recording why it was judged relevant to the topic. These rows belong to the topic rather than to you, and are shared by everyone following it.

Judging what is relevant. That score and reason come from a keyword rule first, which involves no model at all. Only the items the rule cannot settle are sent to Google Gemini, in one batched request per topic per retrieval run, carrying the topic name and each undecided item's title and excerpt — retrieved third-party material, and nothing else. No account identifier, no highlight or passage text, no search history and nothing about who follows the topic travels with it.

Briefs. The items retrieved for a period — their titles, excerpts, source names and publication dates — are sent to Google Gemini to write a short brief that cites them. Claims the model cannot tie back to a retrieved source are dropped before the brief is stored. A brief is written for the topic and shared by its followers; only a brief judged to contain a meaningful update is delivered, and a muted follow is delivered nothing. Briefs are written from third-party sources and can be wrong, incomplete or out of date — read the cited source before relying on one.

What following does not use. Your Saved Highlights, saved items, search history and questions are not read to choose topics, are not used to rank what you see, and are not sent with a retrieval, relevance or brief request. None of the three model steps above receives them. Nothing you follow is set up on your behalf, and no interest is worked out for you.

How we handle and protect it

Everything travels over encrypted connections. Traffic between your browser and our servers, and between our servers and every provider named above, uses HTTPS/TLS.

Server-held data lives in Supabase, which encrypts it at rest. Our application runs on Render. The credentials used to call the providers above are held on the server and are never present in the extension, so nothing on your machine can leak them.

Access to server-held data is limited to the application itself and to us as its operator, for support and troubleshooting. We do not sell it, share it beyond the parties named above, or use it for advertising.

How long it is kept

Your rights

You can delete an individual server-synced Saved Highlight from your account. You can delete a browser-local copy from inside the extension; when you are signed in and the highlight was synced, Lensable also attempts to delete the matching server copy. Browser-local and server-synced copies are separate, so deleting one does not guarantee the other copy has been removed.

Where following is available on your account, unfollowing a topic under Account → Interests removes your follow straight away. Briefs already delivered to you remain in your account until it is closed, and the topic itself — its shared name, its sources and the briefs written for it — is not removed, because other people may still follow it. Closing your account removes your follows and your delivery records.

To request access to, correction of, a copy of, or bulk deletion of server-held data, or to ask us to close your account, follow the instructions on the Delete data page and email us from the address on your account. These requests are handled by us; there is not currently a self-service bulk export or account-deletion control.

If you are in the UK, EEA or Switzerland you have rights under the UK GDPR and GDPR, including the right to object to processing and to lodge a complaint with your supervisory authority. If you are in California you have rights under the CCPA, including the right to know and the right to delete. We do not sell personal information as the CCPA defines it.

Children

Lensable is not directed at children and is not intended for anyone under 13. We do not knowingly collect data from children. If you believe a child has provided us data, contact us and we will remove it.

Changes and contact

If this policy changes in a way that materially affects you, we will say so in the extension before the change takes effect rather than quietly editing this page.

Questions about privacy go to our contact page.