US 7076431

I'll search for authoritative information on US 7076431.

I have solid bibliographic data. Now let me check for any 2026 CAFC docket activity.

Let me verify recent (2025–2026) docket activity and the post-IPR status of the '431 patent.

US 7,076,431 B2 — Analyst Summary

Scope note: I searched the public patent/litigation record (Google Patents, USPTO PTAB/PTACTS filings, Docket Alarm, CourtListener/CAFC opinions, and docket-derived documents). I did not have direct live PACER or USPTO PatentCenter access in this session, and I found no 2026 Court of Appeals for the Federal Circuit docket activity specific to 7076431. The most recent Federal Circuit activity I can confirm is a 2023 decision (below). I flag uncertainties explicitly.

---

1. Bibliographic data

Assignment history (from the Google Patents legal-events record): Original assignee Parus Holdings Inc.; assignment recorded to Parus Holdings, Inc. (July 7, 2010, assignor Kurganov); security agreement to WWIS, LLC (Mar. 18, 2013); assignment to Webley Systems, Inc. (Oct. 9, 2014); assignment back to Parus Holdings, Inc. (Oct. 9, 2014). Note: the "original assignee" field and the 2010 record are not perfectly consistent regarding pre-2010 ownership; treat the chain as approximately as stated rather than definitive.

---

2. Abstract (verbatim)

"The present invention relates to a system for acquiring information from sources on a network, such as the Internet. A voice browsing system maintains a database containing a list of information sources, such as web sites, connected to a network. Each of the information sources is assigned a rank number which is listed in the database along with the record for the information source. In response to a speech command received from a user, a network interface system accesses the information source with the highest rank number in order to retrieve information requested by the user."

---

3. Plain-language overview of the independent claims

The patent has two independent claims — claim 1 (system) and claim 18 (method). The claim text in my source was truncated mid-way through claim 18, so the description of claim 18 below is based on its preamble and the structure of the dependent claims; verify against the printed patent.

Claim 1 — System (independent)

A voice-web-browsing system with these pieces:

• A computer connected to the internet (claim 2 narrows this to the Internet; claim 3 to a LAN).
• A voice-enabled device connected to the computer that receives the user's speech commands (claim 4 lists telephones, IP phones, cellular phones, PDAs, PCs, DVD/TV/CD/MP3 players, etc.).
• A speaker-independent speech-recognition device (claims 5–6: analyzes phonemes; recognizes naturally spoken commands).
• A speech-synthesis device to speak results back.
• An "instruction set" for identifying the information to be retrieved, which contains a plurality of pre-selected web site addresses (claim 7 adds a per-URL content descriptor that pre-defines which portion of the page holds the answer).
• At least one recognition grammar matched to each instruction set and to a speech command; the command is a user-selectable information request. The recognizer picks the matching grammar; the computer then loads the corresponding instruction set.
• Sequential fallback: the computer first hits the first web site in the set; if the information isn't there, it walks through the remaining sites in order until the information is found or all sites are exhausted.
• Finally, the synthesizer produces and transmits an audio message containing the retrieved information.

Notable dependents: claim 8 (computer periodically polls the sites without user instruction to measure availability, response time, and changes in where the information lives, and creates a ranking); claim 9 (periodically searches the network to find and add new web sites); claims 10–13 (per-record rank field, criteria for modifying rank, weighting of criteria, and access in rank order highest-first); claims 14/16/17 (database storage of retrieved info, grammars, and instruction sets; polling of newly discovered sites); claim 15 (re-ranking so the site that actually has the information moves to the top).

Claim 18 — Method (independent)

The method counterpart of claim 1. It recites the steps of: providing a computer operatively connected to the internet and to at least one speaker-independent speech-recognition engine and at least one speech-synthesis engine; providing a voice-enabled device to receive user speech commands; providing at least one instruction set stored in a database containing a plurality of pre-selected web site addresses; and providing a speech command to the speech-recognition engine — followed (per the balance of the claim, which my source truncated) by the grammar selection, sequential site-access, and audio-response steps mirrored from claim 1. Claims 19–32 are dependent claims.

Key concept threading both independent claims: (a) speaker-independent, natural-speech input; (b) an instruction set holding multiple pre-selected URLs for one information category; (c) sequential fail-over across those URLs in a defined order; and (d) spoken output.

---

4. Litigation / validity posture (important caveats)

• IPR2020-00686 (Apple Inc. v. Parus Holdings) — Final Written Decision (Sept. 9, 2021) determined no challenged claims unpatentable (Apple lost). Listed appeal: 22-1143.
• IPR2020-00846 (Google LLC, Samsung, LG v. Parus Holdings) — Final Written Decision (Oct. 19, 2021) held claims 1, 2, 4–7, 9, 10, 13, and 14 of the '431 patent unpatentable as obvious (based on Kovatch and Kurganov-262, among others).
• CAFC Nos. 22-1269 / 22-1270 — Parus Holdings, Inc. v. Google LLC, decided June 12, 2023 (70 F.4th 1365): the Federal Circuit affirmed the Board, holding the '431 claims unpatentable. Parus's antedating evidence was forfeited for violating 37 C.F.R. § 42.6(a)(3), and the Board's written-description finding (that "configured to periodically search … to identify new web sites and to add the new web sites" lacked support in the priority application) was upheld on substantial evidence.
• District court assertions: Parus Holdings Inc. v. Apple (6:19‑cv‑00432), v. Google (6:19‑cv‑00433), v. Samsung (6:19‑cv‑00438), v. LG (6:19‑cv‑00437), v. Amazon.com (6:19‑cv‑00454), all W.D. Tex.; a related N.D. Cal. case (3:20‑cv‑05896). There is also a 2025–2026 fee/malpractice dispute between Parus and its former counsel Mintz (attaching the '431 IPR materials; papers filed in 1:25‑cv‑01722 on Oct. 24, 2025, plus an April 2026 Massachusetts decision) — that is a contract/malpractice matter, not a CAFC patent appeal.

Apparent tension to note: Google Patents still reports the '431 patent as "Expired – Lifetime" with an anticipated expiration of Feb. 5, 2021, while the PTAB/CAFC found several claims unpatentable. The two coexist because patent term expiry and claim cancellation are separate events. I could not confirm from the sources retrieved whether a certificate cancelling claims 1, 2, 4–7, 9, 10, 13, and 14 has issued in the USPTO file — treat that point as unverified.

---

5. Family context

The '431 is one link in a long Parus continuation chain off prov. 60/180,344 (Feb. 4, 2000) and prov. 60/233,068 (Sept. 15, 2000): 09/776,996 (now 6,721,705) → 10/821,690 (7,076,431) → 11/409,703 (now 7,386,455) → 12/030,556 (now 7,881,941) → 12/973,475 (now 8,185,402) → 13/462,819 (now 9,451,084) → 15/269,776 (now 10,096,320) → 15/436,377 (now 10,629,206) → 16/155,523 (pub. 2019/0043505).

Citation sources: https://patents.google.com/patent/US7076431/en ; https://patentimages.storage.googleapis.com/55/06/98/6ca408f46c7e44/US7076431.pdf ; https://www.courtlistener.com/opinion/[9405771](/patent/9405771)/parus-holdings-inc-v-google-llc/ (CAFC 22-1269, June 12, 2023); https://www.docketalarm.com/cases/PTAB/IPR2020-00686/Inter_Partes_Review_of_U.S._Pat._7076431/ ; https://portal.unifiedpatents.com/ptab/case/IPR2020-00846

Uncertainties: (i) the full text of method claim 18 and claims 19–32; (ii) whether formal claim cancellation has been certified; (iii) whether any 2026 CAFC activity exists — none was found, but a PACER-native check would be required to be certain.

I'll research litigation involving US patent 7076431. Let me run several targeted searches.

Let me gather more detail on specific case outcomes and any additional defendants.

Let me search for the later rounds of Parus litigation and their resolutions.

Litigation Involving U.S. Patent No. 7,076,431 ("the '431 patent")

Important scope note: I searched for litigation tied to the specific patent number 7,076,431. That patent ("Robust voice browser system and voice activated device controller") is owned by Parus Holdings, Inc. Note that Parus later asserted a different family of related patents (the '705, '941, '402, '190, '314, '455 patents) in its 2021–2023 filings against Google, Apple, Microsoft, Samsung, Sony, and others — those cases do NOT involve the '431 patent, so I have excluded them from the list below (but flag them so you don't conflate them).

Also note the Google Patents page states the patent "Expired – Lifetime" (anticipated expiration Feb. 5, 2021).

---

A. District Court Infringement Actions (all asserting the '431 patent)

In July–August 2019, Parus filed five suits in the [U.S. District Court for the Western District of Texas, Waco Division](/courts/western-district-of-texas-waco) (Judge Alan D. Albright). Each complaint attached the '431 patent as Exhibit 1 and U.S. Patent No. 9,451,084 ("the '084 patent") as Exhibit 2.

Consolidation: On December 20, 2019, the district court consolidated the five cases into 6:19-cv-00432-ADA as the "Lead Case."

Case-number discrepancy (interpreted literally): Some sources list the Amazon case as 6:19-cv-0484-ADA (Parus's own later complaint, the Mintz malpractice pleading) while court-record aggregators and the Google Patents page list 6:19-cv-00454-ADA. The latter appears in the majority of sources; I report both literally rather than auto-correcting.

B. Refiled/Relocated LG Action

This is the LG dispute that moved out of the Texas consolidation; the complaint again attached the '431 patent (Exhibit 1) and the '084 patent (Exhibit 2). LG filed a Rule 12(c) motion for judgment on the pleadings of invalidity under 35 U.S.C. § 101 (October 15, 2020).

---

C. PTAB Inter Partes Reviews of the '431 Patent

The Board found the challenged claims obvious over prior art including Kovatch (WO 2001/050453) and Kurganov-262 (U.S. Pub. No. 2001/0047262), after rejecting Parus's attempt to antedate Kovatch.

---

D. Federal Circuit Appeals

Holding: Affirmed. The court held (1) that the Board properly refused to consider Parus's antedating arguments/evidence because Parus improperly incorporated them by reference in violation of 37 C.F.R. § 42.6(a)(3); (2) that a patent owner attempting to antedate prior art assumes a burden of production that "cannot be met simply by throwing mountains of evidence at the Board"; and (3) that 35 U.S.C. § 311(b) does not limit the issues (e.g., written description) the Board may consider. The net effect is that the '431 patent claims were held unpatentable and that holding is final.

(Note: the Google Patents page also lists a CAFC case 22-1143 and states "First worldwide family litigation filed." I could not, within this search, independently confirm what 22-1143 involves; I flag it as unverified rather than assume.)

---

E. Related Legal Malpractice Action (arising from the '431 IPR)

• Parus Holdings, Inc. v. Mintz, Levin, Cohn, Ferris, Glovsky and Popeo, P.C. et al. — Parus sued its former IP litigation counsel (attorneys Renaud and McNamara / Mintz) for professional negligence. The complaint alleges that counsel's failure to properly present and cite the antedating evidence in the '431 IPR (the § 42.6(a)(3) violation) caused the invalidation of the '431 patent claims and reduced Parus's settlement/licensing values. This is not an infringement suit, but it is directly about the '431 patent and is docketed in the public record.

---

Current Status Summary

• The '431 patent's asserted claims were held unpatentable by the PTAB in IPR2020-00846 (FWD Oct. 19, 2021) and that decision was affirmed by the Federal Circuit on June 12, 2023 (70 F.4th 1365).
• Because those claims are invalid for all purposes, the earlier 2019 W.D. Tex. infringement actions and the N.D. Cal. LG action asserting the '431 patent can no longer be maintained on those claims. I did not find a document in this search confirming the precise final disposition (dismissal/settlement) of each individual district-court case; I therefore cannot state each case's final procedural end with high confidence.
• The later Parus campaigns (2021 onward) rely on different patents — not the '431 — and should not be attributed to this patent number.

---

Sources

• Google Patents, US7076431B2 (litigation links and family/litigation data): https://patents.google.com/patent/[US7076431](/patent/US7076431)/en
• Unified Patents litigation entries: https://portal.unifiedpatents.com/litigation/Texas%20Western%20District%20Court/case/6%3A19-cv-00433 ; .../6%3A19-cv-00432 ; .../6%3A19-cv-00437 ; .../6%3A19-cv-00438 ; .../6%3A19-cv-00454 ; https://portal.unifiedpatents.com/litigation/California%20Northern%20District%20Court/case/3%3A20-cv-05896 ; PTAB: https://portal.unifiedpatents.com/ptab/case/IPR2020-00846 ; https://portal.unifiedpatents.com/ptab/case/IPR2020-00686
• Parus First Amended Complaint v. Google, No. 6:19-cv-00433-ADA (W.D. Tex.) — Exhibit 1018 in IPR2020-00846
• Parus Holdings, Inc. v. Google LLC, 70 F.4th 1365 (Fed. Cir. 2023): https://storage.courtlistener.com/pdf/2023/06/12/parus_holdings_inc._v._google_llc_1.pdf
• Patterson + Sheridan, Federal Circuit Summary (week ending June 16, 2023)
• Parus Holdings malpractice complaint (Parus v. Mintz): https://fingfx.thomsonreuters.com/gfx/legaldocs/byvrezganpe/Parus%20v%20Mintz%20lawsuit.pdf
• UniCourt dockets: Parus Holdings Inc. v. Amazon.Com, Inc. (6:19-cv-00454) and Parus Holdings Inc. v. LG Electronics Inc. et al. (3:20-cv-05896)

Caveat: I flagged two items I could not fully verify from the search results — (a) the exact final disposition of each district-court case, and (b) the subject of CAFC No. 22-1143. I have not auto-corrected any case numbers, patent numbers, or citations, and where sources conflicted (e.g., the Amazon case number) I reported both rather than reconciling them.

I'll research the PTAB proceedings referenced on the patent page and fill in the available details.

I have a strong picture. Let me verify claim-level outcomes, the Apple appeal, and check for any additional proceedings.

Note on the source block first: the "PTAB proceedings on file" block states the USPTO ODP API returns no AIA trial proceedings for US 7,076,431. That is wrong / stale. The patent's own Google Patents page lists two PTAB matters, and both are independently confirmed below (docket records + a precedential Federal Circuit opinion). I flag them here rather than defaulting to "no PTAB activity." Two IPRs exist; all challenged system claims of the '431 patent were canceled and the cancellation was affirmed on appeal.

---

Proceedings overview

Two AIA trials have been filed on US 7,076,431 (both IPRs — no PGR, no CBM, none pending). One (IPR2020-00846) invalidated every claim it challenged and the Federal Circuit affirmed — final, no live proceeding. One (IPR2020-00686) failed to cancel the claims it challenged; Apple appealed, then the parties jointly dismissed the appeal on 2023-03-07. Bottom line for a defendant: claims 1–17 of the '431 patent are dead or derivative of a dead independent claim — if a demand letter quotes claim 1, 2, 4–7, 9, 10, 13, 14 (or anything depending from claim 1), the patent owner is asserting claims a final, court-affirmed PTAB judgment canceled. The only untested paper left is the method-claim family (¶ claims 18–32), and the '431 patent expired 2021-02-05, so any present-day assertion reaches only historical damages.

⚠️ Claim-numbering caution: the Federal Circuit's opinion is internally inconsistent as to whether claim 9 of the '431 patent was in the canceled set — the panel's Background omits it ("claims 1, 2, 4–7, 10, 13, and 14"), while its § I holding and Parus's own Notice of Appeal both include it. I reproduce the fuller set and flag the discrepancy rather than silently pick one.

---

IPR2020-00846 — Google LLC (with Samsung Electronics Co., Ltd.; [[Samsung Electronics America](/litigations/by-defendant/Samsung%20Electronics%20America), Inc.](/litigations/by-plaintiff/Samsung%20Electronics%20America%2C%20Inc.); [LG Electronics Inc.](/litigations/by-plaintiff/LG%20Electronics%20Inc.); LG Electronics U.S.A., Inc.) v. Parus Holdings, Inc.

• Type: Inter Partes Review
• Filed: 2020-04-18
• Status: Final Written Decision (docket: "Terminated Oct. 19, 2021"). FWD Paper 31 is captioned "JUDGMENT Final Written Decision Determining All Challenged Claims Unpatentable 35 USC 318a."
• Judge panel: APJs David C. McKone, Sheldon M. McGee, and Stacey G. White (Docket Alarm PTAB record; Tech Center 2600, Art Unit 2626)
• Petition grounds: Pre-AIA 35 U.S.C. § 103(a) obviousness. Principal art: WO 01/050453 to Kovatch ("Interactive Voice Response System") and Kurganov-262 (U.S. Pub. App. 2001/0047262, the publication of the application from which the '431 patent claims priority — asserted as prior art on the theory that the '431 claims lack written-description support and are not entitled to the 2000-02-04 priority date).
• Institution decision: Instituted. (Exact institution date is not in the sources I could retrieve; the FWD issued 2021-10-19, consistent with institution in October 2020, within the statutory one-year track.) The Google Patents page lists the matter as "filed (Final Written Decision)."
• Final Written Decision: 2021-10-19 — all challenged claims held unpatentable: claims 1, 2, 4, 5, 6, 7, 9, 10, 13, and 14 of the '431 patent. No claim was sustained. The panel's two dispositive predicate rulings:
  1. Kovatch qualified as prior art. The Board declined to consider Parus's conception/diligence evidence because it was improperly incorporated by reference from declarations into the Response and Sur-Reply, violating 37 C.F.R. § 42.6(a)(3).
  2. Kurganov-262 qualified as prior art. The panel found the limitation requiring a computing device "configured to periodically search via one or more networks to identify new web sites and to add the new web sites to the plurality of web sites" lacked written-description support in the earlier application, so the claims were not entitled to the 2000-02-04 priority date. The panel reasoned that the second embodiment's "devices" merely "appear as 'web sites'" and are not web sites, so that teaching could not be imported to the web-browsing embodiment.
• Settlement / termination: None — resolved by FWD and affirmance.
• Appeal: Yes. Parus appealed; consolidated as Nos. 2022-1269 and 2022-1270 (with IPR2020-00847 on the '084 patent). Affirmed 2023-06-12 in Parus Holdings, Inc. v. Google LLC, 70 F.4th 1365 (Fed. Cir. 2023) (precedential; Lourie, Bryson, Reyna, JJ.). The court held (a) the Board did not abuse its discretion disregarding evidence incorporated by reference (quoting General Access Solutions: "Judges are not like pigs, hunting for truffles buried in briefs"), and (b) the Board did not exceed its § 311(b) authority in deciding written-description priority, and substantial evidence supported the no-written-description finding. An errata to the opinion was filed 2023-07-24.
  • Opinion: https://www.courtlistener.com/opinion/[9405771](/patent/9405771)/parus-holdings-inc-v-google-llc/ · PDF: https://storage.courtlistener.com/pdf/2023/06/12/parus_holdings_inc._v._google_llc_1.pdf
• Defensive value: This is the kill shot. Claim 1 of the '431 patent is canceled, and because claims 2–17 all depend (directly or through claims 9, 10, or 14) on claim 1, the entire system-claim family is unavailable. Any infringement theory built on claims 1–17 is sanction-bait.

IPR2020-00686 — Apple Inc. v. Parus Holdings, Inc.

• Type: Inter Partes Review
• Filed: 2020-03-18
• Status: Final Written Decision (institution decision 2020-09-23)
• Judge panel: APJs David C. McKone (writing), Stacey G. White, Sheldon M. McGee
• Petition grounds: Pre-AIA § 103(a). Four grounds, every one led by Ladd (U.S. 6,269,336) plus Kurosawa (JP H9-311869) and Goedken (U.S. 6,393,423), with secondary references Madnick (5,913,214), Houser (5,774,859), and Rutledge (6,650,998). Support art included Bennett, Khan, Ho, Anderson, Maddalozzo, Driscoll, Muftic, Kenner, Rossides, and the MURAX (SIGIR '93) publication, with Dr. Loren Terveen as Apple's expert.
• Claims challenged / instituted: claims 1, 2, 3, 4, 5, 6, 7, 9, 10, 13, 14, 18, 19, 20, 21, 25, 26, 27, 28, 29, and 30 — all instituted.
• Institution decision: Instituted 2020-09-23 (Paper 9), finding a reasonable likelihood under § 314(a). Parus then filed a Request for Rehearing of the Institution Decision on 2020-10-07, arguing the Fintiv factors now favored denial (parallel W.D. Tex. case, opposed stay motion). I could not confirm the rehearing disposition from the retrieved record; the trial proceeded to FWD, so it did not derail institution.
• Final Written Decision: 2021-09-09 (Paper 37, "Final Written Decision original"). Important caveat — I could not retrieve the claim-by-claim verdict text, so I will not invent one. What the record establishes is inferential: Apple — not Parus — was the appellant in Federal Circuit Nos. 2022-1143/2022-1144. An appellant is the losing party, so the Board did not cancel all the claims Apple challenged; at least a subset survived. Parus's core defense, per its own hearing demonstratives, was that every ground failed on the "speaker-independent speech recognition device" limitation because Ladd's ASR "recognizes a selected speech pattern" and is speaker-dependent under the W.D. Tex. construction ("a speech recognition device that recognizes spoken words without adapting to individual speakers or using predefined voice patterns"). That is the most likely reason Apple lost, but I flag it as inference, not a quoted holding.
• Settlement / termination: The appeal was dismissed under Fed. R. App. P. 42(b) on 2023-03-07, "the parties having so agreed," each side bearing its own costs — i.e., a stipulated dismissal consistent with a settlement. Terms are not public.
• Appeal: Yes — Apple Inc. v. Parus Holdings Inc., Nos. 2022-1143, 2022-1144 (consolidating IPR2020-00686 and IPR2020-00687 for the '084 patent), dismissed by agreed order 2023-03-07 (nonprecedential). https://www.courtlistener.com/opinion/[9388472](/patent/9388472)/apple-inc-v-parus-holdings-inc/
• Defensive value: Double-edged. Apple lost — you cannot point to -00686 as a claim-killing judgment. But Apple is now estopped under § 315(e)(2) on art it raised or reasonably could have raised against claims 18–21 and 25–30, the very claims that remain live. If Apple (or a privy) is your co-defendant or supplier, that estoppel is useful leverage.

Related proceedings not on the '431 patent (context only)

Same panel, same appeal, but different patent numbers — IPR2020-00687 (Apple v. Parus, U.S. 9,451,084, FWD 2021-09-09, appeal 2022-1144) and IPR2020-00847 (Google/Samsung/LG v. Parus, U.S. 9,451,084, FWD 2021-10-19, appeal 2022-1270, affirmed). Google also litigated sibling patents in IPR2022-00279 (U.S. 6,721,705 — the '431's parent), and IPRs on the '941/'402 patents. Those do not invalidate 7,076,431 and should not be cited as if they do.

---

Strategic summary

Canceled vs. sustained vs. untested. Following the affirmed FWD in IPR2020-00846, claims 1, 2, 4–7, 9, 10, 13, and 14 of the '431 patent are canceled (claim 9 per the panel's holding, Parus's Notice of Appeal, and the Federal Circuit's § I holding; the opinion's Background section omits claim 9 — a discrepancy I cannot resolve from the retrieved text). Claims 3, 8, 11, 12, 15, 16, and 17 were never separately invalidated, but every one depends on claim 1, 9, 10, or 14 — all canceled — so they are commercially dead. Claims 18–32 (the method-claim family plus any unchallenged claims) are the only potentially live claims, and claims 18–21 and 25–30 were tested and survived Apple's IPR2020-00686. Claims 22–24 and 31–32 have never been before the Board in either trial.

Estoppel landscape. § 315(e)(2) binds the petitioners and their privies/RPIs — not the world:

• Google, Samsung, and LG (and their privies) are estopped as to claims 1, 2, 4–7, 9, 10, 13, and 14 on any ground raised or reasonably raisable — moot, since those claims are canceled — but they did not challenge claims 18–21 or 25–30, so they are not estopped on the live method claims.
• Apple is estopped as to all the claims it challenged — including the surviving 18–21 and 25–30 — on any art it raised or reasonably could have raised in -00686 (Ladd, Kurosawa, Goedken, Madnick, Houser, Rutledge, Bennett, Khan, Ho, Anderson, Maddalozzo, Driscoll, Muftic, Kenner, Rossides, MURAX).
• For a new, unprivied defendant, the full prior-art field remains open against the surviving claims: Kovatch (WO 01/050453), Kurganov-262 (US 2001/0047262), the Ladd family, and anything else — plus the '431 patent is expired (2021-02-05), so damages are capped by the § 286 six-year lookback and there is no injunctive exposure.

Pattern signals. This was a big-tech-vs.-NPE campaign, not a defensive-aggregator campaign: no Unified Patents (or similar aggregator) petition appears in the '431 record — the Google Patents page links to Unified Patents only as a litigation-data source, not as petitioner. Petitioners were Apple and a Google/Samsung/LG group, all counter-punching against Parus's 2019–2021 W.D. Tex. assertion campaign (6:19-cv-00432/-00433/-00437/-00438/-00454). The patent owner escaped total wipeout because Apple's separate petition failed, then mooted the loss by settling. The Board's -00846 judgment turned on a procedural self-inflicted wound — Parus's counsel's violation of § 42.6(a)(3) — which Parus later litigated in a legal-malpractice suit against Mintz, Levin (Parus Holdings, Inc. v. Mintz, Levin, Cohn, Ferris, Glovsky & Popeo, P.C., filed 2025-10-24, seeking up to ~$1.04B in lost recoveries). The Federal Circuit's affirmance of that procedural ruling is now precedential (Parus v. Google, 70 F.4th 1365), so the § 42.6(a)(3) point is citable precedent against any patent owner who tries to paper over a response with declaration exhibits.

---

Recommended next steps

1. If a demand letter or complaint cites claims 1–17, respond immediately with the operative disposition: IPR2020-00846, Paper 31 (P.T.A.B. 2021-10-19), FWD caption "Determining All Challenged Claims Unpatentable," affirmed in Parus Holdings, Inc. v. Google LLC, 70 F.4th 1365 (Fed. Cir. 2023), in which the court held: "we affirm the Board's holding that Appellees demonstrated by a preponderance of the evidence that claims 1, 2, 4–7, 9, 10, 13, and 14 of the '431 patent are unpatentable." Docket: https://www.docketalarm.com/cases/PTAB/IPR2020-00846/Google_LLC_v._Parus_Holdings_Inc/ · Opinion: https://www.courtlistener.com/opinion/9405771/parus-holdings-inc-v-google-llc/
2. If the assertion targets claims 18–30, treat it as a live (but narrow) fight. The claims survived Apple's IPR2020-00686 — but note that Apple is § 315(e)(2)-estopped on them; if you are Apple, a supplier, or a privy, your invalidity case is artificially constrained and you should consider interpleader/indemnity angles or a licensing posture.
3. Check expiration and damages windows first. The '431 patent is listed as expired 2021-02-05 (anticipated expiration). Any assertion in 2026 can only recover pre-expiration damages within 35 U.S.C. § 286, which typically collapses the case's economic value.
4. No active proceedings to calendar. Both IPRs are terminated and both appeals are closed (2023-06-12 affirmance; 2023-03-07 voluntary dismissal). There is no institution deadline, no oral hearing, and no 1-year § 316 FWD clock running. If the patent owner filed anything new on the '431 patent after 2023, it is not in the sources I could retrieve — but note the patent is expired, which severely limits the value of any such filing.
5. Verify before relying. I could not retrieve the claim-by-claim text of the IPR2020-00686 FWD (2021-09-09, Paper 37) nor the disposition of Parus's 2020-10-07 rehearing request, so I have not asserted which claims Apple failed to invalidate. Pull the FWD and Paper 31 from PTAB E2E (https://ptacts.uspto.gov/ptacts/) and confirm whether the Board's -00846 judgment expressly address claim 9.

I'll research the assignment record for US 7076431. Let me run targeted searches.

Let me try to access the USPTO assignment records directly and search for specific conveyances.

I found a 37 CFR 3.73(c) statement with reel/frame data. Let me retrieve the full document and search for the other links.

Assignment Record Reconstruction — US 7,076,431

Access caveat (read first): I could not open a live, authenticated USPTO Patent Assignment Search session in this run. The findings below are stitched from (a) the Google Patents legal‑events register for US 7,076,431, and (b) primary USPTO records embedded in PTAB/family filings that reproduce reel/frame data. Where I have no reel/frame, I say so rather than guessing. I have not fabricated any reel/frame, correspondent, or date. A live pull at the Assignment Center link at the bottom is required to complete the missing reel/frame fields.

---

Inventors

Both inventors are named on the face of the patent and appear as the assignors in the chain‑of‑title for the family (confirmed by the 37 CFR 3.73(c) statement reproduced below).

Pattern note: There is no evidence of inventor departure in the record — no reassignment away from the inventors to a third party within the 12‑month window; both inventors executed assignments tied to the original developer and/or the later holder. The multiple recordings of the same inventors (2010 and again 2014) are a chain‑of‑title cleanup pattern, not a fire‑sale tell.

---

Original assignee

Parus Holdings, Inc. (named on the issued patent face as Parus Holdings, Inc., Bannockburn, IL; described in the USPTO 3.73(c) statement as "Parus Holdings, Inc., a corporation").

• Primary line of business: Patented voice‑browser / voice‑activated information‑retrieval technology — the commercial system was built by Webley Systems, Inc. (the specification itself says the database "designed by Webley Systems Incorporated"; the source code headers in Tables 3–4 read /usr/local/cvsroot/webley/… and webley: server names). Parus Holdings is the vehicle on the patent face and later the litigation plaintiff.
• Product embodying the claims: The 2000–2001 filings describe an operating system (voice portal, IVR, web‑browsing servers, Nuance ASR, Lernout & Hauspie TTS). Whether Parus Holdings itself ever shipped a commercial product, versus being the holding/assertion entity for Webley's assets, is unclear from the sources retrieved.
• Current status: The corporate chain shows Parus Holdings → WWIS, LLC (security) → Webley Systems, Inc. → back to Parus Holdings, Inc. Parus Holdings remains the named assignee in PTAB proceedings through 2021–2022 (IPR2020‑00686, ‑00846). No bankruptcy filing by Parus or Webley was found in the retrieved record. A 2025–2026 malpractice suit (Parus v. Mintz Levin, ¶ filed 2025‑10‑24) indicates Parus is still an active litigant/rights‑holder.

---

Assignment timeline

Scope warning: The records I could verify by reel/frame come from a 37 CFR 3.73(c) chain‑of‑title statement filed for sibling application 12/973,475 (which became US 8,185,402) on December 1, 2014. That statement lists the family's chain of title; the same recorded documents (same reel) are the ones reflected in the '431 Google Patents register. I could not confirm that the '431 patent number is itemized on the face of reel 033922 without the document itself.

2010‑07‑07 (recorded) — Reel/Frame not retrieved

• Conveyance: Assignment of assignors' interest ("ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
• Assignor: Kurganov, Alexander (Zhukoff not named in this entry)
• Assignee: Parus Holdings, Inc.
• Correspondent: not retrieved
• Context: internal acquisition/consolidation of inventor rights into the named holder.

2013‑03‑18 (recorded) — Reel/Frame not retrieved

• Conveyance: Security Agreement
• Assignor: Parus Holdings, Inc.
• Assignee: WWIS, LLC
• Correspondent: not retrieved
• Context: securitization — Parus pledged the portfolio as collateral to WWIS, LLC. This is a lien, not a title transfer; it is the classic "IP collateral" move by a company monetizing its patent estate.

2014‑10‑09 (recorded) — Reel 033922 / Frame 0418

• Conveyance: Assignment of assignors' interest
• Assignors: Kurganov, Alexander; Zhukoff, Valery
• Assignee: Webley Systems, Inc.
• Correspondent: not retrieved (the 3.73(c) statement is signed by /Reena Kuyper/ as the assignee's authorized signer, dated 2014‑12‑01 — she is the signer, not necessarily the recording correspondent; flag as the only recurrence in this chain)
• Context: chain‑of‑title cleanup — re‑recording the inventor→company link (see next link, same reel, same day).

2014‑10‑09 (recorded) — Reel 033922 / Frame 0431

• Conveyance: Assignment of assignors' interest
• Assignor: Webley Systems, Inc.
• Assignee: Parus Holdings, Inc.
• Correspondent: not retrieved
• Context: internal reorg / cleanup — immediately follows Frame 0418 on the same reel and same day, so the two-step inventor→Webley→Parus sequence was executed together to perfect a clean title record (the 2014 filings exist precisely to document the full chain for the family).

Note on the two 2014 links: because they are consecutive frames (0418, 0431) on the same reel 033922, executed the same day, this is a single‑event chain‑of‑title perfection, not a genuine two‑stage economic transfer. I flag it as such so it is not miscounted as a "cascading transfer" (see signal 4).

Correspondent recurrence: A separate 2021 PTAB chain‑of‑title statement (petition artifact 1554706) bears /Gregory M. Howison, Reg. #30646/ (Oct. 5, 2021). I could not tie that statement to the '431 patent with confidence — it may belong to a different Parus matter or a different petitioner. Report it as unverified; do not treat it as the '431 recording correspondent unless the Assignment Center confirms.

---

Timeline diagram

timeline
    title Ownership of US 7076431
    2000 : Priority applications filed
    2004 : Continuation application filed
    2006 : Patent issued to Parus Holdings
    2010 : Assignment recorded to Parus Holdings
    2013 : Security Agreement to WWIS LLC
    2014 : Inventors assign to Webley Systems
         : Webley System assigns back to Parus
    2019 : Five infringement suits filed
    2021 : IPR cancels system claims
    2023 : Cancellation affirmed on appeal

---

NPE / troll‑pattern signals

1. Shell‑entity transfer — unclear
The chain includes Webley Systems, Inc. (an operating developer) and Parus Holdings, Inc. (a "Holdings" corporation, per reel 033922/0431). No LLC appears as the terminal title holder, and I found no evidence of a registered‑agent service address or single‑member state LLC. The "Holdings" suffix is suggestive but, without product/no‑product evidence for Parus, this does not rise to a finding.

2. Known asserter in the chain — present (moderate)
Parus Holdings, Inc. is the current holder and the plaintiff in at least five W.D. Tex. suits against Apple, Google, Samsung, LG, and Amazon (2019) plus the N.D. Cal. LG case (3:20‑cv‑05896). It litigated the '431 (and '084) at the PTAB against Apple and a Google/Samsung/LG group. That is a classic non‑practicing‑holder assertion posture. I could not, within this run, retrieve a specific RPX/Unified "high‑frequency plaintiff" directory entry naming Parus — so I credit this from the litigation record, not from a list citation. No Acacia / Marathon / IV / Wi‑LAN / Conversant / Vringo / Round Rock entity appears anywhere in the chain.

3. Repeat correspondent across the chain — unclear
Only one individual recurs in the retrieved material: /Reena Kuyper/, signer of the 2014 3.73(c) statement (reel 033922/0418 & /0431), and possibly Gregory M. Howison on a 2021 statement I could not tie to the '431. A single appearance is not the signal — recurrence across multiple links/patents is. The recording correspondents of record for the 2010, 2013, and 2014 recordings were not retrievable, so I cannot establish recurrence. Flag for follow‑up.

4. Cascading transfers (multiple LLCs in <24 months) — not present
There are two recorded links on 2014‑10‑09 (reel 033922/0418 and /0431) to non‑LLC entities. Because both are on the same reel and same day, they are a chain‑of‑title cleanup, not a cascade through chained shell LLCs.

5. Pre‑litigation transfer (within 6 months of first suit) — not present
The last recorded transfer predates the first infringement suit (July 2019) by ~5 years. No transfer was recorded in the six months before the 2019 filings.

6. Bankruptcy fire‑sale — not present
No Chapter 7/11 proceeding by Webley Systems or Parus Holdings appears in the retrieved record. (The only bankruptcy‑adjacent document surfaced — a 2023 NOVVI LLC IP security agreement — is unrelated to this patent and is excluded.)

7. Privateering — unclear
The patent was developed by Webley Systems and ended up asserted by Parus Holdings against Webley's one‑time competitors (Apple/Google/Amazon/Samsung/LG). Structurally that resembles privateering, but I have no SEC filing or Patent Progress/EFF coverage tying a Webley→Parus assertion mandate to this record. Not a finding on the current evidence.

8. Defensive aggregator — not present
The chain terminates at Parus Holdings, Inc., an entity that asserted the patent in 2019–2023 — the opposite of RPX/AST/LOT/Unified neutralization. No defensive aggregator appears.

---

Verdict

NPE — moderate confidence.

Two signals carry the call: (i) Parus Holdings, Inc. is the terminal holder and a non‑practicing assertion plaintiff — it sued Apple, Google, Samsung, LG, and Amazon in 2019 (W.D. Tex. 6:19‑cv‑00432/-00433/-00437/-00438/-00454) and defended the '431 at the PTAB (IPR2020‑00686, ‑00846), which is the assertion posture of an NPE rather than a competitor; and (ii) a 2013‑03‑18 Security Agreement to WWIS, LLC (per the Google Patents register) shows the estate was being used as loan collateral, typical of a monetization vehicle. I stop at moderate rather than high because the chain contains no classic shell LLC, no bankruptcy fire‑sale, and no confirmed repeat‑player correspondent, and I could not retrieve the reel/frame for the 2010 and 2013 recordings — so one of my two signals rests on date/party data rather than a verified reel entry.

Verify at: https://assignmentcenter.uspto.gov/ (search "7,076,431") or https://assignment.uspto.gov/patent/index.html — confirm (a) the 2010‑07‑07 reel/frame, (b) the 2013‑03‑18 WWIS, LLC reel/frame and its correspondent, (c) that the '431 is itemized on reel 033922 frames 0418 and 0431, and (d) the recording correspondent for every entry.

Open items I could not close: reel/frame for the 2010 Parus and 2013 WWIS recordings; the correspondent of record on all four recordings; whether Parus Holdings ever sold a product embodying the claims; the disposition of the 2023–2024 release/reassignment of the WWIS security interest, if any (none was found in the register).

I'll research the prior art cited on US 7,076,431's face and in its PTAB proceedings.

I have the PTAB exhibits. Now let me pin down the references cited on the patent's face (both U.S. and foreign), which is what the task asks about.

Let me retrieve the actual front-page "References Cited" list for the '431 patent.

Prior Art for US 7,076,431 — Analyst Memo

Critical scope note (read first). I was unable to retrieve the verbatim front‑page "References Cited" list of US 7,076,431 in this session (the Google Patents citation tables did not render in the retrieved page text, and my search budget was exhausted before I could pull the patent PDF's front page or the USPTO PatentCenter "References Cited" table). Two consequences flow from that:

1. I will not fabricate a list of examiner‑cited U.S./foreign patent documents. Where I cannot confirm an identifier, I say so.
2. What I can give you, and what is far more probative, is the prior art actually placed of record against the '431 patent in the PTAB — the art that was litigated and that (in IPR2020‑00846) resulted in cancellation of claims 1, 2, 4–7, 9, 10, 13, and 14, affirmed in Parus Holdings, Inc. v. Google LLC, 70 F.4th 1365 (Fed. Cir. 2023). These references are, as a practical matter, the "most relevant prior art" for this patent.

Single most important framing point for your § 102 question: the '431 patent was not invalidated under § 102. Every ground the Board sustained was pre‑AIA 35 U.S.C. § 103(a) obviousness. No cited reference was held to anticipate any claim. So the honest answer to "which claims does reference X anticipate under § 102" is: none as a matter of adjudicated record; below I give the § 102 exposure each reference creates (statutory category and claim mapping), which is a different and weaker thing.

---

A. The two references that actually killed the '431 claims (IPR2020‑00846)

These were asserted by Google LLC, Samsung Electronics Co., Ltd., [[Samsung Electronics America](/litigations/by-defendant/Samsung%20Electronics%20America), Inc.](/litigations/by-plaintiff/Samsung%20Electronics%20America%2C%20Inc.), [LG Electronics Inc.](/litigations/by-plaintiff/LG%20Electronics%20Inc.), and LG Electronics U.S.A., Inc. Grounds and per‑claim outcome per the Board's FWD (Paper 31, Oct. 19, 2021), reproduced verbatim in the Parus v. Mintz complaint exhibit:

Overall outcome: claims 1, 2, 4–7, 9, 10, 13, 14 unpatentable.

Reference detail

1. WO 01/050453 A1 — "Kovatch"

• Full citation: PCT International Publication WO 01/050453 A1 (Kovatch et al.), "Interactive Voice Response System."
• Dates: Published July 12, 2001; earliest possible priority date January 4, 2000 (Fed. Cir. opinion). Note a literal-reading discrepancy worth flagging: the Federal Circuit says Kovatch was "filed on July 12, 2001" while commentary (Weintraub Tobin, June 2023) says it "was filed on July 12, 2001, and possibly had a priority date as early as January 4, 2000." July 12, 2001 is most naturally the publication date of a PCT; the international filing date is not stated in the sources I retrieved. I report all three statements literally rather than reconciling them.
• Brief description: An interactive voice response system for retrieving information by voice — the primary § 103 reference for the '431's system architecture.
• § 102 exposure: As a published PCT designating the U.S., Kovatch is potentially § 102(e) prior art as of its international filing date (pre‑AIA). It is a § 102 candidate against claims 1, 2, 4–7, 10, 13, 14 and 9 — but note the Board found it insufficient standing alone; it required Neal as a secondary reference. A pure § 102 anticipation theory on Kovatch alone is therefore weaker than the § 103 theory the Board actually adopted and the Federal Circuit affirmed.

2. U.S. Published Patent Application 2001/0047262 A1 — "Kurganov‑262"

• Full citation: U.S. Patent Application Publication No. 2001/0047262 A1 (Kurganov et al.) — the publication of application Ser. No. 09/776,996, the '431 patent's own parent.
• Dates: Application filed February 5, 2001; published November 29, 2001.
• Brief description: Discloses the same specification as the '431 patent: a voice-based web browser system and a voice-activated device controller.
• Special status — this is the analytically important one: Kurganov‑262 became prior art only because the Board found the '431 claims were not entitled to the February 4, 2000 priority date for lack of written-description support for the limitation "configured to periodically search via one or more networks to identify new web sites and to add the new web sites to the plurality of web sites." On that basis it qualified as prior art against claim 9 (per the Fed. Cir.) and, per the petition, potentially claim 14.
• § 102 exposure: § 102(e) (U.S. application publication, filed before the invention date). But there is a logical trap here: a reference cannot anticipate a claim it does not describe. Because Kurganov‑262 shares the '431's specification, and the Board held that specification lacks written-description support for the "periodically search … identify new web sites" limitation, Kurganov‑262 cannot be a clean § 102 anticipant of claim 9 — only § 103 art when combined with Chakrabarti. Do not plead Kurganov‑262 as a § 102 anticipant without addressing this.

3. "Neal" — secondary reference (Kovatch grounds)

• Identifier not confirmed from the retrieved record. Used by the Board as the secondary reference in every Kovatch ground (claims 1, 2, 4–7, 9, 10, 13, 14). Verify the full citation and date in the IPR2020‑00846 petition (Paper 1) before relying on it.

4. "Chakrabarti" — secondary reference (claim 9)

• Identifier not confirmed. The Board credited Chakrabarti's web-crawling teachings as supplying the "periodically search … identify new web sites" limitation missing from both Kovatch and Kurganov‑262. Given the subject matter, this is likely the well-known focused-crawling literature (S. Chakrabarti et al., ~1999) rather than a patent, but I will not assert that without verification. § 103 art only (used in combination; not a § 102 anticipant).

5. "DeSimone" — secondary reference (claim 14, not reached)

• Identifier not confirmed. Asserted in grounds 3, 4, and 6 against claim 14. The Board expressly did not reach claim 14 ("neither party contends that claim 14 should be interpreted to require more than temporary storage…"). So DeSimone never carried an invalidity holding. Verify in the petition.

---

B. The prior art Apple asserted and LOST on (IPR2020‑00686)

Apple Inc.'s petition (filed Mar. 18, 2020; instituted Sept. 23, 2020; FWD Sept. 9, 2021) challenged claims 1, 2, 3, 4, 5, 6, 7, 9, 10, 13, 14, 18, 19, 20, 21, 25, 26, 27, 28, 29, 30 on four § 103(a) grounds. The Board did not cancel all challenged claims (Apple, as the losing party, was the appellant — appeal Nos. 2022‑1143/‑1144, dismissed by agreed order Mar. 7, 2023). These references therefore failed against the '431 — they are the least useful art in the record, notwithstanding that several are § 102‑qualified by date.

Reference list, from the Terveen declaration exhibit index (IPR2020‑00686):

Assessment: Apple's field was broad but shallow — the Board did not invalidate the claims. Do not cite IPR2020‑00686 as establishing anticipation or obviousness of any '431 claim. Two useful residuals nonetheless: (a) Apple is now § 315(e)(2)-estopped on this art against claims 18–21 and 25–30, and (b) the surviving method claims are the only live paper.

---

C. § 102 analysis — bottom line per statutory category

Under pre‑AIA § 102 with the '431's asserted priority date of February 4, 2000 (Provisional 60/180,344; second provisional 60/233,068 filed Sept. 15, 2000):

• § 102(b) art (public more than one year before Feb. 4, 2000 → before Feb. 4, 1999): Driscoll (1997), Houser (1998), Kraftsow (WO 98/03923, Jan. 1998), Muftic (1998), Kenner (1999), Madnick (1999), MURAX (July 1993), Chun (1996), Kurosawa (JP H9‑311869, Dec. 1997). None of these was held to anticipate or even invalidate any claim.
• § 102(e) art (U.S. patents/publications with earlier U.S. filings): Ladd, Goedken, Rossides, Rutledge, Khan, Ho, Anderson, Maddalozzo, Biffar, Bennett, Kovatch (PCT designating U.S.), Kurganov‑262.
• § 102(a)/(g) art (knowledge/use by others before invention): nothing in the record establishes such art.

No reference in the record is an adjudicated § 102 anticipant of any '431 claim. The claims died under § 103.

Where actual § 102 exposure is nonzero (for defensive pleading, not for a motion you can win on the papers):

• Kovatch (WO 01/050453) → claims 1, 2, 4–7, 9, 10, 13, 14. Best § 102 candidate by subject matter, but the Board needed Neal, so elect § 103.
• Kurganov‑262 (US 2001/0047262) → claim 9 (and argued claim 14). § 102(e) status only, and self-defeating as a pure anticipation theory for the reason given above; use with Chakrabarti under § 103.
• Ladd (US 6,269,336) → claims 1, 5, 6. Fails on the "speaker‑independent" limitation as construed. Do not assert.
• MURAX and Kraftsow are the only other references with arguable § 102(a)/§ 102(b) reach into claim 1's grammar/retrieval elements — and neither carried a ground.

---

D. Family references that are not prior art (avoid mis-citation)

The '431 specification's "CROSS‑REFERENCE TO RELATED APPLICATIONS" cites Ser. No. 09/776,996 (filed Feb. 5, 2001, now US 6,721,705) and provisionals 60/180,344 and 60/233,068. These are the '431's own priority chain, not prior art. The '431 also names "database 100 designed by Webley Systems Incorporated" and specifies a Nuance Communications speech-recognition engine and a Lernout & Hauspie ("Lemout and Hauspie" as printed) speech-synthesis engine — these are commercial-specification disclosures, not cited references, and none appears on a references-cited face.

---

E. Caveats and verification steps (required reading)

1. I could not retrieve the patent's front‑page "References Cited." The number of examiner‑cited references, and the exact U.S./foreign document list, must be pulled from the patent PDF front page (https://patentimages.storage.googleapis.com/55/06/98/6ca408f46c7af.../US7076431.pdf — note: the URL fragment I have is unverified; retrieve via https://patents.google.com/patent/US7076431/en) or USPTO PatentCenter for application 10/821,690. Do not represent my list above as the examiner‑cited list.
2. Identifiers "Neal," "Chakrabarti," and "DeSimone" are unverified. I have their roles and the claims they were mapped to, but not full citations. Pull IPR2020‑00846 Paper 1 (Petition) and Paper 22 (Petitioner's Reply) from PTAB E2E (https://ptacts.uspto.gov/ptacts/).
3. Dates: I am confident in the issue dates for the § 102(b)/§ 102(e) U.S. references listed (they are long‑settled), but every one should be confirmed against the printed patent or PatentCenter before being cited in a filing. Ladd's (US 6,269,336) and Goedken's dates in particular I flag as to be verified.
4. Claim set caution. Claim 9's inclusion in the canceled set: the Federal Circuit's Background omits claim 9 ("claims 1, 2, 4–7, 10, 13, and 14"), while the Board's own judgment table and ORDER include it ("claims 1, 2, 4‑7, 9, 10, 13, and 14 are unpatentable"). The Board's ORDER controls. Claims 18–32 (method family) were not invalidated by IPR2020‑00846; claims 18–21 and 25–30 survived Apple's IPR2020‑00686; claims 22–24 and 31–32 have never been before the Board.
5. Expiration. The '431 is listed "Expired – Lifetime," anticipated expiration Feb. 5, 2021 — so any 2026 assertion reaches only pre‑expiration damages within the § 286 lookback.

Primary sources used: https://patents.google.com/patent/US7076431/en ; https://storage.courtlistener.com/pdf/2023/06/12/parus_holdings_inc._v._google_llc_1.pdf (Fed. Cir. 22‑1269, June 12, 2023); Board FWD IPR2020‑00846 Paper 31 (Oct. 19, 2021) as reproduced in https://fingfx.thomsonreuters.com/gfx/legaldocs/byvrezganpe/Parus%20v%20Mintz%20lawsuit.pdf ; Terveen declaration exhibit index (IPR2020‑00686), ptacts.uspto.gov; https://www.docketalarm.com/cases/PTAB/IPR2020-00846/ ; https://www.theiplawblog.com/2023/06/articles/uncategorized/dont-ask-judges-to-be-archaeologists/ ; https://www.pattersonsheridan.com/wp-content/uploads/2023/06/Federal-Circuit-Summaries-Week-Ending-June-16-2023.pdf

I'll ground the prior-art references before building the obviousness analysis.

Good grounding. Now let me pin down the -00686 verdict text and the secondary references.

Now let me confirm the -00846 grounds (Neal/DeSimone) and the secondary references' subject matter.

Obviousness Analysis — U.S. Patent No. 7,076,431 ("Robust Voice Browser System and Voice Activated Device Controller")

35 U.S.C. § 103 (pre-AIA § 103(a))

---

0. Scope, sources, and how this builds on the prior sections

This section builds on, and does not repeat, the Patent Summary, Litigation Summary, and PTAB Challenges sections. It assumes those sections' bibliographic and procedural facts (priority Feb. 4, 2000; continuation of App. 09/776,996; expiry Feb. 5, 2021; IPR2020-00846 and IPR2020-00686; CAFC 22-1269/22-1270 affirmed June 12, 2023) and focuses exclusively on the § 103 merits.

Prior-art corpus used. Per the instruction to use the Prior Art results, I work from the reference set already surfaced in the record:

⚠️ Two flag notes on identifiers, per the no-auto-correction rule: (i) the Federal Circuit opinion spells the reference "Kovach" in one passage while the Board and the WO number (01/050453) use "Kovatch" — I use Kovatch and flag the variance; (ii) the Japanese reference appears both as "JP H9-311869" and "JP H09-311869" — same document, reproduced literally.

Methodology. Two of the three combinations below were adjudicated on the public record verbatim; I anchor those in the Board's own findings and label everything I extend to the eleven never-tested claims as my analysis, not a holding.

---

1. Governing law and level of ordinary skill

Framework. Pre-AIA § 103(a) applies (the '431 never entered the AIA regime — priority 2000, filed 2004). The Graham v. John Deere factors control: scope and content of the prior art, differences between the claims and the art, the level of ordinary skill, and secondary considerations.

KSR rationales that drive this patent. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007), supplies the dispositive rationales here:

• "If a person of ordinary skill can implement a predictable variation, § 103 likely bars its patentability."
• Where "a technique has been used to improve one device, and a person of ordinary skill in the art would recognize that it would improve similar devices in the same way," the use is obvious absent proof the application was beyond skill.
• "Any need or problem known in the field of endeavor at the time of invention and addressed by the patent can provide a reason for combining the elements in the manner claimed."

Teaching away — the correct standard. A reference teaches away only if a POSITA "would be discouraged from following the path set out in the reference, or would be led in a direction divergent from the path that was taken by the applicant." In re Mouttet, 686 F.3d 1322, 1333–34 (Fed. Cir. 2012). Critically, "the mere disclosure of alternative designs does not teach away," id. (quoting In re Fulton), and "just because better alternatives exist in the prior art does not mean that an inferior combination is inapt for obviousness purposes."

PHOSITA. For the analysis below I use: a bachelor's degree in electrical engineering or computer science (or equivalent), plus 2–3 years' experience building telephony/IVR or speech-recognition applications, including client-server web access. This is consistent with the parties' competing experts in the record (Lipoff for the -00846 petitioners; Terveen for Apple; Occhirosso for Parus).

Claim construction point that mattered. In the parallel W.D. Tex. litigation, "speaker-independent speech recognition device" was construed as "a speech recognition device that recognizes spoken words without adapting to individual speakers or using predefined voice patterns." That construction is the fulcrum of the one combination that failed (§ 6 below).

---

2. Combination A — Kovatch + Neal (optionally + Chakrabarti, + DeSimone)

Grounds as pleaded (IPR2020-00846)

Final disposition: "ORDERED, based on a preponderance of the evidence, that claims 1, 2, 4-7, 9, 10, 13, and 14 are unpatentable." — IPR2020-00846, Paper 31 (P.T.A.B. Oct. 19, 2021), affirmed in Parus Holdings, Inc. v. Google LLC, 70 F.4th 1365 (Fed. Cir. 2023).

Claim 1 — element-by-element

Dependent-claim mappings found: claim 2 (Internet — Kovatch Fig. 3); claim 4 (standard telephone — element 15); claim 5 (analyze phonemes — obvious per Mr. Lipoff's uncontroverted testimony); claim 6 (naturally spoken commands — "free-form speech"); claim 7 (content descriptor pre-defining the page portion — Kovatch 15:11–14, 15:23–34: where to insert a ZIP code on a weather page and where the results appear); claims 10 and 13 (ranking associated with the addresses and access in rank order — Kovatch 16:28–34, 19:8–10, 23:25–29: ranking Amazon above CD Now and accessing in that order); claim 14 (database storing retrieved information — Kovatch/Neal combination, Lipoff ¶ 176).

Why a POSITA would have combined Kovatch with Neal

Deployed independently by the Board and endorsed on appeal:

1. Same field, same problem, analogous art. Both are keyword/search-based retrieval systems that retrieve records responsive to a user's query. Kovatch retrieves from web sites; Neal from catalogs/datasets. They are squarely in the same field of endeavor.
2. Kovatch's express design point creates the need. HeyAnita "expects a single result when it requests information from web pages." If that single destination fails or returns nothing, the user gets a negative result. Neal's stated purpose is precisely to remedy that failure mode: "The purpose of expanding the search strategies is to avoid the situation wherein the search fails to identify the desired item from any data set, and thereby drops into logical block 222 [where] a negative search result is reported to the user." Neal, 7:28–32.
3. Neal discloses the exact sequential structure. A first data set is searched (Fig. 2, block 206); if no match, a second data set is searched (block 210). Neal, 6:34–52. Neal expressly contemplates repeat application of the same keyword strategy to successive suppliers' catalogs ("if the first preferred supplier did not yield an exact string match … the same search methodology could be applied to a second preferred supplier's catalog," 6:59–65) — i.e., the claimed "sequentially access … until found or … all … accessed" step.
4. Fault tolerance / robustness is a recognized, self-evident engineering objective. The patent itself frames the goal in these terms; the prior art states the identical objective. Where the problem is known in the field, that alone supplies the reason to combine (KSR).
5. Predictable result, no new mechanism. Adding a fallback dataset/site list to an existing query engine is a rearrangement of known elements performing their established functions — the paradigm of predictable variation.
6. Chakrabarti (claim 9). Once the system maintains a list of sites per category, periodically crawling the Web to discover new sites and add them to the list is the standard, known technique for keeping a directory current (the AltaVista/Lycos-era crawler). The Board accepted that "a skilled artisan would have added Chakrabarti's web crawling feature to the combination … such that the HeyAnita server periodically searches the Internet to identify new web sites and to add the new web sites to its listing of web sites."
7. DeSimone (claim 14). Storing retrieved information in memory operatively connected to the computer so it can be reused and converted to voice prompts is the "obvious implementation" of the retrieval path Kovatch already describes (Lipoff ¶ 176).

Parus's counterarguments and the Board's rejections

---

3. Combination B — Kurganov-262 + Chakrabarti (+ DeSimone) — claim 9 (and 14)

This is a structurally different attack from Combination A and reaches a different set of claims by a different mechanism.

The mechanism: claim 9 loses its priority date. The Board found that the limitation requiring a computing device "configured to periodically search via one or more networks to identify new web sites and to add the new web sites to the plurality of web sites" lacked written-description support in the earlier application. Because the shared specification describes the device-browsing embodiment's devices as merely "appear[ing] as 'web sites'" — which "were not 'web sites'" and whose teachings "were not interchangeable" — the disclosure could not be imported into the web-browsing embodiment. Claim 9 therefore was not entitled to the Feb. 4, 2000 date, and Kurganov-262 (published Nov. 29, 2001, from App. 09/776,996) became prior art against it.

The Federal Circuit affirmed this on substantial evidence, and held that § 311(b) does not limit the issues (including written description) the Board may consider. Parus v. Google, 70 F.4th 1365 (Fed. Cir. 2023).

Motivation to combine. Once Kurganov-262 supplies the voice-browser architecture (it is, literally, the same specification) and the priority date is lost, the addition of Chakrabarti's crawler to periodically discover and add new sites is:

• a known technique (web crawling);
• applied to a known directory/listing structure (the ranked web-site records of Kurganov-262);
• yielding a predictable improvement (broader, fresher coverage with no change in the retrieval mechanism).

Why this route is more robust than Combination A. It does not depend on Parus's procedural forfeiture; it depends on a § 112 written-description finding that was litigated on the merits and affirmed on substantial evidence.

---

4. Combination C — Ladd + Kurosawa + Goedken (+ Madnick, Houser, Rutledge) — the combination that failed

Apple's IPR2020-00686 is the cautionary half of this analysis. Its judgment caption is:

"JUDGMENT — Final Written Decision Determining No Challenged Claims Unpatentable — 35 U.S.C. § 318(a)."

Concerning claims 1–7, 9, 10, 13, 14, 18–21, and 25–30: "Petitioner has not proved, by a preponderance of the evidence, that [those claims] are unpatentable."

The grounds Apple pleaded

The theory was sound in outline: Ladd (6,269,336) supplies a network voice-response architecture (ASR unit 254 + communication node 212 + prompt/grammar engine), Kurosawa (JP H09-311869) supplies the URL database with sequential access to a plurality of URLs (the "1j" fallback), and Goedken (6,393,423) supplies the telephony application platform. Madnick went to claim 7's content descriptor, Houser to phonemes/natural speech, Rutledge to claim 9's network searching.

Why it failed — one limitation, fatal

Every ground ran through Ladd for the "at least one speaker-independent speech recognition device" limitation. Under the W.D. Tex. construction ("recognizes spoken words without adapting to individual speakers or using predefined voice patterns"), Ladd's ASR was held not to qualify:

"Ladd specifically discloses that '[w]hen the ASR unit 254 identifies a selected speech pattern of the speech inputs, the ASR unit 254 sends an output signal to implement the specific function associated with the recognized voice pattern.'" (Ladd, 9:35–40.)
"If the node can automatically select the model used in the algorithm based on the user's speech patterns, then the speech recognition is speaker-dependent."

Parus's expert (Occhiogrosso) further argued the language shows "speech pattern"/"voice pattern" mean a key word or phrase matched to a stored vocabulary — not phoneme-level, speaker-independent recognition. And Parus successfully attacked the motivation to combine on latency grounds: Ladd must meet near-real-time timing; Bennett's benchmark was 3–5 seconds; "adding any additional URLs to Ladd to access while executing a speech request would result in a slower system." (The Board never had to reach this because the speaker-independent gap was dispositive, but the argument is preserved in the record and lost.)

How a defendant would repair Combination C

Because the failure was reference-specific, not concept-specific, the repair is to swap in art with express speaker-independent, phoneme/HMM-based recognition — no enrollment, no speaker templates:

1. Replace Ladd with a speaker-independent ASR in a voice-web-browser context. The related IPR2022-00279 record (Google v. Parus on U.S. 6,721,705 — the '431's parent) cites an "Uppaluru" voice-web reference stating: "The voice web system dynamically retrieves the suitable subset of training word patterns from the user's speech profile matching the voice navigation key words in the page being presented to the user," together with Jong (phonemes/syllables), Potter (natural spoken phrases), and Mitsuyoshi (voice recognition + natural language processing). ⚠️ I could not confirm the Uppaluru patent number from the retrieved text — verify before relying.
2. Keep Kurosawa for the sequential-URL "1j" element (it is the element Kovatch alone lacked, and it survived unchallenged in -00686 because the case failed earlier).
3. Keep Goedken for the telephony application platform.
4. Address the latency teaching-away head-on. This is where the Parus "adding URLs slows the system" argument must be neutralized — e.g., by showing that the ranking step itself addresses latency (the top-ranked/fastest site is queried first), so the combination improves rather than degrades responsiveness. That reframing is not in the record as an adjudicated finding; it is an argument a defendant would have to make.

Practical consequence: for a new, unprivied defendant, Ladd's failure is not binding in any res judicata sense — different parties, and IPR judgments bind the petitioner/privies. But it is strong persuasive evidence that the "speaker-independent" element is the patent's real § 103 moat for claims that keep the Feb. 4, 2000 date.

---

5. Claim-by-claim § 103 disposition

Note: claims 5, 6, 7 survived Apple's IPR but were later cancelled by the -00846 FWD (Oct. 19, 2021, issuing after Apple's Sept. 9, 2021 FWD). Both statements are true; the later judgment controls.

Derivative effect: claims 8, 11, 12, 15, 16, 17 are commercially dead regardless of their § 103 merit, because each depends on claim 1, 9, 10, or 14 — all cancelled. But an element-by-element § 103 case still matters for completeness and for the method family.

---

6. The never-adjudicated claims — proposed combinations and motivations

⚠️ This section is my analysis, not a Board or court holding. The 35 U.S.C. § 103 conclusions below are reasoned extensions of the adjudicated record.

Claim 3 — "said internet is a local area network"

Obvious as a bare substitution of a known network topology. Kovatch and Ladd both describe network-connected systems; LAN vs. Internet is a design choice with no new mechanism. The '431's own specification concedes the servers "are also capable of accessing databases stored on Local Area Networks (LANs) or Wide Area Networks (WANs)."

Claim 8 — periodic polling without user instruction for availability, response time, and changes in information location; ranking created from that polling

This is the "robustness" core of the patent (and the subject of the parent's own US 6,721,705 claim 1). Proposed combination:

• Kovatch or Ladd — the base voice-web architecture;
• + a network-monitoring / server-health-checking reference (e.g., the load-balancing and distributed-network-agent art cited in the extended record, such as US 5,941,996 ("Distributed network agents"), US 6,085,217 (load balancing), US 6,101,182, US 6,498,797, US 6,950,821) — for periodic, unattended polling/health checks and response-time measurement;
• + Bennett — for the field's own 3–5 second real-time responsiveness benchmark, which supplies the motivation to poll and rank by speed;
• + Ho — for the bandwidth/latency constraint that made the polling-and-ranking solution desirable.

Motivation: server-pool health checking and response-time-based selection were commercially standard (Cisco LocalDirector, F5 BIG-IP, and comparable load balancers, late 1990s) and were used to improve precisely the attributes claim 8 recites. Under KSR, using a known technique to improve a similar device in the same way is obvious. Detecting "changes to the location of said information" is the direct analogue of content-extraction failure detection in the content-descriptor architecture Kovatch already uses.

⚠️ I have not verified the reference numbers in this sub-list against the Prior Art corpus; they are drawn from the broader record and my background knowledge, and should be confirmed before any filing.

Claims 11 and 12 — criteria for modifying rankings; weighting criteria against one another

Pure optimization. Once claim 8's ranking exists, selecting among responsiveness, availability, and content-location change is a routine engineering trade-off; weighting multiple criteria is the textbook implementation of a multi-factor score. No new mechanism, no unexpected result. Obvious over the claim 8 combination, alone or in further view of any ranking/load-balancing reference.

Claim 15 — re-ranking so the site that has the information moves to the top

Interactive reinforcement of the access history. The system already records the timestamp of last access (dateTime field, Table 1) and maintains a rank field; incrementing rank on success and decrementing on failure is the direct, predictable application of the polling/ranking agent's disclosed function. Obvious over Kovatch + the claim 8 polling art.

Claim 16 — recognition grammars and instruction sets stored in a database

Expressly disclosed by Kovatch (destination tree / executable functions) and by Ladd (grammar/vocabulary lookup; "grammar generator"). Also squarely disclosed by Kurganov-262 itself, and by the '431's own Table 1 (records held in the database). Obvious; was not separately challenged, presumably because it added nothing.

Claim 17 — polling/ranking of newly discovered sites

Combination: claim 9's Chakrabarti crawler + claim 8's polling/monitoring art. Motivation: the same robustness rationale — a newly discovered site is unusable until its availability, speed, and format are known; therefore the same poll-and-rank treatment applies. Predictable application of a known technique.

Claims 18–21, 25–30 (method family)

Reach the same result. Tested by Apple and survived — but note the survival is art-specific (Ladd failed the speaker-independent element), not a merits vindication. A method-claim § 103 case would mirror the Kovatch+Neal structure (with the method steps of grammar selection, sequential access, and audio output) plus a speaker-independent ASR reference.

Claims 22–24, 31–32

Never before the Board in either trial. Their scope cannot be assessed from the text I have (claim 18's text and 19–32 were truncated in my source — see § 8). These are the only genuinely untested paper in the entire patent and the only claims on which no petitioner is estopped. Any future § 103 challenge should be built here and here alone.

---

7. Motivation-to-combine synthesis

The generalizable reasons a POSITA would combine these references — these are the KSR rationales to plead:

On teaching away. Parus's only substantively interesting argument — that adding URLs increases latency and therefore the prior art teaches away — is weak under In re Mouttet and In re Fulton: it is, at most, a disclosure of a trade-off, and the ranking feature at the heart of the patent is itself the latency mitigation. A defendant should turn the argument around: ranking by response time is the predicted remedy for the latency problem, which is exactly why a POSITA would have arrived at it.

---

8. Caveats, contradictions, and open items

1. Claim 9 discrepancy (flagged, not resolved). The Federal Circuit's Background section lists the cancelled-claims set as "claims 1, 2, 4–7, 10, 13, and 14" — omitting claim 9 — while the Board's Order, Parus's Notice of Appeal, and the opinion's own § I holding include claim 9. I reproduce the fuller set (1, 2, 4–7, 9, 10, 13, 14) and flag the inconsistency rather than silently picking one.
2. Fragility of the Kovatch-based holding outside the IPR. The entire Kovatch route to claims 1, 2, 4–7, 10, 13, and 14 rests on (a) Kovatch having an earliest possible priority date of Jan. 4, 2000 — only ~one month before the '431's Feb. 4, 2000 date — and (b) Parus's procedural forfeiture under 37 C.F.R. § 42.6(a)(3). Parus asserted conception by July 12, 1999 and reduction to practice by Dec. 31, 1999, both of which would predate Kovatch. In district court, Rule 42.6(a)(3) does not apply and Parus could properly present and cite its evidence. The Kovatch-based invalidity of those claims is therefore procedurally contingent in a way the affirmed judgment does not show on its face. (That contingency does not reopen the cancelled claims — the affirmance is final — but it matters for understanding the Parus v. Mintz malpractice theory and for assessing any residual claim.) By contrast, the Kurganov-262 route to claim 9 rests on a merits § 112 finding affirmed on substantial evidence and is procedurally sturdier.
3. § 112 may be the stronger attack than § 103 for claim 9. The Board's claim-9 holding is really a written-description holding wearing an obviousness hat: the claim lost its priority date, which made Kurganov-262 prior art. The affirmative § 112 defect is independent and, for claim 9, dispositive.
4. Unverified reference identities. I could not retrieve the patent numbers or disclosure text for Neal, Chakrabarti, DeSimone, the "Uppaluru" voice-web reference, or the full content of Madnick (5,913,214), Houser (5,774,859), and Rutledge (6,650,998). Their roles above are stated at the level the record supports; do not cite them for specific disclosures without pulling the documents.
5. Full text of claims 18–32 not available to me. Claim 18 was truncated mid-sentence in my source and claims 19–32 were not reproduced. My § 103 treatment of the method family is therefore structural, not element-by-element. Pull the printed patent before relying.
6. Expiration and damages. The '431 is listed "Expired – Lifetime," anticipated expiration Feb. 5, 2021. Any present-day assertion reaches only historical damages within 35 U.S.C. § 286, with no injunctive exposure.
7. No 2026 CAFC activity found specific to 7076431; I could not run a PACER-native check.
8. Certificate of cancellation unverified. I could not confirm from the retrieved sources whether a certificate cancelling claims 1, 2, 4–7, 9, 10, 13, 14 has formally issued in the USPTO file. The judgment is final; the ministerial step is separate.

---

Bottom line

• Cancelled by adjudicated § 103: claims 1, 2, 4–7, 9, 10, 13, 14 — Combination A (Kovatch + Neal, plus Chakrabarti for claim 9 and DeSimone for claim 14) and, for claim 9 independently, Kurganov-262 + Chakrabarti after the written-description priority loss. Affirmed, 70 F.4th 1365.
• Survived adjudication, but only on one reference-specific gap: claims 3, 18–21, 25–30 — Apple's Ladd + Kurosawa + Goedken combination failed on the "speaker-independent" element under the W.D. Tex. construction. Repairable by substituting art with express speaker-independent, phoneme-based recognition and by reframing the latency teaching-away argument.
• Never tested, no petitioner estoppel: claims 8, 11, 12, 15, 16, 17, 22, 23, 24, 31, 32 — the only genuinely open § 103 surface left, though the system-side members are commercially dead via dependency on cancelled claim 1, 9, 10, or 14.
• The single most important caveat: the Kovatch leg of the -00846 judgment is bound up with a § 42.6(a)(3) procedural forfeiture over a ~1-month priority margin. The judgment is final and the claims are cancelled; but anyone reasoning about the '431's § 103 strength on the merits — rather than its disposition — should not treat the Kovatch-based holding as a clean merits win.