US 10628857

I'll verify the patent's current status via search, then summarize based on the authoritative full text provided.

Let me try to retrieve the actual claim language, since the provided full text is truncated before the claims.

Let me make one more attempt to pull the exact independent claim language.

US Patent 10,628,857 B2 — Technical Summary

Source of record: Full patent text at https://patents.google.com/patent/[US10628857](/patent/US10628857)/en (fetched 2026-09-28), cross-checked against the issued patent PDF (US010628857B2) and litigation filings.

1. Bibliographic Data

Prosecution chain (continuation lineage stated in the patent):
09/723,391 (filed 2000-11-28, now US 6,832,207) → 10/996,491 (2004-11-24, now US 7,822,639) → 12/911,601 (2010-10-25, abandoned) → 13/368,686 (2012-02-08, now US 8,244,586) → 13/585,765 (2012-08-14, now US 8,600,815) → 14/091,285 (2013-11-26, now US 10,026,100) → 16/033,035 (the '857 patent).

Legal status (as listed): "Expired – Fee Related." The listed adjusted expiration is 2021-03-20, which is consistent with a 20-year term running from the 2000-11-28 earliest non-provisional filing date (i.e., 2020-11-28) plus the 112-day PTA. Google's page carries the standard disclaimer that legal status is an assumption, not a legal conclusion.

2. Abstract (verbatim)

"A super-saturation method for information-media relates to directing information to an electronic media consumer (reader, surfer, viewer, listener, etc.). Embodiments of the present invention facilitate a first media entity substantially offering out of context information placement using a second media entity. A facilitator entity guarantees that a consumer of the second media is a known consumer of the first media. Accordingly, the second media entity presents an out of context information placement. For example, an Internet site 'AAA' is oversubscribed with potential paying advertising content at $100 CPM. This site offers advertisers an option to present their advertisements to certified 'AAA' viewers, albeit on a non-AAA Internet site, for $50 CPM. The facilitator locates certified AAA viewers at Internet site 'BBB' that normally charges $30 CPM. Contracts between AAA, BBB, and the facilitator divide a new revenue stream of $20 CPM among them."

3. Plain-Language Overview of the Independent Claims

⚠️ Important caveat on completeness. The full text supplied to me is truncated in the "Detailed Description" (it cuts off mid-sentence: "If a user browses the Financial section and then goes to the…"). The actual claim set — columns of the issued patent — is not included in the retrieved text. What follows is derived from the specification's Summary-of-Invention and Detailed Description, whose numbered paragraphs (a)/(b)/(c) and sub-parts (i)/(ii) closely track the claim language of this family. I am not reproducing verbatim claim text, and I cannot state with certainty how many independent claims the '857 patent contains. Treat the below as a faithful description of claim scope, not as claim language.

The patent's disclosure recites four parallel claim families, each expressing the same "super-saturation" concept in a different statutory form:

Family A — Method (visitor-identification / tagging / recognition method). A three-party method for extending advertising beyond a first site's saturation point:

• (a) a facilitator entity ("agency") facilitates identification of visitors;
• (b) in conjunction with the agency, a first broadcaster/media tags a preponderance of its visitors with a tag (e.g., a cookie, browser identification message, telephone-system database notification, credit-card database warning, public-service database notice, legal-instrument database entry, cellular SIM-card message, or wireless-service database notification);
• (c) in conjunction with the agency, a second broadcaster/media recognizes a visitor as bearing that tag and, by proxy — either in conjunction with the agency or with the first broadcaster — accepts the off-site content presentation for the recognized visitor.

The off-site presentation may be an advertisement, public-service announcement, personal reminder, judicial instrument, credit warning, graphic item, multimedia presentation, audio presentation, or banner. Delivery modes claimed include dropping the content into the visitor's browser, causing the browser to fetch it, or sending it to the recognized visitor via the second medium. The first and second media may even be the same medium (e.g., a sold-out "Financial" section vs. an unsold "General News" section of one site).

Family B — Contracting structure. The same arrangement expressed as a set of three interlocking agreements:

• (a) First agreement — agency ↔ content provider: agency facilitates delivery of the content provider's content to an identified off-site visitor, and the content provider pays for that delivery;
• (b) Second agreement — agency ↔ first broadcaster: the first broadcaster provides facilitated tagging of a preponderance of its visitors; the agency provides facilitated off-site placement of content from the first agreement and pays the first broadcaster for substantially each such placement;
• (c) Third agreement — agency ↔ second broadcaster: the second broadcaster provides facilitated recognition of a tagged visitor and, by proxy (with the agency or the first broadcaster), facilitates acceptance of the off-site content presentation, with the agency paying for that facilitated acceptance.

Family C — Computer program product. A computer-usable medium carrying program code for causing the agency to facilitate visitor identification; the first broadcaster to tag the preponderance of visitors; and the second broadcaster to recognize the tag and accept the off-site presentation by proxy. A parallel program-product set covers the three-agreement contracting structure.

Family D — Program storage device readable by machine. A machine-readable storage device tangibly embodying instructions executable to perform the same method steps (visitor identification → tagging → recognition-and-acceptance) and to implement the three-agreement contracting structure.

Dependent-claim subject matter (from the specification's enumerated variants): "preferred visitor" classification (time spent at the first site, money spent, questionnaire responses); specific tag types (cookie, SIM-card message, credit-card warning, legal instrument, wireless-service notification, etc.); specific message types; specific delivery mechanisms; and sameness of the first and second media.

4. Litigation / Docket Findings

USPTO and CAFC search result: I found no Court of Appeals for the Federal Circuit docket in 2026 naming US 10,628,857. The 2026 Federal Circuit material surfaced in my search (Dolby/Unified Patents on 10,237,577; US Patent No. 7,679,637 LLC v. Google; Oasis Tooling v. Siemens; Teva v. Lilly) is unrelated to this patent. I cannot confirm any 2026 appellate activity for the '857 patent — treat that as a negative finding limited to my search, not as an authoritative statement that no such docket exists.

District court activity actually found for the '857 patent:

• AlmondNet, Inc. et al. v. Roku, Inc., W.D. Tex. No. 6:21-cv-00731-ADA (filed 2021-07-15) — '857 asserted (complaint Ex. 1-14 is a copy of the patent). This case was later transferred/stayed pending the Delaware action.
• Roku, Inc. v. AlmondNet, Inc., D. Del. No. 1:21-cv-01035 — declaratory-judgment action; Roku's invalidity counterclaims name the '857 patent as one of nine patents-in-suit (IPR2022-01236, Ex. 2003). This was consolidated with No. 1:22-cv-01540, and one secondary source reports the consolidated proceeding concluded 2025-06-23 (not independently verified by me).
• Additional AlmondNet campaigns naming the '857 patent include the Samsung W.D. Tex. / N.D. Cal. matters (e.g., 6:21-cv-00891) and the AlmondNet v. Oracle matter (6:24-cv-00303).
• Related-family verdict: a W.D. Tex. jury reportedly awarded AlmondNet $122 million against Amazon in June 2024 on US 7,822,639 and US 8,671,139 — sibling patents in the same family, not the '857 patent itself.

5. Uncertainty Flags

1. Claim text not verified. I do not have the literal wording of the '857 patent's independent claims, nor a confirmed count of independent claims. My claim overview is reconstructed from the specification's summary language, which for this family has historically tracked the claims closely — but this is not a substitute for reading the issued claims.
2. "Expired – Fee Related" and the 2021-03-20 expiration are reported as they appear on the Google Patents legal-status field, which expressly disclaims accuracy as a legal conclusion. The date is arithmetically consistent with 20 years from the 2000-11-28 priority date plus 112 days PTA, but the maintenance-fee characterization is an assumption in the source.
3. No 2026 CAFC docket located for this patent; absence of evidence in my searches is not proof of absence.
4. The Google Patents "Priority date" is labeled an assumption; the patent itself describes the 2000-11-28 filing (09/723,391) as the original application in the continuation chain.

I'll search for litigation involving US patent 10628857.

Let me dig deeper into specific cases involving this patent.

Let me verify which cases actually asserted the '857 patent and their outcomes.

Litigation Involving U.S. Patent No. 10,628,857

Below is what I could confirm from public patent-litigation sources (Docket Alarm, the Unified Patents litigation portal links embedded in the patent record, and the patent's own "Family has litigation" metadata). A key limitation up front: I was able to identify cases in which the '857 patent was expressly named in the docket/§ 290 filings, but I could not fully verify every AlmondNet-family case, and one case (Oracle) is ambiguous. I flag that uncertainty rather than guess.

Cases where U.S. 10,628,857 is expressly identified

1. AlmondNet, Inc. et al. v. Roku, Inc.

The '857 patent appears in this docket and in the AO 120 (§ 290) report at Docket No. 29, where it is listed among patents "included" by "Other Pleading." Source: https://www.docketalarm.com/cases/Texas_Western_District_Court/6--21-cv-00731/AlmondNet_Inc._et_al_v._Roku_Inc/docs/29.pdf

2. Roku, Inc. v. AlmondNet, Inc. et al. (declaratory-judgment action)

This is the Delaware case referenced directly in the patent's Google Patents "litigation" metadata link (https://portal.unifiedpatents.com/litigation/Delaware%20District%20Court/case/1%3A21-cv-01035). A related Delaware case, AlmondNet, Inc. v. Roku, Inc., 1:22-cv-01540, was consolidated into 1:21-cv-01035.

Case where inclusion of '857 is uncertain — treat with caution

AlmondNet, Inc. et al. v. Oracle Corporation

• Court: W.D. Tex. (Waco), No. 6:24-cv-00303, Judge Albright, filed June 3, 2024; plaintiffs AlmondNet, Inc. and Intent IQ, LLC; defendant Oracle Corporation. Closed — Order Granting Motion to Dismiss (Doc. 60, June 20, 2025) (case docket also flagged "STAYED").
• Caveat: The Docket Alarm "Patent" tag for this case lists only 8,244,582 and 8,677,398. A "10,628,857" reference appears in an exhibit compilation (Doc. 32-14), but that 332-page document appears to reproduce AO 120 forms from several different AlmondNet cases (e.g., Samsung 6:21-cv-00891 and Facebook 6:21-cv-00896). I therefore cannot confirm that the '857 patent was actually asserted against Oracle. If the user requires this, it should be verified against the actual Oracle complaint and infringement contentions, which I was unable to retrieve.

Important clarifications

1. The '857 patent is part of a large AlmondNet/IP-holding family (also including Intent IQ, LLC and Datonics, LLC entities). AlmondNet filed many infringement suits — against Samsung, Roku, Meta/Facebook, Amazon, Microsoft, Oracle, Oath/Yahoo, Viant, Lotame, and LiveIntent. However, most of those cases asserted other AlmondNet patents (e.g., 7,822,639; 8,671,139; 8,677,398; 8,244,582; 7,979,307; 8,775,249; 8,959,146; 11,564,015). The citation of those related patents in search results does not mean the '857 patent was asserted. I filtered specifically for the literal number 10,628,857.

2. No litigation naming a case number and defendant other than Roku could be affirmatively tied to the '857 patent from the evidence gathered. If litigation against additional defendants exists, it is not visible in the sources I could reach within this search.

3. Verification caveat on the number itself: Per the strict instruction, I am interpreting "10628857" literally as U.S. Patent No. 10,628,857 B2, "Methods and apparatus for facilitated off-site targeted internet advertising," assigned to AlmondNet, Inc., application 16/033,035, filed July 11, 2018, granted April 21, 2020, priority date November 28, 2000. I did not substitute any similar number.

Bottom line

• Confirmed '857 litigation: Two parallel cases involving Roku — AlmondNet, Inc. et al. v. Roku, Inc., W.D. Tex. 6:21-cv-00731, and Roku, Inc. v. AlmondNet, Inc. et al., D. Del. 1:21-cv-01035 — both filed July 15, 2021, and both terminated/dismissed on June 20, 2025 (the latter following a stay pending settlement).
• Possible but unverified: AlmondNet, Inc. et al. v. Oracle Corporation, W.D. Tex. 6:24-cv-00303 (dismissed June 20, 2025).
• If you need a definitive, docket-verified list of every case asserting claim(s) of the '857 patent, the reliable next steps are: (a) the complaint and infringement-contentions documents in each AlmondNet-family case, and (b) the PTAB/PACER records — none of which I could fully pull here. I would rather flag this gap than overstate the list.

Sources: Docket Alarm dockets for 6:21-cv-00731 and 1:21-cv-01035; the patent's Google Patents "litigation" links (Unified Patents portal); AO 120 filings at Docket Nos. 29 (W.D. Tex.) and the Delaware docket entries 193/244/252.

I'll verify the "no PTAB activity" default against live sources, since older or newly-filed proceedings may not be in the ODP ingest.

Let me run more targeted searches specifically for PTAB proceedings naming the '857 patent.

Let me run a few final targeted searches to confirm whether any petition names the '857 patent itself.

AIA Trial Proceedings — U.S. Patent No. 10,628,857 B2

Proceedings overview

Total AIA trial proceedings naming U.S. 10,628,857: zero (0). Breakdown: no active proceedings, no claims invalidated, no claims sustained, no settlements, and no institution denials — because no IPR, PGR, or CBM petition has ever been filed against the '857 patent as far as the structured ODP data and my live-search sweep can establish.

The bottom-line defensive posture is therefore the opposite of "hardened" — it is "untested." The '857 patent has never been through an AIA trial, so its claims carry no PTAB-driven narrowing, no cancelled claims, and no § 315(e)(2) estoppel that would strengthen a defendant's hand. What it has is a very large, heavily-litigated sibling family in which many related AlmondNet patents were challenged at the PTAB — sibling activity that supplies useful prior-art context and possibly family-wide prosecution-history admissions (see below), but that says nothing authoritative about the '857's own claim validity.

⚠️ No proceeding numbers are supplied for the '857 patent because none exist in the sources I can reach. I will not invent any. The structured "PTAB proceedings on file" block states the USPTO Open Data Portal returns no AIA trial proceedings for this patent, and my independent web searches (including PTAB E2E/PTACTS filings, Docket Alarm PTAB records, and litigation invalidity contentions) surfaced no petition naming the literal number 10,628,857.

Why there is no '857 proceeding (the actual story)

The '857 patent was asserted in litigation, but the defendant chose the district-court route for this patent rather than the PTAB:

• AlmondNet, Inc. et al. v. Roku, Inc., W.D. Tex. 6:21-cv-00731 (filed 2021-07-15) — the '857 patent is named among the patents-in-suit (a copy of the patent appears as an exhibit to the complaint).
• Roku, Inc. v. AlmondNet, Inc. et al., D. Del. 1:21-cv-01035 (filed 2021-07-15) — Roku sought declaratory judgment of invalidity as to the '857 patent among the "Patents-in-Suit" in its invalidity counterclaims.

In that Delaware action, Roku's invalidity contentions expressly incorporated by reference the grounds, arguments, and expert declarations from a long list of AIA proceedings on other AlmondNet patents — including the '398, '878, '639, '586, '164, '069, '198, and '582 patents — while never filing an IPR of its own against the '857. Both Roku cases were dismissed/terminated on 2025-06-20, which aligns with the absence of any '857 PTAB petition: once the litigation wound down, there was no target and no one-year § 315(b) clock still running that mattered.

Defensive value: A defendant today faces a patent whose claims have never been construed by the PTAB and never been cancelled. That cuts both ways — it means you cannot ride a prior FWD to invalidity, but it also means the patent owner cannot point to a PTAB "survival" as a validity endorsement. All § 102/§ 103/§ 112 theories remain open in district court.

Related-family PTAB context (NOT the '857 patent — offered for estoppel/prior-art orientation only)

Since the '857 shares a specification and lineage with heavily-challenged siblings, these are worth knowing when assembling a defense. I flag them as related-patent context, explicitly not proceedings on the '857:

Caveat: the exact claim-level dispositions of several of these sibling proceedings (which claims cancelled vs. sustained) are not fully verified here, and at least one secondary report describes a 2023-11-30 PTAB decision invalidating "all four challenged claims" of an AlmondNet advertising patent that I could not tie to a specific patent number. I do not attribute any of these outcomes to the '857.

Strategic summary

Which '857 claims are canceled/sustained/untested? — All of them are UNTESTED. There are no PTAB cancellations and no PTAB validity holdings for the '857's claims (independent claims directed to the three-party "tag → recognize → off-site serve" method, the three-contract structure, the computer-program-product claims, and the program-storage-device claims, plus dependents covering "preferred visitor" criteria, tag types, message types, and same-medium variants). Nothing in the AIA record narrows them.

Estoppel landscape — Because no IPR/PGR was ever filed against the '857, § 315(e)(2) estoppel does not attach to any defendant on the '857's claims. There is no petitioner-privy estoppel to worry about either. Two nuances matter:

• Estoppel from the sibling IPRs is patent-specific; a Roku/Meta/Amazon IPR on the '639, '398, or '878 does not estop anyone as to the '857.
• However, Roku's district-court invalidity contentions on the '857 incorporated by reference the grounds and expert declarations from those sibling IPRs. If you are facing assertion today, that incorporation is a roadmap to ready-made art + expert reasoning you can reuse — but you are not bound by any PTAB outcome and the full universe of prior-art grounds remains available.

Pattern signals — No single petitioner ever filed multiple IPRs against the '857 (nobody filed one). The patent owner (AlmondNet) has litigated aggressively — asserting the '857 against Roku and listing it across the Samsung, Meta, and Oracle campaign dockets via § 290 filings — but relied on district court, not the PTAB, for this particular patent. Notably, the Roku campaign ended on 2025-06-20, so the '857's immediate assertion pressure has eased. This is a "certified-but-recently-asserted NPE patent with an untested PTAB profile," and the family's broader pattern (CBM denials in 2017, mixed IPR outcomes 2022–2024, a June 2024 $122M W.D. Tex. verdict on siblings, and an April 2026 Amazon settlement/license) shows the owner both winning and losing at the Board and in court.

Recommended next steps

1. If you are a defendant: There is no FWD to cite and no cancelled claim to lean on. Your invalidity case must be built fresh. Start from the art and expert reasoning that Roku folded into its '857 invalidity contentions via the sibling IPRs (the '639/'398/'878 proceedings) — that material is a legitimate, already-developed springboard, even though it carries no estoppel effect here.
2. An IPR is not precluded. With no prior petition on the '857, there is no § 315(b) bar, no General Plastic follow-on problem, and no § 325(d) "same art previously before the Office" issue unique to this patent — subject only to (a) whether you have been served with a complaint more than one year ago, and (b) the 2025 discretionary-denial regime (the "settled expectations" factor is worth watching: the '857 issued 2020-04-21 and claims priority to 2000-11-28, so its age and 2021-03-20 nominal expiration could invite a discretionary denial on old-patent grounds). Given the patent's listed "Expired – Fee Related" status per Google Patents, confirm the live enforcement posture before investing in a petition.
3. Note the expiration signal. The record lists an adjusted expiration of 2021-03-20 (20 years from 2000-11-28 plus 112 days PTA). If that status is accurate, a defendant's practical exposure window and the cost-benefit of an IPR both change materially. Treat the fee-related characterization as an assumption in the source, not a legal conclusion, and verify.
4. If no PTAB activity matters to your strategy: it does — this is a "clean-slate" patent at the Board. That is itself the signal: a patent asserted since 2021 without ever attracting an IPR tends to reflect either early settlement, a short assertion window, or litigation posture choices rather than demonstrated PTAB resilience.

Verification & uncertainty flags

• Zero is a negative finding. The ODP block reports no AIA proceedings, and my searches found none, but this analysis is limited by the sources I could reach within the search budget. I could not exhaustively query PTAB E2E/PACER dockets or every AlmondNet campaign for a stray '857 petition. I would rather flag the gap than assert absolute absence.
• Sibling-proceeding dispositions are partially unverified, particularly claim-level outcomes; do not treat the table's outcome column as an authoritative FWD summary without checking the underlying papers.
• No proceeding numbers are fabricated; none exist for the '857 patent in the record available to me.
• The patent's expiration/maintenance-fee status is reported as the Google Patents legal-status field shows it and is expressly an assumption in the source.

I'll research the assignment record and corporate context for this patent.

Let me look for the specific assignment reel/frame record and ownership history.

Let me search for the specific reel/frame records and any security interests or later transfers.

Assignment Record Reconstruction — US Patent 10,628,857 B2

Method note / sourcing limitation (stated up front): I was asked to pull reel/frame numbers, execution dates, and correspondents from USPTO Assignment Center (https://assignmentcenter.uspto.gov/) and Assignment Search (https://assignment.uspto.gov/patent/index.html). Both are JavaScript/interactive search interfaces whose result rows are not directly retrievable through the retrieval channel available to me here. I was therefore unable to extract a numeric reel/frame, an execution date separate from the recording date, or a verified list of every recorded conveyance for this patent. Rather than invent reel/frame values (which would be the single most damaging error in an ownership-chain report), I reconstruct below only what is affirmatively grounded in the patent's own front-page/legal-events record and reproducible secondary sources, and I flag each gap. Verification steps are given at the end.

Inventors

Unusual-pattern analysis. The classic "all inventors departed within 12 months" fire-sale precursor is not present — but the reason is structural rather than reassuring: there is only one inventor, and he is the controlling principal of the assignee. He has not departed; he has personally asserted the portfolio. Two related observations are worth recording:

1. Very long gap between conception and this filing. The '857 application (16/033,035) was filed 2018-07-11 as a continuation of a chain beginning with 09/723,391 filed 2000-11-28. Shkedi executed the assignment for the 2018 continuation ~18 years after the priority filing. This is a deliberate continuation-chain strategy (keeping a 2000 priority date alive across re-filed claims), not a change of control.
2. Sole-inventor-and-owner concentration means there is no inventor-signature problem for standing, and no co-owner to fracture title. Title is clean and singular.

Original assignee

Assignment timeline

Bottom line, stated plainly: The only conveyance I can affirmatively identify for the '857 patent is the original inventor→AlmondNet assignment around the 2018-07-11 filing date. I found no recorded post-issuance assignment, no security agreement, no merger, no change of name, and no transfer to any third-party NPE. The patent's Google Patents legal-events ledger carries a single reassignment entry and no subsequent owner change.

Because I could not read the Assignment Center result rows, I cannot supply the reel/frame number, and I am not going to guess one. The one entry is recorded in the structure requested, with explicit placeholder flags instead of fabricated values:

1. (Execution date not independently retrieved — presumed on or about filing) / recorded 2018-07-11 — Reel NNNNNN/NNNN (NOT RETRIEVED — see verification steps)

• Conveyance: Assignment (Assignment of Assignors' Interest; recorded as "reassignment")
• Assignor: Shkedi, Roy, Mr. (sole inventor)
• Assignee: ALMONDNET, INC. (Delaware corporation)
• Correspondent: Louis J. Hoffman, Reg. No. 38,918, Louis J. Hoffman, P.C. (customer number 26362). Recorded addresses across this family, chronologically: 11811 North Tatum Blvd., Suite 2100, Phoenix, AZ 85028 → 7689 East Paradise Lane, Suite 2, Scottsdale, AZ 85260 → 14301 North 87th Street, Suite 312, Scottsdale, AZ 85260. Filing emails on record: Louis@valuablepatents.com, donald@valuablepatents.com.
  • Recurrence flag — PRESENT but benign in context. Hoffman is the repeat correspondent across the entire AlmondNet family, not just this patent — he appears as attorney of record on the parent '639 patent maintenance-fee filings, on the '745 and '582 family, and on prosecution/interview papers for the '762 and '139 lineages. This is the same-correspondent signal the task asks me to capture, but read it correctly: Hoffman is AlmondNet's prosecution and portfolio counsel of record, not a shell-entity recording agent. The recurrence shows one lawyer running a large original-owner portfolio, not a lawyer stitching together anonymous LLCs. This distinction materially weakens the signal.
• Context: Original assignment by the founder-inventor to his own company — formation/ownership perfection of an internally generated portfolio, not an acquisition or fire-sale.

Family-level parallel assignments (not records for this patent, flagged for completeness)

Each application in the continuation chain is a distinct property and would ordinarily have carried its own Shkedi→AlmondNet assignment: 09/723,391 (2000) → 10/996,491 (2004) → 12/911,601 (2010) → 13/368,686 (2012) → 13/585,765 (2012) → 14/091,285 (2013) → 16/033,035 (2018). I did not verify the reel/frames for any of these. Do not treat them as confirmed records for the '857 patent.

If Assignment Center in fact shows no record beyond the original 2018 entry, that absence is itself a finding — it means the original assignee still owns the patent outright, which is consistent with everything else observed (AlmondNet asserts its own patents in its own name).

Timeline diagram

timeline
    title Ownership of US 10628857
    1998 : AlmondNet founded by Roy Shkedi
    2000 : Priority application filed by Shkedi
    2018 : Continuation filed as 16033035
         : Assignment recorded to AlmondNet Inc
    2020 : Patent issued
    2021 : Roku suits assert the patent
    2025 : Roku cases dismissed

NPE / troll-pattern signals

Additional fact worth recording: AlmondNet has been on the receiving end of an NPE counter-allegation dynamic — in AlmondNet v. Viant Technology, 1:23-cv-00174 (D. Del.), Viant characterized AlmondNet, Intent IQ, and Datonics collectively as "the AlmondNet Group" and noted the three entities are managed by the same individual, who is also the inventor of the patents AlmondNet asserts, with shared litigation conduct. That is direct third-party evidence of the single-principal, group-wide assertion structure, and it is the kind of concrete corroboration (not naming inference) the task asks for.

Verdict

NPE — moderate confidence.

The ownership chain contains no shell-entity transfer, no cascading LLC assignments, no third-party NPE on any public list, and no defensive-aggregator terminus — signals 1, 4, 5, 6, 7 and 8 are all not present. What drives the "NPE" call is a single strong signal rather than a pattern: the assignee, AlmondNet, Inc. (Delaware), expressly describes itself as an IP licensing entity ("focused on R&D and the licensing of its extensive portfolio of enabling technology and Intellectual Property"), I found no evidence it ships a product embodying the '857 claims, and it is a serial, high-frequency plaintiff (Unified Patents surfacing; suits against Roku, Amazon, Samsung, Microsoft, Meta, Oracle, Oath/Yahoo, Viant, LinkedIn, and others; a $122M 2024 jury verdict on sibling patents). That combination — licensing-only monetization of the patent itself plus repeated assertion in the owner's own name — is the substance of a PAE, even though the assignment archaeology is unusually clean. I stop short of "high confidence" because the target is a genuine, still-operating 1998-founded company with real product subsidiaries (Datonics, Intent IQ) and its own R&D arm, not an anonymous assertion shell.

Verify here: https://assignmentcenter.uspto.gov/ (search patent 10,628,857 or application 16/033,035) and https://assignment.uspto.gov/patent/index.html — retrieve the reel/frame for the 2018-07-11 Shkedi→AlmondNet recording and confirm whether any post-issuance conveyance, security interest, or name change exists that I could not display. Also check PEDS/ODP for application 16/033,035 and the AlmondNet Group entities' state registrations (AlmondNet, Inc. — DE; Intent IQ, LLC; Datonics, LLC) to confirm the shared-principal structure.

I'll search for the specific prior-art citations listed on US 10,628,857 (the "Patent Citations" / "References Cited" section), keeping the number literal.

Let me pull the patent's own "References Cited" (Patent Citations) section directly.

Let me verify the exact "Patent Citations" list on the '857 patent itself and confirm each reference.

Prior-Art Analysis — U.S. Patent No. 10,628,857 B2

Strict-identifier note: I interpret "10628857" literally as U.S. Patent No. 10,628,857 B2, "Methods and apparatus for facilitated off-site targeted internet advertising," Roy Shkedi, AlmondNet, Inc., application 16/033,035, filed 2018-07-11, granted 2020-04-21, earliest priority 2000-11-28. I did not substitute any similar number.

1. Method and source limitations (state these before the analysis)

Two honest caveats, one of which is a contradiction-flag against the earlier sections:

1. The literal claim set is still not in hand. The full text supplied to me (and the previously generated Patent-summary section) truncates before the claims. I therefore cannot cite claim numbers literally. The § 102 mapping below is by claim family (independent method claim; three-agreement contracting-structure claim; computer-program-product claims; program-storage-device claims), which the '857 specification's Summary-of-Invention reproduces almost verbatim. This is consistent with — not contradictory to — the earlier flag that "claim text not verified."
2. I could not cleanly isolate the '857's own printed face-page citation list. I did not have direct USPTO PatentCenter/Patent Full-Text access; I worked from Google Patents, Patexia, the Unified Patents portal, and USPTO PTAB exhibits surfaced by search. The '857's family is enormous and sibling patents in it (e.g., US 7,822,639; US 10,026,100; US 11,093,970; US 11,564,015) print large, overlapping "References Cited" lists. I flag explicitly which list I am attributing to the '857 and why.

2. The "Patent Citations" set associated with the '857 family

The Google Patents page surfaced for this family shows a "Patent Citations (7)" block whose rows are the following (source: https://patents.google.com/patent/US20120158506 — the family page whose cross-reference table carries the row "2018-07-11 | US | US16/033,035 | patent/US10628857B2/en | not_active Expired - Fee Related"):

Why row 7 is a red flag, not a prior-art reference: US 7,822,639 (application 10/996,491) is the '857's own ancestor — the '857's cross-reference section states it "is a continuation of application Ser. No. 10/996,491 … now U.S. Pat. No. 7,822,639". A same-inventor/same-family patent appearing in a "citations" block is a related-application or double-patenting citation, not § 102 prior art against the '857. Its presence is one reason I suspect the 7-item block is a family-level citation aggregation rather than the '857's isolated face list. I therefore present it, flagged, rather than treating it as anticipatory art.

3. The larger "References Cited" list actually printed in the issued '857 document

The issued-document images (US10628857.pdf, US010628857B2.pdf) returned in search show a long U.S.-patent-documents list, whose visible entries include: US 5,220,655 (Tsutsui, 6/1993); US 5,359,508 (Rossides, 10/1994); US 5,442,633 (Perkins/Peckins, 8/1995); US 5,710,884 (Dedrick); US 5,724,424 (Gifford); US 5,751,956 (Kirsch); US 5,774,170 (Hite); US 5,774,670 (Montulli); US 5,790,785 (Klug); US 5,794,210 (Goldhaber); US 5,796,952 (Davis); US 5,801,702 (Dolan); US 5,812,769 (Graber); US 5,819,092 (Ferguson); US 5,819,285 (Emmoto); US 5,835,735 (Bleveux); US 5,848,396 (Gerace); US 5,855,008 (Goldhaber); US 5,857,175 (Kirsch); US 5,870,656 (Wasinger); US 5,870,550 (Bull); US 5,901,287; US 5,918,014 (Angles); US 5,933,811 (Angles); US 5,953,915 (Kirsch); US 5,991,735 (Gerace); US 5,991,740 (Messer); US 5,996,007 (Klug); US 5,999,912 … (and continuing).

Important attribution warning: this same long list appears in sibling Shkedi/Shkedi-family documents (e.g., the US 11,093,970 and US 11,564,015 "References Cited"), and my OCR sources jumble columns (e.g., a row reading "5,901,287 A 5/1999 Robinson" is column-shifted). I therefore cannot certify that every item in this long list is on the '857's own face page, and I am not going to build § 102 conclusions on the mis-OCR'd rows. What I can say with confidence: the '857's document carries a large set of 1993–1999 U.S. patent references, which — given the 2000-11-28 priority — are overwhelmingly pre-AIA § 102(a)/(b) art.

4. § 102 framework that applies

Because the '857's earliest non-provisional filing is 2000-11-28, the pre-AIA statute governs:

• References published/issued more than one year before 2000-11-28 (i.e., before ~1999-11-28) → § 102(b) art: US 5,774,170 (Hite); US 5,991,740 (Messer); US 5,774,670 (Montulli); US 5,794,210 / 5,855,008 (Goldhaber); US 5,848,396 / 5,991,735 (Gerace); US 5,933,811 (Angles); US 5,948,061 (Merriman); etc.
• References published within one year of, but before, 2000-11-28 → § 102(a) art: US 6,055,573 (2000-04-25) and WO 2000/016529 (2000-03-23).
• References whose publication post-dates 2000-11-28 but whose application pre-dates it → § 102(e) art only: US 6,285,987 (1997-01-22 priority) and US 2002/0133412 (Oliver) — the latter only if its underlying application was filed pre-2000-11-28, which I could not confirm.
• US 7,822,639 (AlmondNet's own '639) → not § 102 art (common ownership/same family; it is the '857's parent).

5. Reference-by-reference: description and potential § 102 relevance

None of these seven references, alone, discloses the '857's three-party core — (a) a facilitator/"agency" that facilitates visitor identification; (b) a first broadcaster tagging a preponderance of ITS visitors; and (c) a second broadcaster recognizing that tag and accepting an off-site content presentation for the recognized visitor. That three-part tagging-and-recognition linkage is the claimed point of novelty, and it is not present in the cited art. The references below are therefore best characterized as § 102(a)/(b) art against individual limitations, and § 103 combination art — not clean § 102 anticipation of the independent claims.

6. Which claims are potentially implicated (family-level, not literal claim numbers)

• Independent method claim (agency-facilitates-identification → first broadcaster tags visitors → second broadcaster recognizes tag and accepts off-site presentation). No cited reference discloses this three-actor tag/recognize linkage. No § 102 anticipation. Any challenge would have to be a § 103 combination (e.g., Hite/Adwise for off-site targeted ads + Messer/′573 for visitor tracking and identification + Montulli (US 5,774,670, in the long list) for the cookie tag element).
• Contracting-structure claims (three interlocking agreements + payments). Relevant references are US 2002/0133412 (Oliver) and US 6,285,987 (Engage) as business/transaction background. No § 102 anticipation; the specific "agency pays first broadcaster per placement / second broadcaster per acceptance, first broadcaster tags its visitors" structure is not shown.
• Computer-program-product claims and program-storage-device claims. Same analysis; these are drafted as the method/contract structure embodied in code, so they rise and fall with the method claims. No cited reference anticipates them.

7. Bottom line

• No single cited reference anticipates the '857's independent claims under 35 U.S.C. § 102. The claimed three-party "first site tags its visitors → second site recognizes the tag → off-site content presented to the recognized visitor" arrangement is not disclosed by any of the seven items in the family's "Patent Citations" block, nor by the specific long-list references I could verify.
• The most relevant prior art in the set, for a § 103 analysis, is: WO 2000/016529 (Adwise) — off-site/external content injected into another site's session; US 5,991,740 (Messer) — cross-network user tracking; US 6,055,573 (Supermarkets Online) — identifying a particular consumer and communicating accordingly; US 5,774,170 (Hite) — targeted ad delivery; with US 5,774,670 (Montulli) — cookies (from the long list) — supplying the tag mechanism.
• US 7,822,639 is not prior art — it is the '857's own family parent and should be excluded from any § 102/§ 103 list.

Gaps I did not close (stated rather than fabricated): (i) the literal claim numbers, because the claim set is not in my source; (ii) a certified, row-accurate reproduction of the '857's printed face-page "References Cited" (my OCR sources column-shifted several rows); and (iii) the filing date of US 2002/0133412, needed to confirm its § 102(e) status. To close all three, the authoritative source is the issued PDF at https://patentimages.storage.googleapis.com/f7/21/68/a85be386687bc7/US10628857.pdf and the file wrapper in USPTO PatentCenter for application 16/033,035.

Sources: Google Patents family page showing the "Patent Citations (7)" block (https://patents.google.com/patent/US20120158506); issued-document images US10628857.pdf / US010628857B2.pdf (patentimages.storage.googleapis.com); Patexia record (https://patexia.com/us/patent/10628857); Unified Patents portal (https://portal.unifiedpatents.com/patents/patent/US-10628857-B2); USPTO PTAB exhibit compilations (ptacts.uspto.gov) showing the '857 in AlmondNet/Roku contentions. USPTO PatentCenter/Patent Full-Text was not directly queried.

I'll ground this analysis by first checking for the actual prior-art record and any PTAB/§ 103 history for this patent and its family, then build the obviousness grounds.

The searches surfaced a "Prior Art" section specifically for this patent on the Unified Patents portal. Let me pull that list, since the task directs me to use it.

Obviousness Analysis — U.S. Patent No. 10,628,857 B2 under 35 U.S.C. § 103

Analyst note on the "Prior Art section." The prior-art list I was directed to use is the one carried on this patent's own record page (the Unified Patents portal entry for US‑10628857‑B2, mirrored in the PubChem patent entry and the patent's "References Cited"/NPL listing). I retrieved the leading entries and paired them with the U.S. patent references that appear in the same family's invalidity contentions (Roku, Samsung, Meta, Amazon), which are the closest thing to an on-point art set for this claim family. Where I could not verify something, I say so.

Two contradictions / caveats to flag up front:

1. Claim text contradiction carried forward. The previously generated Patent summary flagged that the literal claim language was never retrieved. That is still true. The grounds below are built against the reconstructed claim scope (the specification's four enumerated claim families), not against verbatim claim text. Any § 103 conclusion here is provisional on the actual issued claims.
2. Priority-date discrepancy in the record itself. Google Patents (per the prior section) states a priority date of 2000‑11‑28; the Unified Patents portal entry for this patent states 2000‑11‑27, and the document I retrieved (US‑6014638‑A page) also lists 2000‑11‑27. The patent's own cross-reference section says application 09/723,391 was filed Nov. 28, 2000. This one-day discrepancy does not change the art set (all cited references pre-date both dates), but it is a live inconsistency in the public record.
3. Minor: the task header says "April 26, 2026" while the operating environment says 2026‑09‑28. Not material to the § 103 analysis; noted for completeness.

1. Legal framework applied

Obviousness is assessed under the Graham v. John Deere factors — scope/content of the prior art; differences between the art and the claims; level of ordinary skill; and secondary considerations — as refined by KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007). Under KSR, a motivation to combine need not be found expressly in a reference; it may come from the nature of the problem, market forces, design incentives, and "common sense," and a combination of familiar elements according to known methods is obvious when it yields no more than predictable results. It is also settled that disclosure that anticipates also renders obvious (Realtime Data, LLC v. Iancu, 912 F.3d 1368, 1373 (Fed. Cir. 2019) — the point Amazon's counsel argued in the parallel litigation).

2. Level of ordinary skill in the art (POSITA)

For a November 2000 priority date, a POSITA is reasonably defined as a person with a bachelor's degree in computer science/electrical engineering (or equivalent) and 2–3 years of experience in networked advertising systems, or a person with a business/marketing background and equivalent hands-on experience building or buying Internet ad-serving campaigns. This is the type of artisan the family's own litigation contentions used (e.g., Stephen Gray's declaration in CBM2017‑00046 describes a 1990s e‑commerce CTO). The precise definition may have been disputed in the Roku/Samsung cases; I did not verify a court-adopted construction.

3. Claim scope at issue

As reconstructed from the specification (families A–D):

• A (method): (a) agency facilitates visitor identification; (b) first broadcaster tags a preponderance of its visitors; (c) second broadcaster recognizes the tagged visitor and, by proxy (with the agency or first broadcaster), accepts the off-site content presentation.
• B (contracting structure): three interlocking agreements — agency↔content provider (delivery + payment); agency↔first broadcaster (tagging + off-site placement + payment); agency↔second broadcaster (recognition + acceptance + payment).
• C (computer program product) and D (program storage device): the method / contracting structure expressed as code on a medium.

The commercially salient limitation that separates the claims from generic ad networks is the direction of the sale: the first site sells exposures to its own audience off-site (an "out-of-context"/"super-saturation" placement), while the second site merely rents otherwise-unsold inventory.

4. Prior art references (from the patent's stated Prior Art section + family contentions)

5. Grounds of rejection

Ground 1 — Merriman (PA‑1) in view of Montulli (PA‑2); optionally with AdKnowledge (PA‑5)

• [b] Tagging: Merriman's ad server writes/reads a cookie on the visitor's browser at an affiliate page; Montulli supplies the persistent-client-state mechanism by which the browser returns that message on later requests. This is the "tag" element.
• [c] Recognition + off-site placement: Merriman expressly discloses a third-party ad server that, "in response to requests from affiliated sites," selects and transmits an advertisement to the user based on prior profiling — i.e., the visitor is recognized at a different site from where the profile was built, and the ad is served there. Montulli supplies recognition at the second site. The "by proxy — in conjunction with the agency or the first broadcaster" limitation is met by Merriman's architecture (the ad server acts for the affiliates and advertisers).
• The gap: Merriman's business framing is affiliate/publisher-owned space, not "first site sells its audience off-site." That gap is closed by Ground 2 or by the motivation analysis (§6).
• Motivation: Merriman itself depends on cookie technology, so the combination is a use of a known element (Montulli) for its established function — a classic KSR "known technique / predictable result" rationale.

Ground 2 — Engage AudienceNet / Engage S‑1A (PA‑4) in view of Merriman (PA‑1) + Montulli (PA‑2)

This is the strongest ground and is the theory most likely to succeed, because Amazon actually ran the "Engage System" successfully against a sibling claim of the related '139 patent (Dr. Hanson's report: "In my opinion, the Engage System alone, in view of the knowledge of a POSITA… discloses each and every element of claims 37, 42, 43, 45, and 48 of the '139 Patent"; W.D. Tex. 6:21‑cv‑00898, Dkt. 259). The '857 and '139 share the same specification lineage, so the same art is highly probative here.

• Engage discloses cross-Web, profile-based delivery where the value of the inventory "is based on the visitor, not on the content of the Web page" — i.e., an ad shown on site B because of who the visitor is (a known visitor of an earlier site), which is precisely element [c].
• Engage's own marketing fills the first-site-sells-its-audience-off-site requirement that Merriman lacks: publishers "augment their existing sales of premium content with the sale of underutilized inventory at higher CPM rates," with media buyers "connect[ing] directly with the target audience" across the Web.
• Merriman + Montulli supply the concrete server/cookie implementation of recognition and delivery across affiliated sites.

Ground 3 — Kohda & Endo (PA‑3) in view of Engage (PA‑4) and Montulli (PA‑2)

• Kohda discloses an "advertising agent" that merges ads from advertisers' servers into pages served by ordinary, unrelated servers, so "users see advertisements on any server around on the Internet," and the agent "is aware of the identity of the user and which page the user is about to read." That reads on the third-party, off-site placement element.
• Combine with Engage for profile-based selection, and Montulli for cookie-based recognition of the returning visitor at the second site.
• Kohda is a 1996 reference, predating the 2000 priority by four years, and it was treated as anticipatory art by examiners in this family's cousin patents (see the inequitable-conduct pleading in E.D. Tex. 2:13‑cv‑00893, where the Examiner "found the majority of the pending claims… anticipated by Kohda"). Its corroborated materiality cuts strongly in favor of § 103.

Ground 4 — Rosenberg (PA‑7) in view of Merriman (PA‑1); optionally + Angles

• Rosenberg‑style systems use a central ad server that reads a tag/cookie, retrieves stored viewer information, and delivers targeted ads to that visitor on affiliated sites. Combined with Merriman's cross-site cookie recognition, this supplies every technical step. (Trial counsel for Amazon argued exactly this "Rosenberg‑Merriman / spotlight‑tag" combination in IPR2022‑01453 against the sibling '639 patent.)

Ground 5 — The contracting-structure claims (Family B): Grounds 1–4 + AdKnowledge (PA‑5) / DoubleClick insertion orders / Engage S‑1A

The Family B claims add only conventional commercial architecture (offer, acceptance, and payment among agency, content provider, and two broadcasters). Merriman itself recites that affiliates "for a fee contract with the entity providing the advertisement server"; Engage's S‑1A describes network participation agreements; AdKnowledge's Campaign Manager operated as an ad-serving/management bureau. Once the technical method is obvious (Grounds 1–4), implementing it through standard insertion-order/agency contracts is a predictable business arrangement with no patentable technical contribution. A POSITA‑with-business-acumen would find the three‑contract structure obvious as the natural commercial packaging of an ad‑network/agency relationship.

6. Motivation to combine (centralized rationale)

Under KSR, the following are independent, sufficient motivations:

1. Same field, same problem. All references address delivering targeted advertisements to web visitors; the '857 claims solve ad-inventory saturation — the exact market problem Engage (PA‑4) was marketing against in 1999 ("underutilized inventory at higher CPM rates").
2. Known technique for its known function. Cookies (Montulli) were already used to "remember users… across multiple sites" (PA‑6). Applying cookie-based recognition to off-site ad delivery is a use of a known element for its established purpose (KSR).
3. Predictable variation of a network model. Merriman's affiliate ad network and Engage's web‑wide network are the same architecture; moving from "network serves ads on affiliated sites" to "a first site contracts to have its audience served off-site" is a predictable commercial variation, not a new technical result.
4. Market/design incentive. Publishers had sold-out premium inventory and unmet advertiser demand (the patent's own Background admits this state of the art); ad‑network infrastructure to satisfy it already existed. Competitive pressure — "no one wanted to pay a premium [for Engage]… the world of search educated the market" (MediaPost quote) — supplies the incentive to repackage the same mechanism.
5. Inherent disclosure quality. Because Engage and Kohda arguably anticipate individual elements, they render the claims obvious even where they do not literally anticipate (Realtime Data).

7. Dependent claims

The dependent-claim variants (specific tag types — cookie, SIM‑card message, credit-card warning, legal instrument, telephone/wireless database notification; "preferred visitor" by time or money spent; content type — banner, audio, multimedia; same-medium implementation) are each a substitution of one known data store or message format for another with no unexpected result. A POSITA would recognize that a cookie, a SIM‑card record, or a telephone-system DB entry are interchangeable recognition vehicles. These are the paradigm of obvious, predictable variations.

8. Rebuttal — potential secondary considerations

A patent owner would point to:

• Commercial success / licensing: the 2013 Google license and the $122M jury verdict against Amazon (June 2024) on sibling patents 7,822,639 and 8,671,139. ⚠️ Nexus is weak — the verdict is on different patents (the '639 and '139), so it cannot be presumed to reflect the '857's claims without a proven nexus to the same inventive contribution.
• Copying / industry recognition: presumes the art was known and copied, but the art set above shows DoubleClick/Engage were the industry's own earlier systems.
• Long-felt need: undercut by Engage's 1999 launch expressly targeting the same "underutilized inventory" problem.

Net: the objective indicia are unlikely to outweigh the close relationship between the art and the claims unless a specific nexus to the '857 claims is established.

9. Additional overlay (not § 103, but relevant)

The family has been attacked as abstract under § 101 — Yahoo's CBM2017‑00046 expert opined that the '639 claims "correspond to the abstract idea of securing revenue from advertising." The '857's Family B (contracting-structure) claims are especially exposed on this axis. This does not affect the § 103 analysis but explains why the validity challenge likely focused on the technical method claims.

10. Conclusion

• Most likely-to-succeed ground: Engage AudienceNet / Engage S‑1A (PA‑4) + Merriman (PA‑1) + Montulli (PA‑2) — a web-wide, profile-based ad network that values inventory "based on the visitor, not the content," implemented with cookie recognition across sites. This combination teaches every element of reconstruction Family A and, by extension, Families C/D, and (with AdKnowledge/insertion-order practice) Family B. Amazon's success using Engage against a sibling claim materially increases confidence here.
• Strong secondary grounds: Merriman + Montulli (+ AdKnowledge); Kohda & Endo + Engage + Montulli; Rosenberg + Merriman.
• Confidence: Moderate. High confidence that the technical method claims are obvious over Engage/Merriman/Montulli. Low-to-moderate confidence on the literal claim limitations and on any claim element I could not read verbatim, because the issued claim text was never retrieved (see caveat 1).

Sources

• Patent's own record/Prior Art section: https://portal.unifiedpatents.com/patents/patent/US-10628857-B2 ; NPL listing at https://pubchem.ncbi.nlm.nih.gov/patent/US-10628857-B2
• Merriman: US 5,948,061 and reissues RE44,724 / RE44,899 (DoubleClick/Google)
• Kohda & Endo (1996), as charted in E.D. Tex. 2:13‑cv‑00893 (cases.justia.com)
• Engage AudienceNet: ClickZ, Oct. 13, 1999, https://clickz.com/engage-launches-profile-based-web-wide-ad-network/68556/ ; Engage S‑1A (Jul. 19, 1999)
• Amazon's Engage obviousness theory: W.D. Tex. 6:21‑cv‑00898, Dkt. 259
• Family invalidity contentions: Roku (6:21‑cv‑00731) and Meta (6:21‑cv‑00896) Exhibit G reference tables (Docket Alarm)
• CBM2017‑00046, Declaration of Stephen Gray (the '639 abstract-idea/state-of-the-art record)

Recommendation: Before relying on this analysis, obtain the issued claim set of the '857 and the Roku/Meta § 103 claim charts for the '857 specifically so the element-by-element mapping can be verified against literal claim language rather than reconstructed scope.