Google taught screens to recognize your face.
Here is what a real report says about it.
Twelve issued patents. Thirty-three independent claims. Named products, an honest valuation band, and the proof gaps stated plainly — the complete Tier 6 Master Monetization File, published as our public sample so you can judge the thinking before you pay for any of it.
Shown, not described.
Every excerpt below is lifted from the report itself. This is what you would actually receive — the questions answered first, the patentese translated, the evidence charted, the value stated as a band with the assumptions named.
The questions a buyer asks first, answered first.
The report opens with plain answers, not methodology. You know the finding before you know the footnotes.
| What is being analyzed? | A family of Google patents — US 8,965,170 and its continuations — covering connected screens that recognize who is in front of them and change the content accordingly. |
| Who appears to be using it? | Strongest matches: Amazon Echo Show (Visual ID) and Google Nest Hub Max (Face Match). A second group: commercial-PC walk-away presence (Lenovo, Dell), Apple attention-aware / Face ID, smart-TV face switching (Samsung, Panasonic), Sony BRAVIA CAM, and Amazon Astro. |
| What might the possibly infringing uses be worth, if confirmed? | Three independent analyses give most-likely figures of $11.3M, $32M, and $88M, with floors near $1.9M and ceilings up to ~$340M. A defensible middle estimate sits in the low-to-mid tens of millions. |
| What still needs to be checked? | Five items, named one by one — from confirming the patent records against the USPTO to closing the cross-room hand-off evidence gap. A required next step, not finished work. |
Patentese, translated. Same sentence, both languages.
The claim text stays verbatim — that's the law. The explanation sits beside it — that's the point. Written for experts in things other than IP.
"selecting a program from a plurality of programs based on the user identifier"
"associating a time parameter of the program that corresponds to the detection with the user identifier"
"determining that the face of the viewer that has been identified within the second area based on the received second signal corresponds to the user identifier"
Choosing one program out of several, where the choice is driven by the retrieved identifier.
Recording a time marker — the point in the program at the moment interaction stopped — and binding it to the viewer's identifier. In plain terms: a bookmark for how far into the program the viewer had gotten.
Recognizing, in the second area, that the face matches the same identifier. The same viewer just walked into another room.
The claim chart: every element, cited to a public source.
A claim chart puts each requirement of the claim next to the public evidence that a product does it. Three rows from the Echo Show chart — the full report charts ten products this way.
| Claim language (verbatim) | Public documentation |
|---|---|
| "determine that the first image data does not include a face of a user;" | The recognition pipeline performs facial detection before recognition, distinguishing frames with no enrolled face. "The system... first has to detect that a face is present (facial detection) and then determine whose face it is (facial recognition)."amazon.science/blog/the-science-behind-visual-id |
| "determine a user identifier that uniquely identifies the user from a plurality of users associated with the device..." | The device matches the in-view face against the enrolled set — up to 10 members per device — selecting the specific individual. "Visual ID can be set up for up to 10 people per device."amazon.com/gp/help — Visual ID |
| "cause the content item to be presented on the display of the device." | "When Alexa recognizes you, the device shows content from your Alexa profile, such as your calendar, reminders, and recently played music."amazon.com/gp/help — Visual ID |
Value as a band, not a number — with the assumptions named.
Anyone can hand you one flattering number. The report runs three independent models, shows you all three, and tells you exactly which assumptions drive the spread.
| Model | Most-likely value | Floor | Ceiling |
|---|---|---|---|
| Conservative | $11.3 million | $1.9 million | $28.0 million |
| Middle | $32.0 million | $2.4 million | $65.0 million |
| Aggressive | $88.0 million | $28.0 million | $340.0 million |
Labels that say how strong the evidence actually is.
Every named product carries one of four proof-state labels. None asserts infringement; each tells you what the public record shows and what the next step costs. We hold no stake in which label a product earns — the answer is the answer.
Public sources describe the product doing each item on the patent claim's checklist. The strongest the public record gets.
Public sources show the main feature, but one step on the checklist still needs confirmation by teardown, testing, manuals, or discovery.
The right technical area, but the record is not yet enough to say more. Worth pursuing, not yet worth charting as a match.
A required fact is missing, contradicted by the vendor, or the product predates the patent — treated as prior art, not a candidate.
When evidence is missing, the report says so — and names the test.
Amazon's Astro robot follows a recognized person from room to room playing their media. Close to the patents' hand-off claims — but one required element isn't in the public record, and the report refuses to paper over it.
The single most important fact in the report cuts against the biggest number.
A firm paid on the upside buries this paragraph. We are paid for the analysis, not the outcome — so it leads.
That one paragraph tells the owner where the bankable value sits, which patents carry upside that depends on better evidence, and why the aggressive $340M ceiling should be read with both hands on the table.
Read all of it. Then decide.
The full sample runs from the first plain-English finding to the last source URL — status tables, ten claim charts, the valuation models, the open items for counsel. Your own report starts free and takes a few business days.
Illustrative public sample, generated from public materials with substantial AI assistance and only partly verified; full field-by-field confirmation against the issued patents remains an open item for patent counsel. Not a legal opinion, a valuation opinion, or investment advice. Every product named is an illustrative, unverified hypothesis about where the patented system might appear in the market — not an accusation. Whether any product actually practices any claim is a legal question reserved to qualified patent counsel.
Get the sample report
Enter your email and we will open the full report in a new tab.