US 7036128

I'll search for the specific patent number across USPTO and CAFC docket sources.

Let me dig deeper on the litigation history and the claims.

Let me get the claim text and check the specific IPR/appeal numbers tied to the '128 patent.

US Patent 7,036,128 B1 — Analyst Summary

Source basis: The full text you supplied (Google Patents, https://patents.google.com/patent/[US7036128](/patent/US7036128)/en) plus targeted web searches on the litigation/PTAB/CAFC record. I did not have direct access to USPTO PatentCenter/PAIR or to the Federal Circuit's docket system in this session, so the litigation facts below come from secondary docket aggregators and are flagged where uncertain.

1. Bibliographic data (high confidence — corroborated by the full text and search results)

Abstract (verbatim from the patent): A highly mobile, ambient computing environment is disclosed for serving a knowledge worker away from their desk… "An Open Agent Architecture is used to incorporate elements such as GPS agents, speech recognition, and opportunistic connectivity among meeting participants. Communication and cooperation between agents are brokered by one or more facilitators, which are responsible for matching requests, from users and agents, with descriptions of the capabilities of other agents… Extreme flexibility is achieved through an architecture organized around the declaration of capabilities by service-providing agents, the construction of arbitrarily complex goals by users and service-requesting agents, and the role of facilitators in delegating and coordinating the satisfaction of these goals, subject to advice and constraints that may accompany them."

2. Plain-language overview of the claims

Caveat on claim text (important): The full text you supplied contains the specification, abstract, and figures list but not the numbered claims. I therefore cannot quote the exact independent-claim language, and I will not reconstruct it from memory. What follows is derived from the prosecution/litigation record that quotes claim content.

What the record establishes:

• The patent issued with 45 claims (1–45), allowed Dec. 29, 2005 (per the IPA v. Google complaint, https://images.law.com/contrib/content/uploads/documents/394/1640/IPA-Google-Complaint.pdf).
• During prosecution the examiner rejected claims 1–21 and claim 45 under 35 U.S.C. § 101 for reciting "a collaborative community of distributed electronic agents that make up a mobile computing environment" that were deemed non-statutory software environments.
• Claim language is characterized in the pleadings as covering:
  • distributed electronic agents in a mobile computing environment, with capabilities registered using an Interagent Communication Language (ICL) that includes a layer of conversational protocol defined by event types and parameter lists;
  • a facilitator agent to coordinate cooperative task completion among the electronic agents;
  • a location agent to ascertain or provide user location information; and/or
  • a mobile computer interface to forward a user request for resource access to a facilitator agent and provide the user with such resource access.

Practical reading of the independent claims (as a group): they are directed to a distributed agent architecture for a mobile user, in which (a) service agents declare/register capabilities in ICL, (b) a facilitator matches user/agent goals to those declared capabilities and coordinates cooperative task completion, (c) a location/positioning agent (e.g., GPS) supplies the user's position, and (d) a mobile user interface submits requests and presents results — i.e., the "ambient computing" use case of the specification.

Likely independent-claim set (inference, moderate confidence): The litigation record suggests at least claims 1, 22, and 41 are independent (they anchor separate groups; dependent claims 23–26 depend from 22, and claim 40 appears in the claim-22 family). Claim numbers and dependency should be verified against the USPTO claim listing, which I could not retrieve.

Asserted / adjudicated claims noted in the record:

• Amazon case: IPA asserted claims 23, 24, 25, 26, 40 (https://archive.org/download/gov.uscourts.ded.61001/gov.uscourts.ded.61001.161.4.pdf).
• IPR2019-00734 Termination Decision (Oct. 15, 2020): PTAB found claims 22, 41, 42, 44, 45 unpatentable.

3. Litigation and PTAB/CAFC posture

District court (D. Del., before Judge Richard G. Andrews): IPA Technologies (a Wi-LAN/Quarterhill subsidiary) asserted the "Cheyer patents" — US 6,851,115 ('115), US 7,069,560 ('560), and US 7,036,128 ('128) — against multiple defendants:

• IPA v. Amazon.com / Amazon Digital Services, 1:16-cv-01266; IPA v. Microsoft, 1:18-cv-00001; IPA v. Google, 1:18-cv-00318 (see https://npe.law.stanford.edu/case/199001).
• Additional suits: Dell 1:16-cv-00947, DISH Network 1:16-cv-01170, Kyocera 1:17-cv-00263.
• In IPA Techs., Inc. v. Amazon.com, Inc., 352 F. Supp. 3d 335 (D. Del. 2019), the court denied the motion to dismiss as to the '115, '560, and '128 patents (holding them not directed to an abstract idea) while granting it as to three other patents (https://www.courtlistener.com/opinion/[7333543](/patent/7333543)/ipa-techs-inc-v-amazoncom-inc/).
• Claim construction for terms in the '115, '560, and '128 patents issued May 21, 2020.

PTAB: A family of IPRs was filed in 2019 against the '128 patent: IPR2019-00733 and IPR2019-00734 reached Final Written Decisions; IPR2019-00735, -00736, -00738, -00739, -00740 were filed with various outcomes (not instituted on procedural or merits grounds). In IPR2019-00734, the Board terminated with a finding that claims 22, 41, 42, 44, and 45 are unpatentable (https://www.docketalarm.com/cases/PTAB/IPR2019-00734/...).

Federal Circuit (2026): The only 2026 Federal Circuit matter I could confirm involving IPA Technologies is IPA Technologies Inc. v. Google LLC, No. 2024-1246 (consolidated with 2024-1247), a Rule 36 summary affirmance entered January 14, 2026 (panel: Chief Judge Moore, Judges Dyk and Taranto), arising from PTAB IPRs IPR2019-00728 and IPR2019-00731 (https://www.cafc.uscourts.gov/opinions-orders/24-1246.RULE_36_JUDGMENT.1-14-2026_2632664.pdf; oral argument heard Jan. 12, 2026).

⚠️ Uncertainty flag: I could not confirm that IPR2019-00728 / -00731 involve the '128 patent specifically. The numbering suggests they may target the sibling Cheyer patents (e.g., '115/'560). Separately, the patent record lists earlier Federal Circuit case numbers 21-1438, 21-1439, and 22-1193 in the '128 family, but I found no 2026 Federal Circuit docket captioned to US 7,036,128 in this session. I also found no results at all for a literal search on "7036128" alone. Treat the mapping of any specific 2026 appeal to the '128 patent as unverified.

4. Bottom line

US 7,036,128 B1 is an SRI International patent (now held by IPA Technologies) by Adam Cheyer and Luc Julia, filed Aug. 9, 2000, issued Apr. 25, 2006, claiming priority to Jan. 5, 1999, and expired June 19, 2022. It claims a distributed-agent ("Open Agent Architecture") system that registers agent capabilities in an Interagent Communication Language (ICL) with a conversational-protocol layer of event types and parameter lists, uses a facilitator agent for cooperative task delegation, and incorporates location (GPS) agents and a mobile user interface for an "ambient computing" environment. Its claims were heavily litigated (D. Del. 2016–2018 suits and 2019 IPRs) and materially narrowed/challenged at the PTAB; the Cheyer-patent family was still producing Federal Circuit activity in January 2026 (Rule 36 affirmance in IPA v. Google, No. 24-1246), though I could not tie that particular appeal to the '128 patent.

Recommended verification steps (not performed here): pull the official claim set from USPTO PatentCenter for application 09/475,092; and query the CAFC docket and PTAB API (IPR2019-00733/-00734) directly to confirm the exact independent-claim text and the precise 2026 appellate linkage.

I'll search for litigation involving US patent 7036128.

Let me dig into the specific cases and their outcomes.

Litigation Report — U.S. Patent No. 7,036,128

Patent: U.S. 7,036,128 B1 — "Using a community of distributed electronic agents to support a highly mobile, ambient computing environment" (inventors: Luc Julia, Adam Cheyer; original assignee SRI International Inc.; current assignee IPA Technologies Inc.). Application 09/475,092, filed Aug. 9, 2000; priority to Jan. 5, 1999 (CIP of US 09/225,198); granted Apr. 25, 2006; expired – lifetime (adjusted expiration June 19, 2022).

Bottom line: The '128 patent was asserted in three parallel Delaware district-court actions brought by IPA Technologies Inc. (a Wi-LAN/Quarterhill affiliate) against Amazon, Microsoft, and Google, and was the subject of at least seven PTAB inter partes review petitions (two Google petitions instituted; the rest denied) and three Federal Circuit appeals. The two instituted Google IPRs ended in final written decisions holding all challenged '128 claims unpatentable, affirmed by the Federal Circuit. Note that patent numbers below are reproduced exactly as they appear in the sources; I have not corrected any.

1. District Court Litigation

Details worth noting:

• Asserted patents. IPA's complaints asserted a common family of six patents: U.S. 6,523,061; 6,742,021; 6,757,718; 6,851,115; 7,036,128; and 7,069,560 (Siri/OAA "Cheyer patents" acquired from SRI International/Siri Inc.).
• '128 claims at issue against Amazon. IPA's Notice of Reduction of Asserted Claims (Aug. 7, 2020, 1:16-cv-01266) expressly narrowed to '128 claims 23, 24, 25, 26, and 40 (along with '115 and '560 claims). The Amazon case terminated on patent-eligibility grounds, not on the merits of infringement.
• Federal Circuit affirmance in the Amazon case. IPA Technologies Inc. v. Amazon.com, Inc., No. 2022-1193 (Fed. Cir.), appeal from 1:16-cv-01266-RGA; Rule 36 judgment of affirmance entered Jan. 18, 2023 (nonprecedential); combined petition for panel rehearing and rehearing en banc denied Mar. 23, 2023; mandate issued Mar. 30, 2023.
  • https://www.cafc.uscourts.gov/01-18-2023-22-1193-ipa-technologies-inc-v-amazoncom-inc-rule-36-judgment-22-1193-rule_36_judgment-1-18-2023_2064674/
  • https://fedcircuitblog.com/wp-content/uploads/2023/02/22-1193-Denial.pdf
• The Microsoft and Google actions were both stayed on Jan. 10, 2020 pending the IPR campaign (per the parties' joint status update, D.I. 82 in 1:18-cv-00318, July 28, 2021).
• Microsoft case resolution. The Microsoft action, after the IPRs and appeals, proceeded only on U.S. 7,069,560 claims 50, 53, 54 (see D.I. 349, Memorandum Order, Apr. 29, 2024, stating "Claims 50, 53, and 54 of the '560 patent are the only remaining asserted claims"). Jury verdict of $242,000,000 on May 10, 2024; the parties then settled and Judge Andrews dismissed the case on June 20, 2024. Reuters reported Microsoft settled "after losing a $242 million jury verdict."
  • https://www.ciplawyer.com/articles/[153865](/patent/153865).html
  • https://www.bayardlaw.com/insights/skiermont-derby-and-bayard-secure-ipa-win-in-microsoft-patent-infringement-case

⚠️ Verification caveats: (a) I could not confirm the exact calendar filing date of the Amazon complaint (only that it is docketed 1:16-cv-01266, filed in 2016). (b) I could not confirm whether the '128 patent remained asserted in the Microsoft action at the time of trial, or how the '128 claims were formally disposed of there; the reported Microsoft trial was confined to the '560 patent. (c) I could not confirm the final disposition of 1:18-cv-00318 (Google) after the July 2021 status update — no dismissal or judgment was located in the retrieved results. Treat items (a)–(c) as unverified.

2. PTAB Inter Partes Reviews (all on the '128 patent)

• Grounds in the instituted Google IPRs: obviousness of claim 45 over Martin (David L. Martin, Adam J. Cheyer, Douglas B. Moran, Building Distributed Software Systems with the Open Agent Architecture, PAAM 1998); and claims 22, 41, 42, 44 over Martin in view of U.S. 5,528,248 (Steiner). IPA's defense centered on disqualifying Martin as prior art "by another" via Rule 132 inventor declarations.
• The joint status update of July 28, 2021 confirms: "The PTAB denied institution of Microsoft's petitions directed to IPA's U.S. Patent No. 7,036,128," and that the Google '128 IPRs (2019-00733, -00734) reached final written decisions on Oct. 15, 2020 "finding that all of the challenged claims of the '128 were unpatentable."
• FWD (IPR2019-00733): https://www.docketalarm.com/cases/PTAB/IPR2019-00733/Inter_Partes_Review_of_U.S._Pat._7036128/docs/10-15-2020-Board/Final_Decision-54-Final_Written_DecisionDetermining_All_Challenged_Claims_Unpatentable35_USC_%c2%a7_318a.pdf
• IPR2019-00839 docket: https://ipverse.greyb.com/ptab-web/cases/case-details/IPR2019-00839

3. Federal Circuit Appeals

• IPA's notices of appeal in IPR2019-00733 and IPR2019-00734 were both filed Dec. 10, 2020 (Skier-mont Derby LLP as patent-owner counsel; Paul Hastings LLP for Google). The appeals challenged, inter alia, the Board's findings as to '128 claims 22, 41, 42, 44, and 45 and the § 103(a) obviousness holdings over Martin and Steiner.
• https://www.cafc.uscourts.gov/03-11-2022-21-1438-ipa-technologies-inc-v-google-llc-rule-36-judgment-21-1438-rule_36_judgment-3-11-2022_1920353/

Related but NOT involving the '128 patent (do not conflate): the Microsoft appeals 2021-1412 et al. and Google LLC v. IPA Technologies Inc., 34 F.4th 1081 (Fed. Cir. May 19, 2022) (appeal 21-1179), and the Jan. 14, 2026 Rule 36 affirmance in appeals 2024-1246/2024-1247, all concern U.S. 6,851,115 and/or 7,069,560 — not 7,036,128.

4. Effect on the '128 Patent

Because the Federal Circuit affirmed the two Google final written decisions (appeals 21-1438/21-1439), the challenged claims 1–12, 20–22, 41, 42, 44, and 45 of the '128 patent were held unpatentable, and the patent expired (adjusted expiration) on June 19, 2022. The claims IPA had asserted against Amazon in the D. Del. action ('128 claims 23–26 and 40) were not among the claims cancelled in the Google IPRs, but the Amazon action ended in a § 101 invalidity judgment affirmed in full by the Federal Circuit — so those asserted claims did not yield enforceable infringement relief in that case.

5. What I did not find

• No ITC (Section 337) investigation involving the '128 patent, and no district-court case outside D. Del. (including no Western District of Texas action) was located in the retrieved results. I cannot rule these out.
• No litigation outcome could be confirmed for 1:18-cv-00318 (Google) after July 28, 2021 — the case was stayed pending appeal, and I did not locate a termination order, judgment, or settlement report.
• Google Patents' "family has litigation" data for this patent lists exactly the district cases and PTAB/CAFC matters listed above (1:18-cv-00318; 1:18-cv-00001; IPR2019-00733/00734/00735/00736/00838/00839/00840; CAFC 21-1438, 21-1439, 22-1193) plus a Darts-IP "first worldwide family litigation" link — which matches my findings.

Sources consulted: Google Patents (US7036128B1), Stanford NPE Litigation Database (patent 7036128; case 1:18-cv-00318), CourtListener/RECAP dockets (1:16-cv-01266; CAFC 22-1193), Unified Patents litigation/PTAB portal links, Docket Alarm (PTAB IPR2019-00733, -00728, -00810), CAFC opinions/orders pages, Reuters via CI Lawyer, and Bayard P.A.

I'll research the PTAB proceedings listed in the structured data, then fill in outcomes via web search.

Key finding: the "Petitioner: Unified Patents" text in the structured block is a Creative Commons attribution line, not the actual petitioner. The real petitioner for the FWD cases was Google LLC. Let me verify the other proceedings' petitioners and details.

AIA Trial Proceedings on US 7,036,128

Critical data-integrity note before anything else: In the structured "PTAB proceedings on file" block, the line reading "Petitioner: 'Unified Patents PTAB Data' by Unified Patents is licensed under a Creative Commons Attribution 4.0 International License" is not the petitioner. It is the CC-BY source attribution for the PTAB data feed. The actual petitioners, verified from the PTAB Final Written Decisions and district-court record, are Google LLC (IPR2019-00733/-00734/-00735/-00736) and Microsoft Corporation (IPR2019-00838/-00839/-00840). Do not brief "Unified Patents" as the petitioner — that reading is wrong.

Proceedings overview

There are seven AIA trial proceedings on file for US 7,036,128: two where all challenged claims were invalidated (IPR2019-00733, IPR2019-00734), five where institution was denied (IPR2019-00735, -00736, -00838, -00839, -00840), zero active, zero settled, and zero in which any claim was sustained. The bottom line for a defendant is strong: the two proceedings that reached a merits verdict canceled every claim they touched — claims 1–12, 20, 21, 22, 41, 42, 44, and 45 — and the Federal Circuit affirmed on 2022-03-11. Compounding that, the patent is now expired (adjusted expiration 2022-06-19), which caps any damages theory to historical conduct. If a demand letter or complaint cites any of the canceled claims, the patent owner is asserting dead claims.

IPR2019-00733 — Google LLC v. IPA Technologies Inc.

• Type: Inter Partes Review
• Filed: 2019-02-26
• Status: Final Written Decision — all challenged claims held unpatentable (Judgment: "Final Written Decision Determining All Challenged Claims Unpatentable 35 U.S.C. § 318(a)")
• Judge panel: Ken B. Barrett, Trevor M. Jefferson, Bart A. Gerstenblith (Administrative Patent Judges)
• Petition grounds: Challenged claims 1–12, 20, and 21. The Board instituted on all challenged claims and on all proposed grounds of unpatentability (Paper 13). The specific prior-art references for -00733 are not reproduced in the sources I retrieved; I did not verify the primary reference for this case and will not guess.
• Institution decision: Instituted (in full). The panel likewise granted Google's request to file a preliminary reply and Patent Owner's sur-reply limited to whether Exhibit 1011 ("Martin") qualified as prior art under 35 U.S.C. § 102(a) (Order, 2019-07-29).
• Final Written Decision: Issued 2020-10-15 (Paper 54). Verdict, quote: "we determine that Petitioner has shown by a preponderance of the evidence that claims 1–12, 20, and 21 of the '128 patent are unpatentable." No claim in this proceeding survived. Oral hearing held 2020-06-04.
• Settlement / termination: None — decided on the merits.
• Appeal: Yes — IPA Technologies Inc. v. Google LLC, Nos. 2021-1438, 2021-1439 (Fed. Cir.), filed 2020-12-22. Decided 2022-03-11: per curiam (Dyk, Schall, Taranto), AFFIRMED (nonprecedential). Opinion: https://www.courtlistener.com/opinion/[6450827](/patent/6450827)/ipa-technologies-inc-v-google-llc/
• Defensive value: Claims 1–12, 20, and 21 are canceled as to the world following affirmance. Any infringement theory built on those claims is not merely weak — the claims no longer exist to be infringed.

IPR2019-00734 — Google LLC v. IPA Technologies Inc.

• Type: Inter Partes Review
• Filed: 2019-02-26
• Status: Final Written Decision — all challenged claims held unpatentable
• Judge panel: The 2019-07-29 joint order in -00733/-00734 was entered "Before KEN B. BARRETT, TREVOR M. JEFFERSON, and BART A. GERSTENBLITH." I did not independently view the -00734 FWD caption; treat the panel as the same three judges, subject to confirmation on PTAB E2E.
• Petition grounds: Challenged claims 22, 41–42, 44, and 45. Ground(s) included § 103 obviousness over Martin and Steiner (Martin, Ex. 1011, being the reference whose § 102(a) status was briefed in both cases). The mapping of claim 45 to a specific ground is not fully captured in my sources — verify before relying on it.
• Institution decision: Instituted. Preliminary-reply/sur-reply briefing on the "Martin" § 102(a) prior-art date was authorized (Order, 2019-07-29).
• Final Written Decision: Issued 2020-10-15. The Delaware case management statement records: "On October 15, 2020, the Board issued Final Written Decisions in IPR2019-00733 and -00734 finding that all of the challenged claims of the '128 were unpatentable." Claims 22, 41, 42, 44, and 45 are therefore all canceled.
• Settlement / termination: None — merits decision.
• Appeal: Yes — same appeals, Nos. 2021-1438/2021-1439, AFFIRMED 2022-03-11.
• Defensive value: Claims 22, 41, 42, 44, and 45 are dead. Note the odd claim numbering (claims 41–45 in a patent whose first IPR only reached claim 21) — the '128 has a large claim set and not every claim was tested (see strategic summary).

IPR2019-00735 — Google LLC v. IPA Technologies Inc.

• Type: Inter Partes Review
• Filed: 2019-02-26
• Status: Not Instituted - Merits
• Judge panel: Not applicable (no institution). Gerstenblith's judge profile lists -00735 among his assigned cases.
• Petition grounds: Not verified. Google filed four '128 petitions on the same day (-00733 through -00736); the two that were instituted covered claims 1–12, 20–22, 41, 42, 44, 45. The claims targeted by -00735 are not confirmed in my sources.
• Institution decision: Denied on the merits.
• Final Written Decision: None.
• Settlement / termination: N/A.
• Appeal: None identified on this proceeding.
• Defensive value: Because institution was denied, no estoppel attached to Google on this petition's grounds (§ 315(e)(2) estoppel requires a final written decision). But note Google subsequently got a merits FWD in -00733/-00734 anyway, so the practical effect is moot for Google. For a new defendant, the denial means the art in this petition was never adjudicated — it remains "clean" art, but the claims it targeted (whatever they are) are still live unless independently canceled.

IPR2019-00736 — Google LLC v. IPA Technologies Inc.

• Type: Inter Partes Review
• Filed: 2019-02-26
• Status: Not Instituted - Merits
• Judge panel: Not applicable (no institution).
• Petition grounds: Not verified (same caveat as -00735).
• Institution decision: Denied on the merits.
• Final Written Decision: None.
• Settlement / termination: N/A.
• Appeal: None identified.
• Defensive value: Same as -00735 — no estoppel, no adjudication, and no claims were cleared by this denial. A denial is neutral; it neither helps nor hurts a later defendant.

IPR2019-00838 — Microsoft Corporation v. IPA Technologies Inc.

• Type: Inter Partes Review
• Filed: 2019-03-20
• Status: Not Instituted - Procedural
• Judge panel: Not applicable (no institution).
• Petition grounds: Not verified — but note Exhibit 11-13 was titled "Declaration of Dr. Henry Lieberman Regarding US Patent 7,036,128," confirming the '128 was the challenged patent and that a § 103 obviousness theory was supported by expert testimony.
• Institution decision: Denied — the structured data codes this as procedural (as opposed to merits). The D. Del. case management statement confirms the outcome directly: "The PTAB denied institution of Microsoft's petitions directed to IPA's U.S. Patent No. 7,036,128."
• Final Written Decision: None.
• Settlement / termination: N/A.
• Appeal: None identified.
• Defensive value: A procedural denial is the least useful denial — it says nothing about the merits of the art or the claims. Microsoft's '128 petitions are dead ends for a defendant seeking a ready-made invalidity record.

IPR2019-00839 — Microsoft Corporation v. IPA Technologies Inc.

• Type: Inter Partes Review
• Filed: 2019-03-20
• Status: Not Instituted - Procedural
• Judge panel: Not applicable.
• Petition grounds: Not verified.
• Institution decision: Denied (procedural). Confirmed by the D. Del. case management statement quoted above.
• Final Written Decision: None.
• Settlement / termination: N/A.
• Appeal: None identified.
• Defensive value: Same as -00838 — no estoppel, no merits ruling, no canceled claims.

IPR2019-00840 — Microsoft Corporation v. IPA Technologies Inc.

• Type: Inter Partes Review
• Filed: 2019-03-20
• Status: Not Instituted - Merits
• Judge panel: Not applicable. Bart A. Gerstenblith's judge profile lists IPR2019-00840 (assigned 2019-03-20).
• Petition grounds: Not verified. Exhibit 11-13 (Lieberman declaration re: the '128) appears in this proceeding as well.
• Institution decision: Denied on the merits.
• Final Written Decision: None.
• Settlement / termination: N/A.
• Appeal: None identified.
• Defensive value: A merits denial is a mild negative signal — the Board did not think the art raised a reasonable likelihood as to the claims targeted. But it is not a patentability holding, and it is not estoppel.

Strategic summary

Canceled vs. sustained vs. untested. Every claim that was adjudicated is canceled. Claims 1–12, 20, 21 (IPR2019-00733) and 22, 41, 42, 44, 45 (IPR2019-00734) were held unpatentable on 2020-10-15 and the judgments were affirmed by the Federal Circuit on 2022-03-11. Claims sustained: none — not a single challenged claim survived. Untested: claims 13–19, 23–40, and 43 do not appear in either Google petition as I retrieved them, and the Microsoft petitions (-00838/-00839/-00840) never reached the merits. Two cautions: (1) I did not confirm the '128's total claim count, so treat the "untested" list as "not shown to have been challenged in the sources retrieved," not as an exhaustive complement; and (2) because the patent expired 2022-06-19, even the untested claims can support only historical damages (subject to the § 286 six-year lookback).

Estoppel landscape. Google is estopped under § 315(e)(2) as to claims 1–12, 20–22, 41, 42, 44, 45 on any ground it raised or reasonably could have raised in -00733/-00734 — but this is academic, since those claims are canceled. Microsoft obtained no FWD, so no § 315(e)(2) estoppel attached to its petitions. For a currently-asserted defendant, the practical positions are: (a) the canceled claims cannot be asserted at all; (b) as to any remaining claims, Google's instituted art is available to the public record and Kessler/collateral estoppel arguments may apply against the patent owner on identical issues, but a fresh defendant must still confront § 315(b) (one-year-from-service) and § 325(d)/General Plastic discretion if it files its own IPR; and (c) because the patent has expired, IPR is largely pointless — the cheaper posture is a district-court invalidity/expiry defense, not an AIA petition.

Pattern signals. The patent owner is squarely in the NPE chain: IPA Technologies Inc. is described in the PTAB docket as "a wholly owned subsidiary of Wi-LAN Technologies Inc., which is a wholly owned subsidiary of Wi-LAN Inc., which is a wholly owned subsidiary of Quarterhill Inc." The '128 was asserted in IPA Technologies Inc. v. Google LLC, No. 1:18-cv-00318 (D. Del., filed 2018-02-26) and in IPA Technologies Inc. v. Microsoft Corp., No. 1:18-cv-00001 (D. Del., filed 2018-01-02); both suits were stayed on 2020-01-10 pending the IPRs. Two separate Big-Tech defendants filed parallel, same-day petition families (Google: four '128 petitions on 2019-02-26; Microsoft: 2019-03-20), which is the classic multi-front challenge pattern — and Google's two winners did the work. IPA was not defensively aggregated here; both challengers were operating defendants, not Unified Patents. The patent owner did appeal aggressively: IPA took the -00733/-00734 FWDs up, argued, and lost. The structured data also lists a second Federal Circuit case, No. 2022-1193, on file for this family; I could not verify its subject matter and will not infer it — check the docket.

Recommended next steps

1. If you are a defendant and the demand letter cites claims 1–12, 20, 21, 22, 41, 42, 44, or 45 — stop and document it. Those claims were canceled by the PTAB FWDs of 2020-10-15 and the cancellations were affirmed on 2022-03-11. The controlling disposition language, verbatim from IPR2019-00733: "we determine that Petitioner has shown by a preponderance of the evidence that claims 1–12, 20, and 21 of the '128 patent are unpatentable." The -00734 court record: "the Board issued Final Written Decisions in IPR2019-00733 and -00734 finding that all of the challenged claims of the '128 were unpatentable." Federal Circuit affirmance: https://www.courtlistener.com/opinion/6450827/ipa-technologies-inc-v-google-llc/ · PTAB FWD (IPR2019-00733, Paper 54): https://www.docketalarm.com/cases/PTAB/IPR2019-00733/Inter_Partes_Review_of_U.S._Pat._7036128/docs/10-15-2020-Board/Final_Decision-54-Final_Written_DecisionDetermining_All_Challenged_Claims_Unpatentable35_USC_%c2%a7_318a.pdf

2. Assert the expiration date. Adjusted expiration is 2022-06-19 and the legal status is "Expired - Lifetime." Any post-expiry "infringement" is legally impossible; damages are capped to activity before that date and further bounded by § 286.

3. If any untested claim (13–19, 23–40, 43) were somehow asserted, the cheapest paths are (i) confirm the claim set against the patent's reexamination certificate and the PTAB records, and (ii) run Kessler / In re Kessler-style estoppel/issue-preclusion arguments off the -00733/-00734 record rather than filing a new IPR — § 315(b) and § 325(d) discretion make a fresh petition unattractive, and there is no live injunctive relief to buy with an IPR.

4. No active proceedings are pending, so there are no institution/FWD deadlines to track. The only open item is the unverified Federal Circuit appeal No. 2022-1193 — pull that docket on PACER/CourtListener before relying on the "all quiet" posture, since it may relate to a sibling patent in the same IPA/Wi-LAN portfolio (the '115 and '560 patents were litigated alongside the '128).

Fabrication guardrails honored: I did not invent proceeding numbers; the seven numbers above come from the structured block and were cross-checked against PTAB/docket records. I did not attribute the FWDs to "Unified Patents." Where I could not verify a ground, a challenged-claim set (IPR2019-00735/-00736 grounds), the -00734 panel caption, the total claim count, or the subject of CAFC 2022-1193, I have said so explicitly rather than filling the gap.

I'll research the assignment record for US 7036128 before reconstructing the chain.

Let me search for the specific assignment records and reel/frame data.

Ownership / Assignment Reconstruction — US 7,036,128 B1

Data-integrity note up front: The assignment events below are grounded in the patent's own Google Patents legal-events record (authoritative copy supplied) and in the litigation/PTAB record. I was not able to retrieve the USPTO reel/frame numbers or the recorded correspondent for either link from the sources available to me during this session. I am flagging those two fields as unverified rather than guessing. They are checkable in minutes at the Assignment Center search (see Verdict).

Inventors

Inventor-departure pattern: Not present. Both inventors remained at SRI for roughly seven years after the 1999 priority filing (they left with the 2007 Siri spin-out), so the tell-tale "all inventors gone within 12 months" fire-sale precursor does not apply here. The relevant exit event is the 2007 Siri spin-out with a non-exclusive license back to SRI, which is what later made the unencumbered remainder of the portfolio saleable in 2016.

Original assignee

SRI International Inc. (333 Ravenswood Avenue, Menlo Park, CA) — named on the issued patent and on the face of the record.

• Primary line of business: independent non-profit research and development institute (spun out of Stanford in the 1940s); primarily advanced technology/systems, biosciences, computing, and education, monetizing through technology licensing and spin-off ventures.
• Did it ship a product embodying the claims? No. The OAA technology was commercialized through a spin-out, Siri, Inc. (2007), which received only a non-exclusive license to the portfolio; Siri was acquired by Apple in 2010. SRI itself did not market an OAA product.
• Current status: Operating. No bankruptcy, dissolution, or acquisition of SRI itself. It is the assignor (not assignee) in the 2016 transfer.

Assignment timeline

The Google Patents legal-events record for this patent shows exactly two ownership-relevant transfers: the inventor→SRI assignment and the SRI→IPA Technologies assignment. Reel/frame and correspondent values are not reproduced here because I could not verify them — see note below.

• Execution date not stated / recorded 2002-03-04 — Reel NNNNNN/NNNN (unverified)

  • Conveyance: Assignment of assignors' interest (inventor-to-employer)
  • Assignor: CHEYER, ADAM; JULIA, LUC
  • Assignee: SRI INTERNATIONAL
  • Correspondent: not retrieved — unverified. No recurring-correspondent flag can be made without the recorded value.
  • Context: Internal/standard employment assignment — the inventors' rights conveyed to their employer, SRI International.

• Execution date not stated / recorded 2016-09-26 — Reel NNNNNN/NNNN (unverified)

  • Conveyance: Assignment (portfolio sale)
  • Assignor: SRI INTERNATIONAL
  • Assignee: IPA TECHNOLOGIES INC.
  • Correspondent: not retrieved — unverified. This is precisely the field that matters most for the NPE analysis below; the correspondent on this 2016 recording should be pulled directly.
  • Context: Transfer-to-asserter. The transfer was publicly announced on 2016-05-09 as WiLAN's second transaction with SRI ("WiLAN Completes Second Transaction with SRI International"; nine patents acquired by wholly-owned subsidiary IPA Technologies Inc.), and the recording followed in September 2016.

• 2022-06-19 — not an assignment; listed only as the patent's adjusted expiration legal-status event. No further recorded ownership changes appear in the legal-events record, meaning IPA Technologies remained assignee of record at expiration.

Timeline diagram

timeline
    title Ownership of US 7036128
    1999 : Inventors employed at SRI International
    2000 : Application filed 09 Aug
    2002 : Inventor assignment to SRI recorded
    2006 : Patent issued 25 Apr
    2016 : SRI sells portfolio to IPA Technologies
    2018 : IPA sues Google and Microsoft
    2019 : PTAB IPRs filed by Google and Microsoft

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT (qualified). The patent moved from an operating research institute (SRI) to a licensing-only entity with no products: IPA Technologies Inc. The qualification: IPA is not an anonymous single-member Delaware/Texas LLC — it is a disclosed, wholly-owned subsidiary of a public-company chain, so the "designed to look unrelated" anonymity tell is absent. The licensing-only character of the transferee is documented (Google Patents event 2016-09-26, assignee IPA TECHNOLOGIES INC.).

2. Known asserter in the chain — PRESENT (strong). The PTAB case caption itself recites the ownership chain verbatim: "IPA Technologies Inc., which is a wholly owned subsidiary of Wi-LAN Technologies Inc., which is a wholly owned subsidiary of Wi-LAN Inc., which is a wholly owned subsidiary of Quarterhill Inc." (e.g., IPR2019-00839; Court of Appeals for the Federal Circuit, Nos. 21-1438 / 21-1439 / 22-1193). Wi-LAN is on the reference NPE list supplied. This is a match against a public NPE list with a specific, citable source.

3. Repeat correspondent across the chain — UNCLEAR / NOT DETERMINABLE. I did not retrieve the recorded correspondent for either recording, so I cannot test the recurrence test. Do not treat this as negative — it is a data gap. The 2016 SRI→IPA recording's correspondent is the single highest-value field still to pull.

4. Cascading transfers — NOT PRESENT. Only two recorded transfers exist across ~16 years (2002 and 2016). There is no chain of consecutive LLC-to-LLC assignments within 24 months on this patent's record.

5. Pre-litigation transfer — PRESENT (strong). The acquisition was announced 2016-05-09 and the assignment recorded 2016-09-26; IPA's first enforcement wave (Dell, HP, Toshiba, Acer, ASUS) followed within months, with the second wave (Amazon, Alco, DISH, HTC, TCL) in December 2016. The chain was consummated immediately before assertion — consistent with venue/standing housekeeping (all cases filed in D. Del., Judge Andrews).

6. Bankruptcy fire-sale — NOT PRESENT. SRI is an operating non-profit with no Chapter 7/11 event; this was a negotiated portfolio sale, not a distressed disposition.

7. Privateering — NOT PRESENT (strict definition). Privateering requires an operating competitor transferring to an NPE that asserts on its behalf. SRI is a research institute, not a market competitor of Google/Microsoft/Amazon, and it retained no apparent enforcement role. The transaction is better characterized as outright monetization of a spin-out-encumbered portfolio (Apple/Siri holds only a non-exclusive license).

8. Defensive aggregator — NOT PRESENT. The chain terminates at an asserting entity (IPA/WiLAN/Quarterhill), not at RPX, AST, LOT, Unified Patents, or OIN. The patent was not neutralized by acquisition into a defensive pool.

Enforcement outcome (context, not an assignment signal): Delaware invalidated the ’061/’021/’718 siblings under Alice; the ’115/’128/’560 survived § 101; extensive IPR activity followed (IPR2019-00733 through -00840; CAFC 21-1438/21-1439/22-1193), and a May 2024 Delaware jury verdict of $242M against Microsoft on ’115. Defense-side participation by RPX/Unified confirms the interpreted NPE status from the other side of the "v."

Verdict

NPE — high confidence.

Two independent strong signals converge. First, a known asserter is in the chain: the patent's assignee of record is IPA Technologies Inc., a wholly-owned subsidiary of Wi-LAN / Quarterhill — a name on the supplied NPE reference list, recited in the PTAB caption (e.g., IPR2019-00839; CAFC Nos. 21-1438, 21-1439, 22-1193). Second, the 2016-09-26 recorded transfer from SRI to the licensing-only IPA was executed immediately before a multi-front assertion campaign (first wave within months; second wave December 2016), which is the classic pre-litigation arrangement pattern. The chain does not terminate at a defensive aggregator, and IPA is not a product company.

Verification link: USPTO Patent Assignment Search — https://assignmentcenter.uspto.gov/ (also https://assignment.uspto.gov/patent/index.html) → search patent number 7036128. Pull the correspondent of record on the 2016-09-26 SRI→IPA recording and the reel/frame for both links; that single query resolves signal #3, which I could not verify.

Caveat (restated): Reel/frame numbers and correspondent names are not asserted here because I could not confirm them. Everything else above is anchored to the patent's own legal-events record and to the litigation/PTAB record cited.

I'll search for the prior-art citations tied to US 7,036,128 and cross-check against the patent's own record.

Let me pull the actual "Patent Citations" and "Non-Patent Citations" sections from the patent record.

Let me target the Justia and Google Patents citation tables directly.

US 7,036,128 B1 — Prior-Art (Citation) Analysis

Task run under the strict rule: every ID below is interpreted literally, and no patent number has been "corrected." The number searched was 7036128, and the hits returned are for US 7,036,128 B1 ("Using a community of distributed electronic agents to support a highly mobile, ambient computing environment," Julia & Cheyer), consistent with the patent text you supplied and the bibliographic section already generated. No similar-but-different number appeared in the results, so the mapping is clean.

⚠️ Source-completeness caveats you must read first

1. The full text you supplied to this session contains the specification, abstract, and figure list — but not the numbered claims. Therefore, when the task asks "which claim(s) a reference potentially anticipates under §102," I can only point to the claim families identified in the earlier-generated section (independent claims anchoring ~claims 1, 22, and 41; dependent family 23–26; claim 40; and the claims held unpatentable in IPR). I will not invent verbatim claim language.
2. I could only retrieve a partial view of the "References Cited" table for the '128 patent (my search access returned the U.S. Patent Documents list truncated around US 6,594,765). The foreign-patent-document list and the "Other Publications" (non-patent literature) list were not fully returned. I flag exactly where the list ends below rather than filling gaps from memory.
3. I could not open USPTO PatentCenter/PAIR directly in this session; the citation list below is taken from the patent's public front-page citation table as mirrored on its Justia page (https://patents.justia.com/patent/7036128).

A. References cited by US 7,036,128 (U.S. Patent Documents — partial, ascending)

These are the examiner/applicant-cited references appearing on the face of the patent. Dates shown are the patent dates (publication), which is what governs availability as §102 art; I note the priority-date implication in the last column.

Reading rule for the dates: because this is a pre-AIA patent (priority Jan 5, 1999; filed Aug 9, 2000), items published before Jan 5, 1999 are unambiguously §102(a)/(b) art. Items published after that date are only art if their own application/priority date qualifies under pre-AIA §102(e) (or if the challenged claims are only supported by the Aug. 9, 2000 CIP disclosure and thus have a later effective date). Several rows above (e.g., US 5,890,123, US 5,948,040, US 5,963,940) sit in this gray zone and must be date-verified before being asserted as anticipatory. I did not verify those underlying filing dates in this session.

B. Non-patent literature cited (partially corroborated, not from the '128 front page)

I could not retrieve the '128 patent's own "Other Publications" sub-list in this session. However, the same disclosures/family are repeatedly cited with these NPL items in later Siri-era patents, and they are the de facto substantive prior art for this invention:

• Cohen et al., "An Open Agent Architecture" (1994), 8 pages — the canonical OAA paper; directly describes capability declaration + facilitator/blackboard coordination.
• Cheyer & Martin, "The Open Agent Architecture" (Autonomous Agents and Multi-Agent Systems, vol. 4, 2001) and Cheyer et al., "The Open Agent Architecture: Building communities of distributed software agents" (SRI presentation, retrieved Feb. 21, 1998).
• Cheyer et al., "Multimodal Maps: An Agent-based Approach," Int'l Conf. on Cooperative Multimodal Communication, 1995 — directly relevant to the multimodal-map claims.
• The specification itself cites Cheyer et al., "Development Tools for the Open Agent Architecture," PAAM 96, London, April 1996, and references KQML as the prior conversational-act language the ICL improves upon. KQML (Finin/Genesereth line of work) is the closest prior art to the claimed "layer of conversational protocol defined by event types and parameter lists."

⚠️ Flag: I could not confirm in this session which of these NPL items appear on the '128 front page versus only on sibling/descendant patents. Treat the NPL list as indicative, not as a verified front-page transcription.

C. PTAB / IPR prior art (the art actually litigated against these claims)

The ipverse/docket records confirm IPR2019-00839 (Microsoft v. IPA Technologies, respondent app. 09/475,092, patent 7036128, Tech Center 2100) was filed Mar 20, 2019 and institution was denied Nov 8, 2019 — the denial reason in the earlier-generated section was characterized as procedural. The earlier section also records IPR2019-00734 ending in a Final Written Decision holding claims 22, 41, 42, 44, and 45 unpatentable, and IPR2019-00733 reaching an FWD. I did not retrieve the specific references relied on in those petitions/decisions in this session, so I cannot name the anticipation art used there without risking fabrication. That is the single most important gap to close (see verification steps).

D. What is most relevant, ranked (my assessment)

Within the retrieved set, the references that most plausibly bear on the independent-claim families (agent architecture + ICL capability registration + facilitator delegation + location/mobile UI):

1. US 6,144,989 (Hodjat et al.) — adaptive service agent; the closest single "agent" reference in the list.
2. US 5,748,974 (Johnson) — multimodal NL interface across applications; maps to the multimodal-interpretation claims.
3. US 5,519,608 (Kupiec) — anaphora/reference resolution; maps to the "me"/"this hotel" resolution disclosure.
4. US 6,226,666 (Chang et al.) and US 6,519,241 (Theimer) — agent + distributed/mobile architecture.
5. US 5,948,040 (DeLorme) and US 6,553,310 (Lopke) — navigation/mapping for the GPS-driven embodiments.
6. The OAA/KQML non-patent literature — the most dangerous art for the ICL/facilitator claims, because it predates the priority date and describes the same architecture.

Legal framing reminder: none of the retrieved references alone reads cleanly on the full claim families (each claim family requires the combination of agent-capability registration in ICL + a facilitator + a location agent + a mobile interface). So the realistic challenge posture for most of these is §103 obviousness in combination, not standalone §102 anticipation — with the possible exceptions of the OAA/KQML NPL against the broadest ICL/facilitator claims.

E. Contradiction / consistency check vs. previously generated sections

• No contradiction with the prior section's bibliographic and litigation facts. The IPR numbers (2019-00734; the 2019-008xx Microsoft filings) and the claim set (unpatentability of claims 22, 41, 42, 44, 45) line up.
• One caveat carried forward: the prior section already flagged that the claim text was not in the supplied full text, and that the 2026 Federal Circuit linkage to the '128 patent is unverified. Both remain true; the claim-mapping column in the table above inherits that limitation.

F. Recommended verification steps (not performed here — hit my search-step limit)

1. Pull the official "References Cited" and "Other Publications" tables for application 09/475,092 from USPTO PatentCenter/PAIR (or the patent's Google Patents page) to complete the truncated list (the table above cuts off at US 6,594,765 and omits the foreign-citation and NPL blocks).
2. Pull the filing/priority dates for the gray-zone U.S. patents (US 5,890,123; 5,948,040; 5,963,940; 6,144,989; 6,226,666; 6,519,241; 6,553,310) to determine whether they qualify under pre-AIA §102(e).
3. Retrieve the IPR2019-00733 / -00734 petitions and Final Written Decisions to obtain the exact anticipation/obviousness references used against claims 22, 41, 42, 44, 45 and the Microsoft IPR2019-00839 petition art.
4. Confirm the exact NPL items on the '128 front page (Cohen 1994 OAA; Cheyer & Martin 2001; Cheyer 1995 Multimodal Maps).

Honest bottom line: I identified the patent-number match correctly and retrieved a substantial portion of its cited U.S. patents, but I could not obtain the complete citation table (foreign + NPL) nor the IPR petitions' art in this session. The most probative prior art for the '128 claims is the pre-1999 Open Agent Architecture / KQML non-patent literature and, among the patent references, US 6,144,989, US 5,748,974, US 5,519,608, US 6,226,666, and US 5,948,040 — with the date caveats noted above. I have deliberately not asserted specific claim-number anticipation beyond the claim families, because the verbatim claim text was not available to me.

I'll pull the actual prior-art citations tied to this patent before building the § 103 analysis.

I have the KQML/facilitator art, but I still need the actual prior-art citation list and the IPR record. Let me dig further.

Obviousness Analysis — US 7,036,128 B1 under 35 U.S.C. § 103

Analysis date: April 26, 2026 (note: the environment's session timestamp reads 2026-09-28; I flag the mismatch rather than silently reconciling it. It does not affect the analysis, since the patent expired June 19, 2022 and all § 103 events of interest are historical.)

Cross-reference note: This builds on the earlier-generated bibliographic, claim-structure, and litigation sections. Where the earlier sections are superseded or need correction, I flag it in §10 rather than restating them.

1. What the "Prior Art" material on this page actually gives me

Two distinct bodies of prior-art material are relevant here, and I want to be precise about which I could verify:

(a) The patent's own PRIOR RELATED ART discussion (part of the authoritative page text you supplied). This is, legally, an admission of what was known, and it expressly brackets four categories:

(b) The page's citations table. The Google Patents page shows "Patent Citations (37)," but the enumerated list of those 37 documents was not present in the text I was given, and my searches did not recover it (I confirmed only the count, via a citing document that reproduces the header). I therefore will not fabricate a list of the 37 cited patents. Everything below rests on (i) the spec's own admissions, (ii) references I could independently verify in search results, and (iii) the IPR record.

Verified prior-art references I can actually ground:

⚠️ Caveat: I did not open the Martin paper or the Steiner patent directly in this session; my characterization of their disclosures is drawn from the PTAB petition summary and the Board's disposition. Treat element-by-element mappings to Martin/Steiner as high-confidence but second-hand.

2. Governing law and the priority-date question that decides everything

Pre-AIA § 102/§ 103 governs (effective filing 1999/2000). The single most consequential § 103 issue here is which filing date each claim gets, because the '128 is a continuation-in-part of Ser. No. 09/225,198 (filed Jan. 5, 1999), with the '128 itself filed Aug. 9, 2000.

• If a claim's subject matter is fully supported by the Jan. 5, 1999 parent → priority date Jan. 5, 1999. Martin (March 1998) is then ~10 months earlier → at most § 102(a)/102(e) art, which must be "by others."
• If a claim depends on new matter added in the CIP — and the entire "highly mobile, ambient computing environment," car-panel, GPS, passenger-recreation and setup-panel disclosure (FIGS. 17–24) is new matter relative to the 1999 parent — the effective filing date is Aug. 9, 2000.

This matters enormously: March 1998 is more than one year before Aug. 9, 2000, so Martin becomes a § 102(b) printed publication. Section 102(b) has no "by others" requirement, which extinguishes IPA's principal defense that Martin is the inventors' own work and therefore not § 102(a)/(e) art. That is almost certainly why the Board was able to reach a merits holding over Martin notwithstanding the inventor declarations, and it is the load-bearing inference in this whole analysis. I flag it as an inference, because the Petition summary states only that IPA had "disqualified Martin by submitting inventor declarations from the '128 patent's co-inventor (A. Cheyer) and another individual" during prosecution of a related application — it does not state the Board's own § 102(b) reasoning.

3. Person of ordinary skill in the art (POSITA)

For a Jan. 1999/Aug. 2000 priority window: a practitioner with at least a bachelor's degree in CS/EE (or equivalent) and roughly 2–4 years' experience building distributed software systems, including at least working familiarity with one agent-communication framework (KQML or equivalent), client-server/distributed-object middleware (CORBA/COM/RMI), and one or more human-interface modalities (ASR, pen, gesture). This is a software-integration POSITA, not a specialist in any single modality — which matters because it makes cross-domain combining (GPS ↔ agent middleware) routine.

4. Claim families (cross-reference; not restated)

Per the earlier section, the asserted/adjudicated claims cluster as: independent claim 22 (+ deps 23–26, 40) and independent 41 (+ 42, 44, 45). IPR2019-00734 challenged 22, 41, 42, 44, 45; the Board found all unpatentable (FWD Oct. 15, 2020; panel Barrett, Jefferson, Gerstenblith; appeal 2021-1438). IPR2019-00733 likewise found all challenged '128 claims unpatentable.

5. The core § 103 combination family

Ground 1 — Martin alone vs. claim 45 (and its independent siblings)

Claim 45's limitations, as the petition mapped them, are the "architecture" limitations, not the "car" limitations:

Motivation / rationale: No combination is needed — this is a single-reference § 103 (or even § 102) theory. The petition's fallback was the KSR "known programming technique" rationale: to the extent the parameter-list refinement is ambiguous, implementing event refinement via parameter lists was a "fundamental concept in software development."

My assessment: This is the strongest ground, because the '128's own specification concedes the architecture was known ("Several agent-based projects have helped to evolve the notion of facilitation," KQML performatives, blackboard tuples) and the asserted novelty reduces to the application of that architecture to a mobile setting. If Martin is § 102(b) art (§ 2 above), Ground 1 is close to dispositive for claim 45.

Ground 2 — Martin + Steiner vs. claims 22, 41, 42, 44 (the location/GPS claims)

This is the combination the task specifically asks about, and it is the cleanest one:

Three independent motivations to combine, each KSR-sufficient:

1. Martin teaches the destination. Martin expressly discloses using the OAA for a "Multimodal Map" application. A mobile map application without the user's current position is an incomplete map application; adding position sensing is the single most predictable enhancement. This is KSR's "use of a known technique (GPS position fixing) to improve a similar device (a map application) in the same way."
2. Same field, same problem, common ownership of the problem. Both references address mobile/portable computing with a PDA; both are designed to present spatial information on a small display. The combination is a "mere aggregation of known elements" performing their known functions with no change in principle.
3. Finite, predictable solutions ("obvious to try"). By 1998, obtaining position for a mobile mapping application had a small, identified set of solutions (GPS, dead reckoning, cell-ID), and GPS was the recognized best. KSR permits a finding of obviousness where the prior art identifies a finite number of predictable solutions and there is a reasonable expectation of success.

Reasonable expectation of success: Both references are software/hardware components with published interfaces; wrapping a GPS receiver as a service-providing agent is exactly the "wrapper around a pre-existing API" pattern the '128 spec itself describes for application agents.

Ground 3 — KQML/Finin + Steiner (the fallback if Martin is disqualified)

This is the most important redundant ground for a validity opinion, because Martin is the inventors' own publication and its prior-art status is the contested issue.

• KQML/Finin supplies: the facilitator as a distinct class of agent performing "maintaining a registry of service names, forwarding messages to named services, routing messages based on content, providing 'matchmaking' between information providers and clients, and providing mediation and translation services"; the layered KQML language (content layer / message layer with performatives / communication layer); parameter lists (:content, :force, :reply-with, :ontology, :sender, :receiver); agent self-registration and de-registration with a local facilitator at startup; and multiple facilitators with symbolic-address routing — i.e., the multi-facilitator/hierarchical-topology concepts of FIGS. 14–15. Sources: the two Finin papers and DTIC ADA346991 above.
• Steiner supplies: the location agent and the map/icon display.
• Motivation to combine: KQML's own literature expressly contemplates hierarchical ontologies and facilitator-agent-servers for scaling (Finin, "KQML as an Agent Communication Language," discussion of DNS-analogous agent-name mapping). A POSITA seeking to build a location-aware assistant on a KQML substrate would naturally register a GPS-provider agent with the facilitator and let matchmaking route location queries — the ICL oaa_Solve pattern with an address parameter is a direct, predictable implementation of KQML's broker/recruit performatives.

Why this ground matters: the '128 spec itself frames ICL as an improvement over KQML ("greater expressiveness than an approach based solely on a fixed selection of speech acts, such as embodied in KQML"). An express improvement-over-a-known-system statement, where the improvement is an orthogonal parameterization of performatives, is the textbook fact pattern for obviousness — the patent has effectively stipulated the difference and characterized it as a design choice.

Ground 4 — KQML/Finin + blackboard art + mobile-agent art (for the facilitator/global-data-store claims)

For any claim reciting the facilitator's global persistent database (spec FIG. 7, item 720) or blackboard-style interaction: the spec concedes that "blackboard architectures typically allow multiple processes to communicate by reading and writing tuples from a global data store," that "a blackboard architecture provides one solution to eliminating the tightly bound interaction links," and that a publicly readable/writable data solvable "acts essentially as a global data repository" enabling "a 'blackboard' style of communication." Combining a KQML facilitator with a shared tuple space is the combination of two references that the applicant himself describes as addressing the same problem (decoupled interprocess coordination), with a predictable result.

Ground 5 — Car/ambient embodiment (FIGS. 17–22, 24) at the levels where those features are claimed

Where claims reach the in-vehicle panels, passenger-private multimedia areas, voice-commanded device control, and speaker-verification-gated setup: motivation is supplied by the art itself (in-vehicle navigation/speech systems and the NUANCE/DCG-NL engines the spec names as pre-existing), combined with Ground 1/2's architecture. The VO*V* three-slot model (FIG. 24) and the multimedia prompting mechanism are, on the record, the weakest-obviousness-evidence items because I have no verified prior-art reference mapping to them; I would not assert obviousness of the VO*V* limitations on this record.

6. KSR rationale ledger (for the record)

7. Anticipation vs. obviousness

Ground 1 (Martin alone) is pleaded as obviousness, but if Martin discloses the mobile interface, the community, the registration and the ICL parameter lists in one reference, the correct characterization may be § 102 anticipation of claim 45 — and a reference that anticipates also renders obvious. I note that the Board's FWD for IPR2019-00734 was rendered under § 103 (all challenged claims unpatentable), not § 102, which suggests the Board found at least one limitation not literally disclosed by Martin and bridged it with the KSR "known technique" reasoning or with Steiner.

8. Rebuttals the patent owner would raise (and how they fare)

1. "Martin isn't prior art — it's our own work" (§ 102(a)/(e) 'by others'). Strong as to a Jan. 5, 1999 priority date; collapses to the extent the asserted new-matter claims get the Aug. 9, 2000 CIP date, making Martin § 102(b) art with no "by others" requirement. This is the pivotal dispute (see § 2).
2. Teaching away / different field. Weak. KQML/Finin expressly designs for a "wide variety of interesting agent architectures" and hierarchical facilitator scaling; the OAA is an instance, not a departure.
3. Claimed improvement is non-obvious (orthogonal parameterization beats fixed speech acts). Weak-to-moderate: KQML already carried a parameterized content/message/communication layer, so the delta is generalization, not invention.
4. Secondary considerations. None of the objective indicia appear well-developed on this record. To the extent IPA could point to the OAA→Siri lineage as commercial success, the nexus is to Apple's later product, not to the claimed combination, and any such evidence would need to be tied to the specific claim limitations; I found no objective-indicia findings in the '128 IPR record I retrieved.
5. § 325(d) / procedural challenges. Several related petitions were not instituted on procedural or merits grounds (−00735, −00736, −00738, −00739, −00740), which is a meaningful asymmetry: the non-institution of sibling petitions is not a merits vindication of the claims, and should not be cited as such.

9. What the adjudicated record already establishes

• IPR2019-00734 (Google LLC v. IPA Technologies Inc., Patent 7,036,128, challenged claims 22, 41, 42, 44, 45; instituted Oct. 17, 2019; FWD Oct. 15, 2020): all challenged claims unpatentable. Petition grounds as summarized: Ground 1 = claim 45 over Martin; Ground 2 = claims 22, 41–42, 44 over Martin + Steiner; motivations expressly tied to Martin's Multimodal Map. Decision: https://www.docketalarm.com/cases/PTAB/IPR2019-00734/ ; petition summary: https://ai-lab.exparte.com/case/ptab/IPR2019-00734/doc/1002
• IPR2019-00733 also found all challenged '128 claims unpatentable (per the joint D. Del. letter, https://www.docketalarm.com/cases/Delaware_District_Court/1--18-cv-00318/IPA_Technologies_Inc._v._Google_LLC/docs/82.pdf).
• On appeal: the Federal Circuit's March 11, 2022 per curiam order in IPA Technologies Inc. v. Google LLC, Nos. 2021-1438, 2021-1439 (appeals from IPR2019-00733 and -00734) — panel Dyk, Schall, Taranto — AFFIRMED. https://www.courtlistener.com/opinion/[6450827](/patent/6450827)/ipa-technologies-inc-v-google-llc/

So the § 103 analysis I have constructed bottom-up (Martin; Martin + Steiner; KQML as fallback) is not hypothetical — it is substantially the analysis the Board adopted and the Federal Circuit affirmed for the claims that were challenged.

10. Flags, corrections, and contradictions

1. Petitioner identity (correction to the earlier section). The Google Patents page's "Petitioner: 'Unified Patents PTAB Data'" string is a data-source/license attribution, not a party. The verified petitioner in IPR2019-00734 is Google LLC, and the D. Del. joint letter states the five instituted IPRs (2019-00728, -00730, -00731, -00733, -00734) were instituted "pursuant to petitions filed by Google." A separate IPR2019-00810 against the '115 patent was filed by Microsoft Corporation.
2. Unverified — IPR2019-00728/-00731 and the '128. The earlier section's uncertainty stands. The D. Del. letter confirms the five Google IPRs collectively challenge "a subset of the issued claims of" the '115/'560/'128 family, but still does not tie -00728/-00731 specifically to the '128. Still unverified.
3. ⚠️ Direct contradiction between a search result and the authoritative patent text. A PatSnap blog post characterizes US7036128B1 as "speech navigation with multimodal error feedback" and assigns the "community of electronic agents for ambient computing" title to US6757718B1. That conflicts head-on with the authoritative document you supplied, whose title, abstract, and claims are directed to the mobile/ambient Open Agent Architecture community with a facilitator. The same post also lists "US6951115B2" (a refrigerant composition) among the "speech & agent" patents — an evident identifier/table corruption. I reject that characterization; per the hierarchy in your instructions, the full patent text controls. Note also that the earlier section quoted a different 2026 Federal Circuit disposition (Rule 36, Jan. 14, 2026, No. 24-1246/1247, Moore/Dyk/Taranto), while the verified appellate outcome for the '128-specific IPRs is the March 11, 2022 affirmance in Nos. 21-1438/21-1439. These may both be real but are different appeals; do not conflate them.
4. Unretrieved: the page's "Patent Citations (37)" list and non-patent-citation list. I could not enumerate them. Any claim-by-claim § 103 chart keyed to prosecution-cited art should be re-run once that list is obtained from PatentCenter/PAIR for application 09/475,092.
5. Martin's § 102(b) status is my inference, not a retrieved Board holding. Verify against the actual IPR2019-00733/-00734 Final Written Decisions (PTAB API / Docket Alarm) before relying on it.

11. Bottom line

The '128 claims are obvious over at most three references, and the strongest combination requires only two. A single prior publication by the patent's own co-inventor (Martin, PAAM 1998) discloses the entire claimed architecture — agent community, facilitator, capability registration, and a parameterized event/ICL conversational layer for a mobile multimodal interface. Adding Steiner (US 5,528,248) supplies the only architectural element Martin lacks (a location agent plus a map with a position icon), and the motivation to combine is supplied by Martin's own disclosed "Multimodal Map" application plus Steiner's known GPS-PDA integration — a predictable, same-field, finite-solution combination squarely within KSR. If Martin's prior-art status is defeated on the "own work / CIP-priority" theory, the combination of KQML/Finin facilitator art + Steiner, and secondarily KQML + blackboard/tuple-space art, independently renders the facilitator, registration, ICL-parameterization, and global-data-store limitations obvious — a route made unusually clean by the applicant's own specification, which expressly concedes KQML facilitation and blackboard coordination as known and frames ICL as an orthogonal-parameter improvement over KQML's fixed speech acts. This conclusion is corroborated by the adjudicated outcome: all challenged claims (22, 41, 42, 44, 45) were held unpatentable in IPR2019-00734 (and in IPR2019-00733), affirmed by the Federal Circuit on March 11, 2022. The genuinely weak spots in the record are the VO*V* three-slot multimodal model (FIG. 24) and the multimedia prompting features, for which I found no verified prior-art mapping and on which I would not opine without the full citation list.