US 10320981

Let me search more specifically for litigation and CAFC activity tied to this exact patent number.

Let me check for any Federal Circuit appeal arising from the Delaware case and the current status of that litigation.

US Patent 10,320,981 (US10320981B2) — Summary

Core bibliographic data

• Title: Personal voice-based information retrieval system
• Patent number: US10320981B2 (8-digit form: 10,320,981)
• Assignee: Parus Holdings, Inc. (Bannockburn, IL)
• Inventor: Alexander Kurganov (Buffalo Grove, IL)
• Application: No. 15/707,951, filed 2017-09-18 (continuation)
• Issue date: 2019-06-11
• Priority date: 2000-02-04 (Provisional App. 60/180,343)
• Legal status (per Google Patents/USPTO record): Expired – Fee Related; anticipated expiration listed as 2021-02-06
• Continuity chain: 09/777,406 → US7516190B2 → 11/711,773 (abandoned) → 12/787,801 → US9377992B2 → 15/193,517 → US9769314B2 → 15/707,951 → US10320981B2; child application 16/436,764 (US20190297189A1, abandoned). Source: Google Patents record for US10320981B2.

Abstract

"The present invention relates to a system for retrieving information from a network such as the Internet. A user creates a user-defined record in a database that identifies an information source, such as a web site, containing information of interest to the user. This record identifies the location of the information source and also contains a recognition grammar based upon a speech command assigned by the user. Upon receiving the speech command from the user that is described within the recognition grammar, a network interface system accesses the information source and retrieves the information requested by the user."

Plain-language overview of the independent claims

The patent has 28 claims; the independent claims are 1, 5, 10, 19, and 24.

• Claim 1 (method): A speech-recognition engine receives a user's spoken command over a network via an IVR application offering a "user-defined search." The engine selects a user-assigned recognition grammar tied to a web request and a website URL. A media server retrieves a stored information-source-retrieval instruction for that grammar; a web-browsing server then accesses the site, executes a content extractor that uses a content-descriptor file to locate and separate only the relevant portion of a web page (using a named object), retrieves only that portion, converts it to audio via a speech-synthesis engine, and transmits the audio message to the user's voice-enabled device.

• Claim 5 (voice-browsing system): The system-side counterpart of claim 1 — a speech-recognition engine coupled to a media server (which initiates a voice-response application on connection), a database storing retrieval instructions keyed to the established grammar, a web-browsing server that accesses only the relevant portion of a web page using a content extractor and content-descriptor file with a named object, and a speech-synthesis engine that converts the retrieved information to audio for the voice-enabled device.

• Claim 10 (method of selective retrieval): Identifies one of a plurality of speech commands in a speech-recognition lexicon from audio of the user's spoken words; uses the identified command to access a corresponding descriptor file that identifies (i) a web-accessible information source and (ii) request information; fetches response data containing a named object with content; extracts the content using the named object; generates audio response data with a message responsive to the command and based on the extracted content; and directs playback of that audio on the user's electronic-communication device.

• Claim 19 (apparatus): A system with a transceiver, a database of descriptor files (each identifying a web-accessible source and request info), a speech-recognition engine that identifies a speech command from audio, a media server that accesses the descriptor file, a web-browsing server that retrieves responsive data and extracts content by the name of a named object, and a synthesizer that generates/transmits audio response data — programmed to direct playback on the user's device.

• Claim 24 (apparatus, user-device side): An electronic-communication device having a microphone and speaker, in communication via a network with a remote computer system (comprising a speech-recognition engine, media server, web-browsing server using a named object to extract content, and speech synthesizer) that is programmed to direct playback of the audio response on the device's speaker.

Litigation / CAFC status (search results)

• The patent is asserted in Parus Holdings Inc. v. [Amazon.com, Inc.](/litigations/by-plaintiff/Amazon.com%2C%20Inc.), 1:23-cv-00190 (D. Del., filed Feb. 17, 2023) (Judge Gregory B. Williams), where the complaint reports U.S. Patent Nos. 7,516,190 B2; 9,377,992 B2; and 10,320,981 B2 — the last being this patent. The infringement allegations target Amazon's Alexa/Alexa Voice Service ("skills" such as Tide Pool retrieving portions of a website and converting content to speech). Sources: Justia docket 1:2023cv00190; Stanford NPE Litigation Database (patent 10320981); docket Filing 3 (Report to Commissioner).
• CAFC 2026: I found no Federal Circuit appeal specifically directed to patent 10320981 in my searches. The CAFC 2026 matter that surfaced — Google LLC v. Parus Holdings, Inc., Appeal No. 2024-2381 (oral argument audio posted; CAFC vacated a PTAB decision per IPWatchdog, July 2026) — concerns a different Parus patent family (U.S. 6,721,705 and descendants 7,881,941 and 8,185,402), not US10320981.

Uncertainty notes

• I could not independently confirm a CAFC 2026 docket entry naming US10320981; none was located in the searches performed. The most recent district-court docket data I retrieved for 1:23-cv-00190 dates to March 2023; the current disposition of the Delaware case (e.g., any later judgment, settlement, or appeal) was not confirmed.
• The "Expired – Fee Related" status and 2021-02-06 anticipated expiration come from the Google Patents record and may reflect non-payment of maintenance fees/terminal-disclaimer effects; verify against USPTO PAIR/PatentCenter for current maintenance-fee status.

Based on my searches, here is what is known about litigation involving US Patent 10,320,981 B2 ("Personal voice-based information retrieval system," assigned to Parus Holdings, Inc.):

Known litigation

1. Parus Holdings Inc. v. [Amazon.com, Inc.](/litigations/by-plaintiff/Amazon.com%2C%20Inc.)

• Plaintiff: Parus Holdings Inc.
• Defendant: Amazon.com, Inc.
• Jurisdiction: U.S. District Court for the District of Delaware (D. Del.)
• Case number: 1:23-cv-00190 (also cited as 1:2023cv00190)
• Filing date: February 17, 2023
• Patents asserted: The complaint (D.I. 1, filed 2/17/2023) expressly identified U.S. Patent Nos. 7,516,190 B2; 9,377,992 B2; and 10,320,981 B2 — so US 10,320,981 was one of three patents asserted in this case. (Sources: Justia docket for 1:2023cv00190; Ex Parte case database; Stanford NPE Litigation Database.)
• Assignments/judges: Assigned to Judge Gregory B. Williams; Magistrate Judge Christopher J. Burke also listed on the docket.
• Status: Closed. Both the Stanford NPE Litigation Database and the Ex Parte case database list the case status as "Closed." I was not able to confirm the precise terminal disposition (e.g., stipulated dismissal, settlement, summary judgment, or trial outcome) from the sources retrieved, so I cannot state with confidence how it resolved beyond the closed status. A docket search on PACER or the Delaware District Court would be needed for the final order.

Related matter — not litigation of 10,320,981 (excluded for clarity)

• Google LLC v. Parus Holdings, Inc., No. 24-2381 (Fed. Cir.) — This is a Federal Circuit appeal (filed September 30, 2024; argued June 5, 2026; opinion issued July 13, 2026) of a PTAB IPR decision. However, the CAFC opinion and the claims at issue (polling mechanism, website rank numbers, "unexpected response") concern a different family member — the '705 patent (per the opinion's discussion of the "′705 patent," col. 19–20) — not US 10,320,981. I therefore did not count it as litigation of 10,320,981.

Caveats

• The Google Patents page for 10,320,981 flags the family as having litigation and links to a Delaware District Court case (1:23-cv-00190) and a Darts-ip family record — consistent with the Amazon case above.
• I found no other district-court cases specifically identifying 10,320,981 in the sources searched. The JDJournal article about Mintz's fee suit against Parus references Parus's broader enforcement campaign (including suits against Apple, Google, Samsung, LG), but those cases appear to involve other patents in the family and were not shown to include 10,320,981; I did not verify any specific case numbers for them.
• To confirm the final outcome of the Amazon case and to rule out any additional cases, a PACER query on the Delaware docket 1:23-cv-00190 and a search of the PTAB (IPR/PGR) records for 10,320,981 would be the next steps.

The first search surfaced a Docket Alarm link to IPR2022-00805 (Google LLC v. Parus Holdings) — but that petition cites US 9,769,314 (a sibling in the same family). I need to verify whether any AIA proceeding specifically names US 10,320,981, and map the family-wide PTAB activity. Let me dig deeper.

Excellent — the family-wide picture is emerging. There's a Google IPR (IPR2022-00805) against the sibling '314 patent, an Apple IPR (IPR2022-00948), and Google IPRs where the Board denied institution on Parus' U.S. 7,516,190 (the direct parent of '981). I need to pin down which patents each proceeding targeted, and specifically whether any proceeding names 10,320,981 itself.

I've hit my research-step limit, so I'm answering from the structured ODP data plus the web-verified family record. Here is what the evidence supports.

---

Proceedings overview

There are zero AIA trial proceedings (IPR / PGR / CBM) on file for US 10,320,981 itself — the USPTO ODP list is empty, and web search surfaced no petition naming the '981 patent. The default stands: no PTAB activity on this patent. That said, the search did surface heavy PTAB activity on the '981 patent's continuation sibling (US 9,769,314, which was fully invalidated, all claims 1–26) and institution denials on its direct parent (US 7,516,190). Bottom line for a defendant: the '981 patent has never been through an IPR and no claim of it has been canceled by the PTAB, but its substantively identical sibling has a complete invalidation roadmap sitting in a public Final Written Decision, and the '981 patent itself expired on 2021-02-06 (per Google Patents legal-status data) — so exposure is damages-only and the "troll's" strongest claims are the very ones a PTAB panel already found unpatentable in the sibling.

---

Family-adjacent PTAB proceedings (NOT on US 10,320,981 — flagged per instructions)

IPR2022-00805 — Google LLC v. Parus Holdings, Inc. (U.S. 9,769,314 B2 — continuation sibling of '981)

• Type: Inter Partes Review
• Filed: 2022-04-04
• Status: Final Written Decision — all challenged claims (1–26) held unpatentable (DocketAlarm: "Final Written Decision Determining All Challenged Claims Unpatentable 35 U.S.C. § 318(a)"; Law360 headline: "Google Wins PTAB Invalidation On Voice-Command Patent")
• Judge panel: David C. McKone, Patrick M. Boucher, James J. Mayberry — Boucher authoring (per FWD)
• Petition grounds: Claims 1–26 of the '314 patent under 35 U.S.C. § 103(a) — obviousness over Wise + Woods, Wise + Dasan (claims 1–25), and Wise + Dasan + Uppaluru (claim 26); the petition also pressed a broken priority-chain / written-description challenge to whether Woods (Ex. 1012) was prior art (DocketAlarm / AboutBlaw FWD excerpts)
• Institution decision: Instituted on all challenged claims — 2022-11-07 (Patexia)
• Final Written Decision: Issued 2023-11-06 (DocketAlarm FWD PDF; Law360 reported 2023-11-08). Verdict at claim level: claims 1–26 all unpatentable. Panel quote: "Based on the record before us, Petitioner has shown, by a preponderance of the evidence, that claims 1–26 are unpatentable." The FWD also contains a duty-of-candor / show-cause section directed at Parus's counsel (Scott W. Hejny) over priority-chain representations to the Board (Banner Witcoff copy of FWD)
• Settlement / termination: None — full merits FWD
• Appeal: Docketed at the Federal Circuit as Appeal 2024-1329 (per Patexia's case summary). Disposition as of my sources: not confirmed
• Defensive value: This is the single most valuable document for anyone defending the '981 patent. The '314 claims (voice-command retrieval using a speech-recognition grammar, an information-source-retrieval instruction, a web-browsing server accessing "a portion" of a web page, a content extractor, and a content-descriptor file) are substantively the same claim architecture as '981 claims 1, 5, 10, 19, and 24. The Wise/Woods/Dasan/Uppaluru combinations are a ready-made invalidity roadmap. (Note: this proceeding is estoppel-binding only on Google and its privies, not on a new defendant.)

IPR2022-00948 — Apple Inc. v. Parus Holdings, Inc. (U.S. 9,769,314 B2 — same sibling patent)

• Type: Inter Partes Review
• Filed: on or about 2022-05-05 (post-institution fees paid 2022-05-05; DocketAlarm shows Apple filing Google's '314 IPR petition as Ex-1017 on that date)
• Status: Terminated without a Final Written Decision — the Board approved a refund of Apple's post-institution fees on 2023-03-15 (DocketAlarm Notice of Refund), which indicates the proceeding ended before a merits decision (settlement or joinder/termination; the specific basis is not confirmed in my sources)
• Judge panel: Not confirmed in my sources
• Petition grounds: Apple likewise asserted Woods and a priority-chain contention against the '314 patent (referenced in the Google FWD's show-cause discussion: "Apple's IPR2022-00948, which likewise asserted Woods and presented a [priority-chain contention]")
• Institution decision: Not confirmed; the post-institution fee refund implies no FWD followed
• Final Written Decision: None
• Settlement / termination: Terminated by 2023-03-15 with fee refund — terms (if any) not public in my sources
• Appeal: None identified
• Defensive value: Confirms that a second major tech defendant (Apple) independently targeted the same family with the same art — reinforcing that the Wise/Woods/Dasan cluster is the strongest attack path against '981-style claims.

Google IPRs on the direct parent — US 7,516,190 (and US 7,386,455): institution denied

• Type: Inter Partes Review (petitions by Google LLC)
• Status: Institution denied — per Parus's own counsel: "The current victory follows prior wins in which the Board denied Google's requests to review the claims of Parus' U.S. Patents Nos. 7,386,455 and 7,516,190" (McKool Smith press release, 2024-08-15)
• Proceeding numbers: Not available in my sources — do not guess; the PTAB E2E search for Parus Holdings as patent owner will yield the numbers
• Defensive value: Important caveat — a denial of institution is not a merits holding that the claims are patentable. The '190 parent (from which '981 descends) survived the threshold § 314 screening, but that creates no estoppel for anyone and no PTAB finding of validity. For a defendant, this is a two-sided signal: the Board wasn't persuaded by Google's initial showing on '190, but the identical-in-substance '314 claims were later wiped out on similar art — so the merits question on '981-family claims is genuinely live.

---

Strategic summary

Claims of '981 — CANCELED vs. SUSTAINED vs. UNTESTED. On the '981 patent's own PTAB record: zero claims canceled, zero claims sustained on the merits, all 28 claims untested (independents 1, 5, 10, 19, 24; dependents 2–4, 6–9, 11–18, 20–23, 25–28). The PTAB has never institutionally reviewed — let alone reached — the '981 claims. That absence is itself meaningful: the '981 patent granted in 2019 and expired (per Google Patents) on 2021-02-06, so the window in which a challenger would bother petitioning for IPR largely closed while the family was being litigated on its siblings instead. The practical exposure is past damages only, and the claims most likely to be asserted mirror the '314 claims that a PTAB panel unanimously found unpatentable.

Estoppel landscape. § 315(e)(2) estoppel runs only to the petitioners and their privies in the proceedings where grounds were raised or reasonably could have been raised. Google (IPR2022-00805) and Apple (IPR2022-00948) are the only parties estopped from re-litigating Wise/Woods/Dasan/Uppaluru against the '314 patent — and neither estoppel extends to a new defendant asserting against the '981 patent. For a defendant being sued on '981 today, all prior-art grounds remain fully available — including exactly the art that felled the sibling: Wise (Ex. 1007), Woods (Ex. 1012), Dasan (Ex. 1013), and Uppaluru, in the combinations and with the claim-element mappings laid out in the IPR2022-00805 FWD. Google's April 2022 stipulation in the W.D. Tex. case (agreeing not to pursue Wise-based obviousness in court if the IPR instituted) is a useful template, not a constraint, on a new defendant.

Pattern signals. This is a coordinated, multi-year, multi-front attack on the Parus voice-browser portfolio by Google and Apple — with Google as the lead aggressor. Across the portfolio the results are mixed: full cancellation of the '314 sibling (IPR2022-00805) and of the related '945,1084 family (CAFC affirmed, appeal 22-1270); institution denied on the '190 parent and '455; and in a different Parus family (the '705 patent), the CAFC just vacated a Parus win and remanded (IPWatchdog, 2026-07-14) — showing the Board and Federal Circuit are actively second-guessing each other on these voice-browser patents. Parus fights hard (McKool Smith, with a sanctions show-cause episode in IPR2022-00805 over priority-chain candor). No defensive aggregator (e.g., Unified Patents) appears in the chain — these were direct competitor-filed petitions. The Amazon litigation in Delaware (Parus Holdings Inc. v. Amazon.com, Inc., 1:23-cv-00190, D. Del.) asserts family patents including 7,516,190 and 9,377,992, and the Stanford NPE Litigation Database also lists '981 as involved — the district court, not the PTAB, is where this patent is being fought.

---

Recommended next steps

• If you are a defendant: Do not wait for an IPR that will never come — the patent is expired (2021-02-06 per Google Patents' legal-status record; confirm exact expiration/maintenance-fee history via USPTO Patent Center), so only past damages within the § 286 six-year window are at stake, and there is no injunction risk. The highest-value move is district-court invalidity on the IPR2022-00805 record: pull the FWD (DocketAlarm copy: https://www.docketalarm.com/cases/PTAB/IPR2022-00805/Google_LLC_v._Parus_Holdings_Inc/docs/11-06-2023-Board/Final_Written_Decision__original-35-Determining_All_Challenged_Claims_Unpatentable_35_USC_%C2%A7_318a.pdf) and the petition's exhibit set, and map Wise/Woods/Dasan/Uppaluru onto the '981 claims — the claim architecture is the same, and no estoppel bars you.
• If the demand letter cites claims tracking '314 claims 1–26 (speech-recognition grammar → media server selects an information-source-retrieval instruction → web-browsing server accesses "a portion" of a page via a content extractor using a content-descriptor file): quote the FWD's sentence — "Petitioner has shown, by a preponderance of the evidence, that claims 1–26 are unpatentable" — and put Parus to its burden of showing those claims are materially different from the canceled sibling claims; on this record that is a steep climb.
• Monitor the pending threads: Federal Circuit Appeal 2024-1329 (Google's appeal of the '314 FWD) — if the CAFC affirms, the invalidity roadmap is appellate-blessed; if it reverses, the roadmap weakens. Also monitor the Delaware case (1:23-cv-00190) for claim-narrowing and for which '981 claims Parus actually asserts.
• Confirm the procedural record before relying on it: the two items I could not verify from primary sources are (a) the docket numbers of Google's institution-denied IPRs on 7,516,190 / 7,386,455 (search PTAB E2E under "Parus Holdings" as patent owner), and (b) the precise termination basis of IPR2022-00948 (the fee refund indicates termination without a FWD, but the underlying order should be pulled from PTAB E2E). No active PTAB trial-stage milestones exist for '981 itself because no trial was ever instituted on it.

I'll attempt to verify the recorded assignment data before writing anything up.

Good progress — I found family reel/frame data. Let me pin down the chain-of-title specifics for this application and Parus's corporate status.

I was able to retrieve the family chain-of-title data and the prosecution/legal-event record, but I hit the research-step limit before I could query the Assignment Center directly for application 15/707,951. I flag every gap explicitly below rather than filling it in.

Inventors

Pattern notes:

• No co-inventor departures to flag. Unlike the sibling patents in this family (e.g., US 7,881,941, which names Kurganov and Valery Zhukoff), the '981 names Kurganov alone. There is therefore no multi-inventor team whose departure could signal a pre-sale attrition pattern.
• What is unusual is the 16-year gap between the priority filing (2000-02-04) and the filing of this continuation (2017-09-18), with the sole inventor still in place. That is a family-maintenance pattern (repeatedly re-filing continuations to keep a claim set alive and re-issuable), not a fire-sale pattern. It is consistent with the Royalty/Assertion picture set out below.
• I could not determine Kurganov's employer of record at the 2000 priority date beyond the litigation record: he was at Vail Systems, Inc. in the mid-1990s, and a recorded assignment "From: Alex Kurganov To: Vail Systems, Inc." appears in a USPTO CBM exhibit for a related Parus patent. The '981 specification instead credits Webley Systems Incorporated for the database (col. 5) and Bannockburn, IL is the assignee address. The Vail → Webley relationship is asserted in litigation ("Parus' predecessor, Webley Systems, Inc."; and Parus as "successor-in-interest to Vail Inc.") but I could not verify a recorded instrument documenting it.

Original assignee

Parus Holdings, Inc. (as printed on the face: "Assignee: PARUS HOLDINGS, INC., Bannockburn, IL (US)"). The '981 face also lists Applicant: Parus Holdings, Inc. and Attorney/Agent/Firm: Patent Law Works LLP.

• Products embodying the claims — yes, on the record, but party-supplied. Parus's own infringement pleading (reproduced in the 2025 Mintz fee complaint) states: "Since its founding, Parus has offered through its operating division a robust diversity of products and services to customers in a wide variety of markets. Parus' products and services have included audio and video conferencing, email management, voice messaging, polling and transcription, IM/presence, collaboration, softphone, and virtual assistant solutions" — specifically naming ParusOne, ParusOffice, and ParusSpeak. The predecessor product line was the Webley Assistant, which Parus's IPR exhibits show was deployed with a Nuance ASR engine and an L&H text-to-speech engine by 1998–1999 — i.e., substantively the system claimed here.
• Primary line of business: voice-enabled / unified-communications and interactive-voice-response services.
• Current status — operating, private, and relocated. No bankruptcy, dissolution, or acquisition surfaced in any source. Parus is no longer at the printed Bannockburn, IL address: the W.D. Wash. transfer opinion notes "Parus is in the midst of transferring all operations to the Austin area," and Mintz's October 2025 complaint describes Parus as "a Texas-based technology company specializing in voice-enabled and AI-driven communication technologies." I found no SEC filings for Parus — it appears to be a private company, so no 10-K/8-K corroboration of revenue is available.

Classification caveat: The Stanford NPE Litigation Database lists Parus Holdings Inc. under asserter category "8 Product company" — i.e., the leading public NPE directory treats it as an operating company, not a classic NPE. That is the single most important data point for the verdict below.

Assignment timeline

Retrieval status — read this first. I was not able to query the USPTO Assignment Center (https://assignmentcenter.uspto.gov/ or https://assignment.uspto.gov/patent/index.html) directly before exhausting my research steps. I therefore cannot confirm the reel/frame for any instrument recorded against application 15/707,951 itself. What follows is: (a) one recording tied specifically to this application, evidenced by the patent record; and (b) the family-level chain-of-title instruments, evidenced by a sworn 37 CFR 3.73(c) statement filed in a sibling application. Items are tagged with confidence. Per the constraints, I have not invented reel/frame numbers for the '981 application.

• Execution date unknown / recorded 2017-12-19 — Reel/frame NOT RETRIEVED

  • Conveyance: Assignment ("Assignment of assignee interest"; "See document for details" per Google Patents legal events)
  • Assignor: Alexander Kurganov (sole inventor)
  • Assignee: PARUS HOLDINGS, INC.
  • Correspondent: Not retrieved. Probable, not confirmed: Patent Law Works LLP — the firm of record prosecuting this application (see face of patent) and the firm whose attorney Reena Kuyper signed the family's chain-of-title statements (Dec. 1, 2014). Flag: this firm recurs across the Parus family's takeover filings.
  • Context: Confirmatory / perfecting assignment from the sole inventor to the company that was already the applicant of record — recorded ~3 months after the 2017-09-18 filing of the continuation. Normal housekeeping for a continuation filed 16 years after the parent; not a transfer to a new owner.
  • Source: Google Patents legal events for US10320981B2 ("2017-12-19 Assigned to PARUS HOLDINGS, INC. … Assignors: KURGANOV, ALEXANDER"). Execution date not shown.

• Recorded circa 2007–2008 (exact date unverified) — Reel 033922/0418

  • Conveyance: Assignment (conveyance type as characterized in the citing statement; underlying instrument not seen)
  • Assignors: Alexander Kurganov; Valery Zhukoff
  • Assignee: Webley Systems, Inc.
  • Correspondent: Not retrieved.
  • Context: Original inventor assignment into the operating predecessor company. Coverage caveat: this instrument names two inventors and was cited as chain of title for application 12/973,475 — it is a portfolio/inventor-level instrument. I cannot confirm it covers 15/707,951 or its parent chain, since the '981 names Kurganov alone.
  • Source: Statement Under 37 CFR 3.73(c), application 12/973,475 ("Robust Voice Browser System and Voice Activated Device Controller"), signed /Reena Kuyper/, December 1, 2014.

• Recorded circa 2007–2008 (exact date unverified) — Reel 033922/0431

  • Conveyance: Not confirmed — the citing statement lists it as the next link in an inventor→assignee chain of title but does not characterize it as a change of name, merger, or asset transfer. Treat conveyance type as unverified.
  • Assignor: Webley Systems, Inc.
  • Assignee: Parus Holdings, Inc. (the 3.73(c) statement names Parus Holdings, Inc. as the terminal assignee of a two-step chain)
  • Correspondent: Not retrieved.
  • Context: Internal consolidation / predecessor-to-successor transfer — Webley Systems to Parus Holdings, same business, same Bannockburn, IL locus, same inventor. Note: Parus's litigation filings call Webley "Parus' predecessor," but I found no instrument establishing whether this was a merger, asset purchase, or name change.
  • Source: same 3.73(c) statement, application 12/973,475, Dec. 1, 2014.

Related prosecution-control facts (not assignments, but ownership-relevant)

• Parent-chain prosecution, Dec. 2006: the file history of application 09/777,406 (the '981's original grandparent) contains a notice that "The Power of Attorney to you in this application has been revoked by the assignee who has intervened as provided by 37 CFR 3.71. Future correspondence will be mailed to the new address of record." The address of record became STEVE Z. SZCZEPANSKI, KELLY, DRYE & WARREN LLP, 333 West Wacker Drive, Suite 2600, Chicago, IL 60606 (attorney docket 47242-00028USPT). This confirms assignee control of prosecution from 2006 onward — relevant to standing and to who directed the continuation chain.
• No post-issuance assignments found. Nothing indicates the patent ever left Parus Holdings, Inc. No IP-holding subsidiary, no security interest, no license-and-assign instrument surfaced.

Timeline diagram

timeline
    title Ownership of US 10320981
    2000 : Provisional application filed Feb 4
    2001 : Parent application filed Feb 6
         : Kurganov assigns to Webley Systems
    2008 : Webley Systems transfers to Parus Holdings
         : Exact recordation date unverified
    2017 : Continuation application filed Sept 18
         : Kurganov assignment to Parus recorded Dec 19
    2019 : Patent issues June 11
    2023 : Amazon Delaware suit names the patent

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT. The patent never leaves its original assignee. Assignee of record is Parus Holdings, Inc., the same corporate family that appears on the face of the 2019 patent and in the Reel 033922/0431 predecessor transfer. There is no "IP / Patents / Licensing / Ventures" transferee, no registered-agent service address, and no single-member Delaware/Texas LLC anywhere in the chain. The "Holdings" suffix is present but attaches to the original developing company, not to a downstream acquirer — naming alone, which the task correctly instructs me not to treat as a finding. Counterweight: I could not verify that Parus currently ships product or generates product revenue; its product evidence rests on its own pleadings and Webley-era deployments.

2. Known asserter in the chain — NOT PRESENT (as a listed entity); UNCLEAR (as a functional category). Neither Parus Holdings, Inc. nor Webley Systems, Inc. appears on any of the enumerated NPE lists (Acacia, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, Spangenberg entities). The contrary evidence is affirmative: Stanford's NPE Litigation Database classifies Parus under asserter category "8 Product company" (https://npe.law.stanford.edu/party/parus-holdings-inc). Unclear factor: Parus is unambiguously a high-frequency serial plaintiff — 2019 wave (Apple, Google, LG, Samsung, Amazon), 2021 wave (Google, Microsoft, Apple, Samsung), a 2021 wave against Bank of America, Capital One, Charles Schwab, and Fidelity, a 2014 suit against SunTrust, and the 2023 Amazon case. Suing retail banks is a monetization hallmark rather than a competitor-dispute hallmark, and RPX has editorialized on Parus "reviv[ing] its first litigation campaign."

3. Repeat correspondent across the chain — NOT PRESENT as an NPE-recording signal; but note a recurring firm pattern. The correspondents I recovered are Patent Law Works LLP (prosecution of the '981; Reena Kuyper signing the family's 3.73(c) statements on Dec. 1, 2014) and Steve Z. Szczepanski, Kelly, Drye & Warren LLP, Chicago (assignee-intervened prosecution of the parent from 2006). This is a prosecution/chain-perfection recurrence, not the shell-LLC recording-agent recurrence the signal targets — the same lawyer is not papering a series of differently named assignees. I flag it only because it is the sole recurrence I can document, and because it shows a single firm controlling title perfection across the family for over a decade. Separately, the enforcement side shows a different recurrence: Mintz, Levin, Cohn, Ferris, Glovsky & Popeo P.C. (attorneys Michael McNamara, Michael Renaud, William Meunier) filed the 2019 wave under a contingent-fee "Monetization Program," per Mintz's 2025 complaint; McKool Smith P.C. (John B. Campbell, Austin) handled the IPR defense. Mintz subsequently sued Parus in D. Mass. (Oct. 2025) for > $2 million in unpaid success fees, alleging Parus "wrongfully retained the benefits" of the monetization work.

4. Cascading transfers — NOT PRESENT. One recorded instrument tied to this application (the 2017-12-19 Kurganov→Parus recording), plus a single predecessor consolidation at Reel 033922/0431. No <24-month LLC chain, no shared correspondent addresses across multiple transferees.

5. Pre-litigation transfer — NOT PRESENT. The only recorded transfer is 2017-12-19, roughly three months after the 2017-09-18 filing and more than five years before the first suit naming this patent (Feb. 17, 2023, D. Del.). That is a filing-driven recording, not venue or standing engineering.

6. Bankruptcy fire-sale — NOT PRESENT. No Chapter 7/11 proceeding involving Webley Systems, Inc. or Parus Holdings, Inc. appears in any source. No sale-of-portfolio-in-bankruptcy instrument.

7. Privateering — NOT PRESENT (with an adjacent monetization overlay worth noting). Parus asserts its own patents against actual competitors in voice assistants (Apple, Google, Amazon, Samsung, Microsoft, LG) — the definitional opposite of privateering, which requires an operating company to hand patents to a third-party NPE to assert on its behalf. There is no such NPE. What is present is outsourced, contingency-fee monetization: the Mintz complaint describes a program in which Mintz was to "represent Parus in petitions for inter partes review and in the review itself," selected the targets, and chose the Waco Division of the Western District of Texas as venue — i.e., counsel-driven enforcement strategy.

8. Defensive aggregator — NOT PRESENT. The chain terminates at Parus Holdings, Inc., not at RPX, AST, LOT Network, Unified Patents, or OIN. (Unified Patents' portal tracks the Delaware case against Parus as a litigation-data entry only — that is not ownership.)

Verdict

Operating-company assertion.

The chain never leaves its originator: Parus Holdings, Inc. is both the assignee printed on the face of the 2019 patent and the terminal assignee of the predecessor transfer at Reel 033922/0431, and the only other recorded instrument (executed 2017, recorded 2017-12-19) is a confirmatory inventor-to-applicant assignment of this very application. Every classic NPE structural signal is absent — no shell-entity transferee, no cascading LLC chain, no known-NPE assignee, no defensive aggregator, no bankruptcy sale, and the Stanford NPE Litigation Database affirmatively categorizes the holder as a "Product company." Its enforcement targets are its own competitors in voice-assistant technology.

Two honest caveats that keep this from being a clean call. (i) The "product company" evidence is largely party-supplied — Parus's own pleading resurrects ParusOne/ParusOffice/ParusSpeak, and I could not verify current product availability or revenue; the court in the Microsoft transfer fight noted Parus was in the midst of relocating all operations to Austin, and Mintz's 2025 complaint describes Parus principally through its enforcement posture. (ii) The enforcement pattern is NPE-shaped in intensity, including suits against retail banks and a contingency-fee "Monetization Program" run by outside counsel with venue selection by the law firm. If Parus's product line is in fact discontinued, the correct reclassification would be NPE — moderate confidence, on signal 2 alone. That determination turns on facts I could not verify.

To verify (next steps): (1) Query the USPTO Assignment Center by patent number 10320981 and by application 15/707,951 to capture the reel/frame, execution date, and correspondent of record for the 2017-12-19 recording and to confirm whether Reels 033922/0418 and 033922/0431 in fact cover this application. Search page: https://assignment.uspto.gov/patent/index.html (or https://assignmentcenter.uspto.gov/). (2) Pull the underlying Reel 033922/0431 instrument to determine whether Webley Systems → Parus was a merger, asset purchase, or change of name — this materially affects the "internal consolidation" characterization. (3) Check Delaware/Illinois corporate registries and the USPTO Assignment Center for any security interest or license recorded against Parus's portfolio.

I've hit the search-step limit, so I'll synthesize the answer from the sources already retrieved (Google Patents record for US10320981B2, the parent US9769314B2 PDF, and cross-referenced patent pages). I'll flag confidence levels explicitly.

---

Prior Art Analysis — US Patent 10,320,981 B2 ("Personal Voice-Based Information Retrieval System")

A. Source & reliability notes

• What I searched: USPTO/Google Patents records for patent number 10320981 (interpreted literally — no similar numbers), the parent US9769314B2 PDF (which carries the family's "References Cited" page, Primary Examiner Lisa Hashem), and cross-referenced "cited-by" pages for individual references.
• Citation count: The face of US10320981B2 lists 393 references ("Citations (393)" on Google Patents). The complete list was truncated in the retrieved record (visible portion ends at US5416834A). I could not retrieve the USPTO PTOL-892 form or file wrapper in this session, so the examiner's per-claim designations (X/Y/A) are not confirmed; the claim-level §102 mapping below is my technical analysis of each reference's disclosure against the independent claims, not the examiner's formal rejection.
• Statutory framework: The '981 patent is a continuation claiming priority to Provisional App. 60/180,343 (Feb. 4, 2000). Because the chain's effective filing date is before March 16, 2013, pre-AIA §102(a)/(b)/(e) governs. A reference is §102 prior art only if it was (i) published or known/used before Feb. 4, 2000 (§102(a)/(b)), or (ii) a U.S. patent/application filed before that date and ultimately published (§102(e)).

---

B. The independent claims in one line each (for mapping)

• Claim 1 (method): speech command → speech-recognition engine selects user-assigned recognition grammar tied to a URL/web request → media server retrieves stored information-source-retrieval instruction → web-browsing server executes a content extractor using a content-descriptor file and a named object to pull only the relevant portion of the page → speech-synthesis output to the user's device.
• Claim 5 (system): system-side counterpart of claim 1.
• Claim 10 (method): identify speech command from a lexicon → access a descriptor file identifying (i) a web-accessible source and (ii) request info → fetch response data containing a named object with content → extract content using the named object → generate audio response → direct playback.
• Claim 19 (apparatus): transceiver + database of descriptor files + speech-recognition engine + media server + web-browsing server (named-object extraction) + synthesizer.
• Claim 24 (apparatus, device-side): microphone/speaker device in communication with that remote system.

---

C. Most relevant prior art (ranked)

1. US 5,819,220 A — "Web triggered word set boosting for speech interfaces to the world wide web"

• Full citation: US Patent 5,819,220 A; inventors Sarukkai et al.; assignee Yahoo! Inc.; granted Oct. 6, 1998 (pre-2000).
• Description: A speech interface to the World Wide Web in which the speech-recognition vocabulary is dynamically "boosted" with terms extracted from web pages; a user speaks a natural-language request, the system performs a web retrieval, and results are returned in audible form. This is a voice-browser reference squarely in the same field as the claimed invention.
• Confirmed as cited: Appears in the cited-by list of the '981 patent's family records (it is cited by US9769314B2 and appears in the same family citation record that includes US10320981B2).
• Potential §102 mapping:
  • Claim 10: Strong — steps (a) identify spoken command, (b) access a web-accessible source using request info, (c) fetch response data, (e) generate audio response, (f) direct playback are all arguably present. The weak point is step (d) — extraction "using a named object" — unless the reference's page-parsing/vocabulary-boosting is read to use named/tagged objects.
  • Claims 1 and 5: Moderate — discloses speech→web-retrieval→audio, but does not clearly disclose the user-defined recognition grammar + URL stored in a user-defined record, the content-descriptor file, or the clipping client creation flow. Anticipation of claims 1/5 fails on those elements absent a broad reading.
  • Claims 19 and 24: Moderate — discloses the speech-recognition engine, web server, and synthesizer architecture, but the named-object extraction and descriptor-file database are not clearly present.

2. US 5,884,266 A — "Audio interface for document based information resource navigation and method therefor"

• Full citation: US Patent 5,884,266 A; inventor Waldman et al.; assignee IBM; granted Mar. 16, 1999 (pre-2000).
• Description: An audio interface that converts documents — including web pages — into audio, parsing the document's structure so that only relevant content is read aloud, and accepting voice commands to navigate among documents/links. It discloses content extraction based on document structure (links, headings — arguably "named objects").
• Confirmed as cited: US10320981B2 appears in the cited-by list on the Google Patents page for US5884266A.
• Potential §102 mapping:
  • Claim 10: Strong candidate — identifies speech commands (lexicon), accesses a document/information source, fetches response data, extracts content based on structural elements, generates an audio response, and directs playback.
  • Claims 19 and 24: Strong — the apparatus (transceiver, speech-recognition, web fetching, synthesizer, playback command) maps well; the descriptor-file database limitation is the main gap.
  • Claims 1 and 5: Weaker — no user-assigned recognition grammar/URL record and no content-descriptor-file created by a clipping client.

3. US 6,233,318 B1 — "System for accessing multimedia mailboxes and messages over the internet and via telephone"

• Full citation: US Patent 6,233,318 B1; assignee Bosch Telecom GmbH (later Ipcom GmbH & Co. KG); granted May 15, 2001; filed late-1990s (German priority 1997/1998 — filing date not independently confirmed in this session; treat with caution). If its U.S. filing date is before Feb. 4, 2000, it qualifies under §102(e).
• Description: A system that lets a user retrieve multimedia messages (voice mail, e-mail, fax) over the Internet and via a telephone using speech commands, with text-to-speech playback of retrieved content — a voice-activated network-retrieval system.
• Confirmed as cited: US10320981B2 appears in the cited-by list on the Google Patents page for US6233318B1.
• Potential §102 mapping: General mapping to claims 1, 5, 10 (speech command → retrieve network-stored content → audio playback). It lacks the web-page content-extraction with named objects and the user-defined grammar/URL record specifics, so anticipation of the independent claims in full is unlikely without a very broad reading.

4. US 2001/0054085 A1 — Kurganov, "Personal voice-based information retrieval system" (applicant's own parent publication)

• Full citation: US Patent Application Publication 2001/0054085 A1; inventor Alexander Kurganov; published Dec. 20, 2001; claims priority to Feb. 4, 2000 (parent of this very patent).
• Description: The published application of the '981 patent's own specification (same disclosure).
• Cited-by marker: Appears with a * (examiner citation) in family records.
• Potential §102 mapping: As a same-family publication, it is normally overcome by the priority claim and would only be §102(e)-type "self-collision" prior art against claims not entitled to the benefit of the earlier filing dates. On its face it discloses everything in the '981 specification (identical disclosure) — so if priority were lost, it would be a complete anticipation of all 28 claims.

5. US 5,325,421 A — "Voice directed communications system platform"

• Full citation: US Patent 5,325,421 A; assignee AT&T Bell Laboratories; granted June 28, 1994 (in the visible "Citations" list).
• Description: A telephony platform where callers direct system functions by voice — voice-directed routing and service selection. Relevant to the IVR + speech-command aspects of claims 1, 5 (the "interactive voice response application" and speech-command selection).
• Potential §102 mapping: Anticipates only the IVR/speech-command fragments of claims 1 and 5; does not disclose web retrieval, URL records, grammars tied to web content, or content extraction. Realistically an "A"-type background reference.

6. US 5,054,054 A — "Voice applications generator"

• Full citation: US Patent 5,054,054 A; assignee IBM; granted Oct. 1, 1991 (in the visible "Citations" list).
• Description: A generator for voice applications that lets developers define speech-recognition grammars and voice-driven application flows — relevant to the "recognition grammar" concept in claims 1 and 5.
• Potential §102 mapping: Discloses recognition grammars and voice-driven application generation, but not the user-assigned-grammar-to-URL/web-content linkage or content extraction. Background/"A" reference.

7. US 5,365,574 A — "Telephone network voice recognition and verification using selectively-adjustable signal thresholds"

• Full citation: US Patent 5,365,574 A; assignee VCS Industries, Inc.; granted Nov. 15, 1994 (in the visible "Citations" list).
• Description: Telephone-network speech recognition with adjustable thresholds — relevant to the speech-recognition-engine/telephony interface elements of claims 1, 2, 5, 7.
• Potential §102 mapping: Limited to speech recognition over telephony; no web/content-extraction elements.

8. US 4,930,150 A — "Telephonic interface control system"

• Full citation: US Patent 4,930,150 A; assignee First Data Resources Inc.; granted May 29, 1990 (in the visible "Citations" list).
• Description: A voice-controlled telephonic interface for operator/services selection — early IVR/voice-command art.
• Potential §102 mapping: Background for the voice-command/IVR aspects of claims 1 and 5 only.

9. Other references in the visible "Citations" list (lower relevance)

The bulk of the 393 citations are telephony, voice-mail, IVR, and speech-processing patents from the 1970s–1990s (e.g., US3,728,486 "Voicegram service"; US4,371,752 "Electronic audio communication system"; US4,757,525/4,761,807 (VMX voice-command/voice-authentication); US4,776,016 "Voice control system"; US4,850,170 "Real time computer speech recognition system"; US5,054,054 above; US5,177,958 "Integrated voice messaging/voice response system"; US5,195,086; US5,325,421; US5,365,574). These are best characterized as §102/§103 background for the speech-recognition and IVR limitations, not standalone anticipations of any independent claim, because none discloses the combination of user-defined recognition grammar + URL + content-descriptor-file/named-object web extraction + speech-synthesis output.

10. References in the family's "Families Citing" list — not prior art to '981 (excluded)

Patents such as US6,775,264 B1 (Webley Systems), US6,904,450 B1 (Geodata), US6,937,986 B2 (Comverse), US2001/0017855 A1 (Matsushita) appear in Google Patents' "Families Citing this family" section — i.e., they cite the Parus family, so they are later work, not prior art cited by the '981 patent. (Caveat: US6,775,264 B1, filed Mar. 3, 1997 and granted 2004, could qualify as §102(e) art against the 2000 priority date if it were actually on the '981 reference list; I could not confirm it is among the 393 citations. It is the same Webley/Parus ecosystem and would be a strong §102(e) candidate if present.)

---

D. Bottom-line §102 assessment

Key takeaway: No single pre-2000 reference in the confirmed record appears to disclose the full combination of the independent claims — particularly (i) a user-assigned recognition grammar stored with a URL in a user-defined record, and (ii) content extraction from only the relevant web-page portion using a content-descriptor file and a named object. The two closest anticipatory candidates are US 5,819,220 and US 5,884,266 (both pre-2000 voice-browser systems), with claim 10 the most vulnerable of the independent claims if "named object" extraction is read broadly.

E. Unresolved items (for a complete analysis)

1. Full 393-reference list — the retrieved record was truncated; the complete list (and any post-2000 §102(e) references filed before Feb. 4, 2000) should be pulled from the USPTO PatentCenter file wrapper or the issued PDF.
2. Examiner designations (X/Y/A) on the PTOL-892 — not accessible in this session.
3. Filing date of US 6,233,318 B1 — needed to confirm §102(e) status against the Feb. 4, 2000 priority date.
4. Whether US 6,775,264 B1 (Webley) is among the 393 citations — if yes, it is a significant §102(e)/§103 reference given the shared Webley/Parus lineage.

I'll first try to firm up the identities of the PTAB-tested art (Wise/Woods/Dasan/Uppaluru) so the §103 combinations rest on named references, then build the analysis on the Prior Art section's confirmed references.

That gives me the actual PTAB-tested references. Let me confirm their bibliographic identities before mapping them onto the '981 claims.

Let me confirm the identities of Woods and Dasan so the combinations rest on named references.

Obviousness Analysis — US 10,320,981 B2 under 35 U.S.C. § 103

Scope note. This analysis is built on (i) the Prior Art section already generated for this patent, and (ii) the PTAB art identified in the PTAB section of this same page, which I have now tied to bibliographic identities. I have not re-verified the underlying exhibits myself; where a mapping is my extrapolation rather than a Board finding, I say so. Nothing here is legal advice or a formal invalidity opinion.

---

A. Threshold issue: which critical date governs § 103?

Because the '981 chain's earliest filing predates March 16, 2013, pre-AIA § 103(a) applies. The Graham factors (scope/content of prior art; differences; PHOSITA level; secondary considerations) control with the KSR gloss.

The critical § 103 question is whether the Feb. 4, 2000 provisional date holds. It almost certainly does not, for the whole family:

• The chain is 15/707,951 ← 15/193,517 (= the '314 patent) ← 12/787,801 ← 11/711,773 ← 09/777,406 ← Provisional 60/180,343. Every member passes through application 09/777,406.
• In IPR2022-00805 the Board entered a show-cause order finding that the patent at issue "was entitled to claim priority to a provisional patent application where application was filed one day after the provisional application expired, breaking the priority chain" (Banner Witcoff PTAB Highlights). The arithmetic is consistent: the '343 provisional was filed 2000-02-04; twelve months expired on a weekend, extending the § 119(e) deadline to Mon. 2001-02-05 under 35 U.S.C. § 21(b)/37 C.F.R. § 1.7; 09/777,406 was filed 2001-02-06, one business day late.
• Parus ultimately conceded the break (counsel: "Parus drops all challenges to the status of Woods as prior art").
• Critically, the § 119(e) defect is a timing defect, not a claim-drafting defect peculiar to the '314. It attaches to 09/777,406 itself, and therefore to every continuation in the family, including '981.

Consequence for § 103: the effective filing date for '981's claims falls to the '406 filing date, 2001-02-06 (assuming the otherwise-intact § 120 chain). That converts the prior art window from "before Feb. 4, 2000" to "before Feb. 6, 2001," and admits the entire 2000–2001 voice-portal art — most importantly Woods (US 6,510,417, filed 2000-03-21) as § 102(e) art.

---

B. PHOSITA as of the critical date

A person of ordinary skill in February 2000 / February 2001, in this field, would have had a bachelor's degree in CS or EE plus ~2 years of experience (or an M.S.), and working familiarity with: (1) server-side telephony/IVR platforms and ASR engines (the '981 specification names Nuance; the parent's examiner used Wise + Rhie); (2) HTML/HTTP document retrieval and parsing; and (3) TTS rendering (the specification names Lernout & Hauspie). That artisan knew that the technical problem was a packaging problem — how to let a telephone caller reach a small, user-selected slice of a Web page — and all three component technologies were off-the-shelf by 1999.

---

C. The common claim core

Claims 1, 5, 10, 19 and 24 are drafted differently but reduce to eight elements:

The novelty, if any, lives in E3–E5 and E7 only. Everything else is admitted prior art in the specification itself.

---

D. Primary combination: Wise (US 5,884,262) in view of Woods (US 6,510,417)

This is Ground 1 of IPR2022-00805 against the '314, which carries the same specification and substantially the same claim architecture.

• Wise, US 5,884,262 — "Voice interface for electronic documents" (Wise et al.), granted 1999-03-16, filed 1997. It is § 102(b) art and is therefore prior as of either candidate critical date. It teaches a telephone user accessing and navigating electronic documents audibly; parsing the document "to identify any text and any links"; building a hierarchical data structure; and generating and speaking an audio representation of the document (with user control over how much descriptive information is read). Wise is the backbone the Board used for "most limitations."
• Woods, US 6,510,417 — "System and method for voice access to Internet-based information" (Woods, Carriere, Quillet; AOL), filed 2000-03-21, granted 2003-01-21. Woods' voice portal (UI 110, database 170, update engine 160) stores user-relevant Internet information; spiders pull web pages and "rules are applied to identify characteristic information or attributes"; and — decisively for E7 — "a search page is found and patterns are used to isolate the area on the page containing relevant information," with "an appropriate form ... special routines are invoked to extract actual data and information."

Mapping onto the E-elements:

The Board's own summary of the two-pronged rationale is the motivation-to-combine statement a challenger should quote:

"although Petitioner relies on Woods or Dasan for certain alternative positions, both sets of challenges rely on Wise for most limitations and on either Woods or Dasan primarily for those aspects of the claims that relate to separation and retrieval of a portion of information."

Supporting motivation evidence credited by the Board: a POSITA "would have been motivated to implement Woods's data organizing tool 1025 as a clipping engine into Wise because it would enhance the accuracy and efficiency of retrieving the requested information," and to fold Woods' website-search functionality into Wise's parser "to deliver information garnered from the Internet to users." Result: claims 1–26 of the '314 held unpatentable, the Board concluding "Petitioner has shown, by a preponderance of the evidence, that claims 1–26 are unpatentable."

---

E. Substitute / supplemental combinations

E.1 Wise + Dasan. Dasan supplies E7's "only a portion" and E4's parameterized request: Petitioner relied on Dasan's form-filling interface in which a user "labels a topic of interest 'by filling in the field ... provided by the fill-in forms feature of HTML or CGI'" and "specifies search terms, case sensitivity, and 'search context(s)'"; and on Dasan's "personal newspapers" generated from user profiles. Google fell back on Dasan precisely "to the extent that Wise does not explicitly disclose separating 'only a portion of the webpage containing information of interest.'" A POSITA would combine them because Dasan's user-configured topic forms produce exactly the narrowed, user-specific page region Wise's audio browser needed in order to avoid reading whole documents aloud over a telephone call — a direct efficiency/accuracy rationale. The Board found claims 1–25 obvious over Wise + Dasan.

E.2 Wise + Dasan + Uppaluru (US 5,915,001). Uppaluru adds the personal-grammar / user-specific command vocabulary element for the claim-26-type limitation; the Board found claim 26 unpatentable over this three-way combination.

E.3 The page's own prior-art section, combined independently of the PTAB record. Even ignoring the IPR entirely, the references already catalogued here combine cleanly:

• Sarukkai, US 5,819,220 (Yahoo!, granted 1998-10-06) — a speech interface to the WWW in which the recognition vocabulary is dynamically boosted from the page being retrieved; maps E1–E3, E6, E8.
• Waldman, US 5,884,266 (IBM, granted 1999-03-16) — an audio interface for document-based information resources that parses document structure so only relevant content is voiced; maps E7.
• US 5,054,054 (IBM, 1991) — a voice-applications generator for defining recognition grammars; maps E3/E5.
• US 5,325,421 (AT&T, 1994) and related telephony/IVR art in the 393-reference list — maps E1/E2 and the IVR-application limitation of claim 1(a).

Motivation: Sarukkai and Waldman are in the identical field (voice access to hypertext documents), address the same deficiency (reading an entire document is impractical over an audio channel), and one supplies what the other lacks (Sarukkai lacks structural isolation; Waldman lacks command-driven retrieval and vocabulary). Post-KSR, combining two references that each address the same problem with predictable, complementary teachings is the paradigm case for obviousness.

Caution — near-identical numbers. The PTAB's Wise is US 5,884,262 ("Voice interface for electronic documents"). The earlier Prior Art section on this page lists US 5,884,266 (Waldman/IBM, granted the same day). These are different patents; do not conflate them in a brief. A third data point — "5,884,232 A ... Wise et al." appearing in the reference list of US 9,971,774 — is almost certainly an OCR transposition of '262, but treat it as unverified rather than silently correcting it.

---

F. Consolidated motivation to combine (why a POSITA would have done this)

1. Same field, same problem, same solution type. Every reference is a voice interface to network/database information. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007), expressly rejects the notion that a combination is non-obvious merely because it was not the inventor's own path.
2. Complementary, non-overlapping teachings with predictable results. Wise/Waldman solve audio rendering; Sarukkai/Woods/Dasan solve which bytes to render. Adding a page-region isolator to an audio document browser is the mechanical application of a known technique to a known structure, with no change in the principle of operation.
3. Recognized, articulated design need. Telephone users cannot scan a page; they need the answer, not the document. Woods' own premise is "quick and accurate voice access to Internet-based information"; Dasan's "personal newspaper" premise is user-filtered content. Both supply the very motivation asserted for the '981 claims.
4. Market and competitive pressure. By 1999–2001 AOL (Woods), Yahoo! (Sarukkai), AT&T, IBM and others were all racing to build voice portals — objective evidence of a known, pressing problem and a finite, predictable set of solutions. See KSR (market demand; "design incentives").
5. Reasonable expectation of success. Each component was independently proven; the specification itself concedes the components were off-the-shelf (Nuance ASR engine; Lernout & Hauspie TTS; Pentium III servers; "content descriptor file ... stored within the web browsing server").
6. No teaching away. Nothing in Wise, Waldman, Sarukkai or Woods teaches away from user-configured page-region retrieval; Wise in fact encourages presenting less than the whole document ("the user may select the amount of descriptive information recited").
7. Read references together, not in isolation. The Federal Circuit's July 2026 Google v. Parus decision (appeal 2024-2381) squarely reversed a Board that "evaluated the Burrows reference in isolation rather than considering how a relevant artisan would have combined its teachings," holding that "references in an obviousness combination must be read together, not in isolation when, as here, the particular combination argued by the petitioner calls for such a reading." That holding helps a challenger, not Parus.

---

G. Dependent claims

Dependents fall with one or two additional, well-known references:

Note that claims 11, 18, 23 and 28 (named object determines beginning/end regardless of position) are drafted to capture exactly the '981 specification's adaptive parsing paragraph — and are correspondingly broader, hence easier to invalidate, than claim 1's "content-descriptor file" limitation.

---

H. Self-collision art if priority is lost

If the '343 provisional benefit is unavailable (Section A), then US 2001/0054085 A1 (Kurganov), published 2001-12-20 from the '406 application, is § 102(e) art as to the later continuations only if a § 120 chain link is also defective; ordinarily a parent's publication is not self-collision art for a proper continuation. Treat this as a secondary ground. The primary effect of the priority break is not self-collision — it is unlocking Woods (2000-03-21) and the rest of the 2000–2001 art.

---

I. Secondary considerations

There is no evidence in the record of a nexus between any specific claim and the asserted commercial success. The predecessor platform was deployed (Webley Systems; Allstate, US West per the inventor's IPR declaration), but (a) commercial success of a platform does not attach to the distinct claim architecture of '981 claims 1/5/10/19/24 without proof of nexus, and (b) the field was crowded with simultaneous independent voice-portal development (AOL, Yahoo!, AT&T, IBM, Parus), which undercuts the inference of non-obviousness. Parus's own litigation history — its counsel's Federal Circuit sanction and the mooted professional-negligence allegations against prior counsel — further suggests that any "unexpected results" record would have to be built from scratch.

---

J. Where the combination is weakest

1. E4's "user-assigned" grammar stored with the URL in a user-defined record. The PTAB never reached claim 10 or claim 24 of '981 — those are '981-specific architectures not present verbatim in the '314's claim set. A challenger must still show a POSITA would have arrived at a user-authored name for a stored page-URL + request record. Woods' rule-writer tool and Dasan's user-labeled topics do this, but the mapping is inferential.
2. "Content-descriptor file" as a separate physical/logical file. Parus's own expert conceded "the term 'content-descriptor file' doesn't appear in the provisional." That cuts both ways: it defeats the priority claim but also means the limitation is what must be met on the merits. Google's reply position was that "there is no requirement that the content descriptor file be a separate file from a webpage," which the Board appears to have accepted for the '314 analogs.
3. Claim 24's unusual format (an apparatus claim that recites microphone/speaker and then defines the remote system) presents an independent § 112 vulnerability that the PTAB never tested; it may be attackable on written description/support rather than on § 103.
4. Estoppel does not help a new defendant, but it also does not hurt. § 315(e)(2) estoppel binds only Google (IPR2022-00805) and Apple (IPR2022-00948) as to Woods/Wise/Dasan/Uppaluru against the '314. A new defendant sued on '981 is free to run the identical combination. Conversely, if the pending Federal Circuit appeal of the '314 FWD (Appeal 2024-1329) reverses, the roadmap weakens materially.

---

K. Bottom line

Combination most likely to render '981's claims obvious: Wise (US 5,884,262) in view of Woods (US 6,510,417), optionally with Dasan for the "only a relevant portion" limitation and Uppaluru (US 5,915,001) for the personal-grammar limitation — the same combination a unanimous PTAB panel applied to the substantively identical sibling '314 claims, resulting in a Final Written Decision that "claims 1–26 are unpatentable." As an independent fallback, Sarukkai (US 5,819,220) + Waldman (US 5,884,266) + US 5,054,054 supplies every element from this page's own prior-art catalogue, with a motivation grounded in the shared problem of voice access to hypertext documents.

The single highest-leverage move is procedural, not technical: establish the broken § 119(e) chain — the same 09/777,406 timing defect that Parus conceded in IPR2022-00805 — which moves the critical date to 2001-02-06 and makes Woods § 102(e) prior art.

Confidence: high on the legal framework and on the existence and content of the IPR2022-00805 findings; moderate on element-by-element mapping of Wise/Woods/Dasan, because I worked from the FWD excerpts and decisional summaries rather than the exhibits themselves, and because the Board's analysis was directed at the '314's claim wording, not '981's. Unverified: the patent number for Dasan (identified in the record only as Ex. 1013, a "personal newspaper"/keyword-search and HTML-fill-in-form reference); the assignee of Wise '262; and the final disposition of Appeal 2024-1329. Note also a metadata inconsistency: the task header states the current date as April 26, 2026, while the system date is September 21, 2026 — this does not affect the analysis, but the July 2026 Federal Circuit decision postdates the former.