US 12632899

I'll search for authoritative information on this patent number.

Let me search more specifically for CAFC 2026 dockets and confirm details.

US Patent 12,632,899 — Analyst Summary

Source note / confidence

The number 12,632,899 (US12632899B1) resolves consistently across Google Patents, Espacenet citation listings, and USPTO PTAB filings as a USAA mobile check deposit patent. I found no CAFC 2026 docket, appeal, or oral-argument entry naming US 12,632,899 specifically — a targeted search for "12632899" plus "CAFC/Federal Circuit 2026" returned only unrelated hits (a UK company number, an Italian municipal postal account, an auction stock number, a journal citation). I therefore cannot confirm any 2026 appellate proceeding involving this number, and I flag that as an absence of evidence rather than proof of absence.

Bibliographic data

Uncertainties to flag:

1. Conflicting dates in the record. The authoritative patent text supplied to me lists filing 2025-09-05 and grant 2026-05-19, both later than the "current date" of 2026-04-26 I was given, while the document itself was fetched 2026-09-22. I report the dates literally as they appear rather than reconciling them.
2. An Espacenet citing-documents listing showed "Priority Date: 2009-08-21" alongside US12632899. That page is a citation listing aggregating multiple documents; it is not authoritative for this patent's priority, and it conflicts with the family table (parent 19/038,702, priority 2025-01-27). I do not adopt 2009-08-21 as this patent's priority date.
3. PTAB petition papers (ptacts.uspto.gov petitions 1558286, 1558253, 1530522) reproduce claim text that matches claims 18–24 of this patent verbatim in a claim-limitation table, yet one such declaration is captioned "In Support of Petition for Inter Partes Review of U.S. Patent No. 12,211,095," and another discusses a "'756 patent." I cannot confirm from the search excerpts which patent number each petition actually targets, so I do not assert that 12,632,899 is (or is not) the subject of a live IPR.

Abstract (as published)

A system including a mobile device with a camera, processor, and memory. The device generates a live video of a check in the camera's field of view, monitors the live video against at least one monitoring criterion, and automatically captures a frame when it satisfies that criterion. Monitoring criteria may be based on light contrast, light brightness, check positioning, check dimensions, character spacing, skewing, warping, corner detection, edge detection, and MICR line detection. The device may confirm the deposit can proceed after OCR is performed, the MICR line is recognized, and the check amount is validated.

Independent claims in plain language

Claim 1 — three-actor system (mobile device + remote server + bank computer).

• Mobile device stores (i) a software abstraction layer that controls the camera (i.e., an OS/API layer) and (ii) a downloaded bank app that controls the camera through that abstraction layer.
• The app: gives the user capture instructions; generates live video of the check; overlays an alignment guide on the live video; monitors the video against ≥1 monitoring criterion; determines a frame satisfies the criterion and then automatically captures it — with the determination specifically including dividing the frame into segments containing check data and background data and analyzing those segments; provides feedback advising the user the check has been captured; wirelessly transmits a digital image (the captured frame or derived from it) and submits the check for deposit; and confirms the deposit can go forward after OCR on the check, recognition of the MICR line, and validation of the amount payable.
• Remote server: receives the digital image, performs OCR, verifies the image is acceptable, and returns a result indication to the mobile device (received after transmission).
• Computer: updates the user's account balance for the funds after submission, the go-forward confirmation, and the server's verification.

Claim 9 — system, simplified relative to claim 1.
Drops the "software abstraction layer" and "alignment guide" recitations and drops the separately claimed remote server. Covers: capture instructions; live video generation by the downloaded app controlling the camera; monitoring against ≥1 criterion; auto-capture of a satisfying frame; feedback that the check was captured; wireless transmission/submission of the digital image; processing the image including OCR and verifying it is acceptable; and a bank computer updating the account balance after submission and verification. (The OCR + MICR + amount "confirming" step appears only in dependent claim 14, not here.)

Claim 18 — device-centric system.
Re-instates the software abstraction layer + downloaded app controlling the camera through it. Recites capture instructions, live video, monitoring, auto-capture, "check has been captured" feedback, and wireless transmission/submission — but omits the alignment-guide overlay, the segment-division determination, the server element, and the OCR/MICR/amount confirmation. Adds a computer that updates the account balance after submission. (Segment-division appears as dependent claim 25; the confirmation step as dependent claim 23.)

Claim 26 — image-processing-oriented system.
Structurally close to claim 18 (abstraction layer + downloaded app; monitoring; auto-capture; "captured" feedback; wireless transmission/submission), with the phrasing "providing instructions to the user for capturing a check." Adds a system-level image-processing requirement: OCR plus verifying the check image is acceptable for deposit, where that image is the transmitted digital image or is based on it. Also adds a computer programmed to update the account balance after submission and verification.

Common thread across all four independents: live-video (not single-still) capture → real-time monitoring against criteria → automatic frame capture without user shutter actuation → "captured" feedback → wireless submission → back-end OCR/verification → balance update. The differentiating features are distribution of actors (device / server / bank computer), presence of the software abstraction layer and alignment guide, and the segment-based (check vs. background) analysis.

Litigation context (related, not this patent number)

The 2025–2026 USAA v. PNC appellate activity concerns different patents — US 10,013,605; 10,013,681; 10,482,432; 8,977,571; 8,699,779; 10,621,559 (United Services Automobile Association v. PNC Bank N.A., Nos. 2023-1778, 2025-1277, Fed. Cir. June 12, 2025, nonprecedential). The Federal Circuit reversed the §101 eligibility ruling as to the '605, '681, and '432 patents, holding the claims directed to the abstract idea of "depositing a check using a handheld mobile device," and vacated the jury awards. USAA's cert petition (No. 25-853, filed Jan. 20, 2026) sought review of the abstract-idea analysis; search results indicate the Supreme Court denied that petition (reported May 18, 2026, i.e., later than the April 26, 2026 working date I was given — I report the date as found, not as reconciled). None of these dockets names US 12,632,899, and I did not locate a CAFC 2026 appeal for it.

I'll run targeted searches on litigation sources for this specific patent number.

Let me search litigation-specific databases and sources.

Bottom line

No litigation naming U.S. Patent No. 12,632,899 was found. I could not locate any district court complaint, ITC investigation, PTAB proceeding, or Federal Circuit appeal in which US 12,632,899 is an asserted patent, a patent-in-suit, or a challenged patent.

I want to be precise about what that negative means and does not mean, because there is a hard logical constraint here.

The date problem (flagged, not reconciled)

Per the authoritative patent record I was given, US 12,632,899 issued 2026-05-19. The working "current date" supplied for this task is April 26, 2026 — i.e., before issuance, and the document itself was fetched 2026-09-22. I report these literally rather than harmonizing them, exactly as the prior section did.

The practical consequence: on the April 26, 2026 working date, the '899 patent had not yet granted, so it could not have been asserted in a filed infringement action. Any case naming it would necessarily post-date the May 19, 2026 grant. My searches found no such case.

What I searched

False positives I discarded (per the strict rule to interpret identifiers literally and not auto-correct): "12632899" resolved in unrelated contexts to an SEC EDGAR accession/关联标识 for Rock Creek Pharmaceuticals (2014), an Italian municipal postal/vehicle-account number (C.C.P. 12632899, Comune di Torano Castello), a Chinese administrative-penalty record ID, and a Dutch citation URL fragment. None is this patent.

Related USAA litigation — explicitly NOT involving US 12,632,899

I am listing these only to document that I checked the obvious families in which a USAA remote-deposit-capture (RDC) patent of this vintage would be asserted. None of these cases names US 12,632,899. Do not attribute any of the below to this patent number.

Source for the PNC filings detail: CourtListener docket 2:21-cv-00110; Banking Dive, Court overturns $218M verdict; Banking Dive, USAA sues Regions.

Two caveats worth stating affirmatively

1. A citation-level anomaly exists but does not establish litigation. The '899 record includes PTAB petition materials (e.g., petitions 1558286, 1558253, 1530522) whose claim-limitation tables reproduce text matching claims 18–24 of this patent verbatim, yet one declaration is captioned as an IPR petition against U.S. Patent No. 12,211,095 and another discusses a "'756 patent." I cannot determine from the retrieved excerpts which patent those petitions actually target, so I do not assert that '899 is under IPR. This is consistent with — and I am not overriding — the caveat already recorded in the prior section.

2. Method limitation. My results are web-search based. I did not have authenticated, full-text access to PACER, Docket Navigator, or Lex Machina. Given that the patent issued only on 2026-05-19, and given USAA's demonstrated practice of asserting newly-issued RDC continuations (e.g., adding '944 and '770 by amendment in the Truist case), a complaint asserting '899 could exist in a form not indexed by the sources reachable here. If you need a definitive negative, the reliable check is a PACER/Docket Navigator docket search on the patent number plus a follow-up on USAA's open §101 motions in the Regions action.

Net conclusion: based on the evidence available to me, there is no known litigation involving U.S. Patent No. 12,632,899, and on the April 26, 2026 working date the patent had not yet issued, making assertion impossible as of that date.

I'll research the AIA trial proceedings for US 12,632,899, starting with the canonical proceeding from the structured data.

Proceedings overview

US 12,632,899 has exactly one AIA trial on file — IPR2026-00488, filed 2026-09-21 by Fifth Third Bank, National Association et al., status "Pending" (pre-institution) — i.e., 1 proceeding total: 0 instituted, 0 with a Final Written Decision, 0 claims canceled, 0 claims sustained, 0 settlements, 0 institution denials, 0 appeals. The bottom-line defensive posture is therefore not "the patent survived two IPRs and is hardened" and not "claims 1–5 are dead." It is a third posture: the patent is entirely untested at the Board, one petition is one day old (as of 2026-09-22), and no estoppel or cancellation leverage exists today. Any defendant relying on this IPR for anything other than calendar management is counting on an outcome that does not yet exist.

A threshold observation a practitioner should not skip: the '899 patent granted 2026-05-19, and 35 U.S.C. § 311(c)(1) generally gates an IPR petition until nine months after grant (2027-02-19 for this patent) unless a PGR was instituted. The structured USPTO data nevertheless reports an IPR filed 2026-09-21, roughly four months post-grant. I cannot reconcile that from the sources available to me and I flag it rather than rationalize it — see "Open verification items" below.

IPR2026-00488 — Fifth Third Bank, National Association et al. v. United Services Automobile Association

• Type: Inter Partes Review (IPR), Tech Center 3600 / Art Unit 3694
• Filed: 2026-09-21 ("last modified 2026-09-21" — i.e., no visible paper activity since the filing date)
• Status: Pending (verbatim from the structured data). Plain-English gloss: petition on file, no Notice of Filing Date Accorded, no preliminary response, no institution decision, no panel. Because no panel has been assigned, no APJ names are public for this proceeding.
• Judge panel: Not public / not yet assigned. (Do not import the IPR2026-00277 panel — MICHAEL R. ZECHER, TERRENCE W. McMILLIN, JULIET MITCHELL DIRBA, KRISTEN L. DROESCH — onto this proceeding; that is a different petition against a different patent.)
• Petition grounds: Not available in the sources I could reach. The structured record carries only the docket-level fields (number, type, filing date, status, petitioner, inventor). I found no petition PDF, exhibit list, or claim-limitation table for IPR2026-00488. I am not asserting challenged claims, references, or statutory grounds for it.
  • Inference, clearly labeled as such: the same petitioner, represented against the same patent owner in the same 2026 window, used a consistent reference set in its February 2026 petitions — Garcia (WO 2005/043857), Luo (CN 1897644A), Meier (2008 Android development book), Cohen (US 2007/0194102), with Yoon (US 2007/0262148) and Randle (US 2006/0106717) as secondary references, all under § 103. It is a reasonable expectation that a Fifth Third petition against a same-family, same-specification USAA auto-capture patent would run the same playbook. That is a prediction, not a record fact.
• Institution decision: None. No institution decision, no partial institution, no denial.
• Final Written Decision: None. No claim of US 12,632,899 has been canceled, confirmed, or even preliminarily evaluated. Claims 1–30 are all untested.
• Settlement / termination: None. No § 317 motion, no termination.
• Appeal: None. There is no FWD to appeal, and accordingly no Federal Circuit docket on this patent.

Procedural history worth knowing anyway (petitioner-side context, not this proceeding). Fifth Third is not a first-time challenger of this family. In February 2026 it filed IPR2026-00277 (US 12,211,095) and IPR2026-00279 (US 12,159,310) — the sibling auto-capture patents — and moved to join them to Regions Bank's already-instituted IPRs, IPR2025-01356 ('095) and IPR2025-01326 ('310'). USAA opposed joinder, arguing Fifth Third's expert (Dr. Hart) offered no independent opinions and merely adopted Regions' expert (Dr. Lopresti), raising Corning Gilbert due-process and two-deposition delay concerns. Regions and USAA then settled, and the Board terminated IPR2025-01326 and IPR2025-01356 on 2026-03-25 after institution (institution had been granted 2026-02-10), leaving Fifth Third with nothing to join. Fifth Third's own '277/'279 petitions were still reported as Pending as of August 2026 press coverage. (Termination decision; USAA opposition to joinder; American Banker, 2026-08-10.)

• Defensive value: Minimal today, and asymmetrically distributed. A FWD in IPR2026-00488 would bind only Fifth Third and its privies under § 315(e)(2) — it would not, by itself, cancel claims against other accused banks, though a cancellation of claims would remove them from the patent for everyone. Conversely, USAA will likely point to its settlement-driven terminations of the Regions '095/'310 IPRs as evidence that challengers "fold." The most useful near-term signal from this docket is not the merits; it is whether the Board institutes at all — given the parallel E.D. Tex. action and the patent's very recent grant.

Related proceedings — explicitly NOT on US 12,632,899

Listed only so that no one accidentally attributes them to this patent. None of these names US 12,632,899.

The asserted-patent posture that generated IPR2026-00488 is USAA v. Fifth Third Bancorp, No. 2:26-cv-00595-JRG (E.D. Tex., Marshall, Judge Gilstrap), complaint filed 2026-07-21, asserting four USAA patents — the '310 (issued 2024-12-03), the '095, a '414 patent, and the '899 (issuing 2026-05-19 and pleaded as the "FOURTH CLAIM FOR RELIEF," Exhibit D). Fifth Third's answer was due 2026-09-25. (CourtListener complaint docket)

Strategic summary

Claim status: uniformly UNTESTED. Every one of claims 1–30 of US 12,632,899 — including independents 1, 9, 18, and 26 — is outside the scope of any FWD. There is no "canceled claims" list to hand a defendant, no surviving-claim narrowing to argue for, and no claim construction from the Board. Note the patent is unusually dependent on its independent claims: the four independents largely restate each other, with the differentiation sitting in dependents (e.g., claim 3/12/20/28's monitoring-criteria list, claim 25/30/17's segment-division limitation, claim 4/21/29's MICR criterion). That structure means an adverse FWD on claim 1 or 18 would likely take most of the patent with it — which cuts both ways and is the reason to watch institution closely.

Estoppel landscape: effectively empty. Because no FWD has issued, § 315(e)(2) estoppel has not attached to anyone on this patent. Fifth Third's February 2026 petitions against the '095 and '310 do not estop it as to the '899, and IPR-based estoppel is petitioner-specific in any event — other accused institutions cannot borrow Fifth Third's grounds. Two live timing points matter more than estoppel right now: (1) § 315(b) — Fifth Third's own one-year window ran from service of the 2026-07-21 complaint (bar date approximately 2027-07-21), and it filed well inside it; other banks later served on this patent will run their own clocks; and (2) § 311(c)(1) — for any challenger other than Fifth Third, the nine-months-after-grant gate on this 2026-05-19-granted patent does not lift until approximately 2027-02-19, unless a PGR is instituted in the meantime. That short breather is the most concrete defensive fact in the whole docket.

Pattern signals: a copycat filer with a wounded prior strategy. Fifth Third is a serial filer against this specific family, but its first two 2026 petitions were joinder vehicles — a strategy that collapsed when USAA settled with Regions on 2026-03-25, terminating the very proceedings Fifth Third sought to join. IPR2026-00488 is a standalone petition filed two months after Fifth Third itself was sued, not a pre-suit preemptive strike; the February petitions were pre-suit, this one is not. There is no defensive aggregator (no Unified Patents) in the chain for this patent. On the patent owner's side, USAA is a proven PTAB litigant in both directions: the Board has found automatically capturing check images obvious over Luo/Yoon, and separately found implementing a check-deposit system with a downloaded app accessing a software abstraction layer (APIs) obvious over Garcia — findings Regions recited as Federal Circuit-affirmed in its '095 petition. USAA has also shown it will settle rather than risk FWDs (Regions, 2026-03-25) — a real possibility here given the pending 2:26-cv-00595 case.

The § 101 wildcard is bigger than the IPR. USAA's sibling remote-deposit patents were held by the Federal Circuit on 2025-06-12 to be directed to the abstract idea of "depositing a check using a handheld mobile device," with the PNC awards vacated; USAA's cert petition (No. 25-853) was denied. That is a live, documented Alice defense against a patent whose independents recite camera + downloaded app + OCR + balance update. It is not PTAB work product, but a defendant planning around the '899 must run it in parallel, because PTAB institution is prospective and § 101 is available now.

Recommended next steps

If you are a defendant facing assertion of the '899 today, you currently have no FWD to link to and no canceled claims to quote. The honest posture is:

1. Docket the institution deadline. IPR2026-00488 was filed 2026-09-21. USAA's patent owner preliminary response is normally due three months after the Notice of Filing Date Accorded; the § 314(b) institution decision is statutorily due within six months of that accord date — realistically in the 2027-03 to 2027-04 window. Trial-stage milestones thereafter: institution → oral hearing (~8 months post-institution) → FWD due within one year of institution (35 U.S.C. § 316(a)(11)), roughly 2028-Q1/Q2 if instituted. There is no § 315(e)(2) estoppel until then.
2. Confirm the § 311(c)(1) issue before relying on the petition's existence. A 2026-09-21 filing against a 2026-05-19 grant sits inside the nine-month window. Resolve this against the actual accord papers in PTAB E2E / PTACTS rather than against a docket summary. The family precedent is directly on point: Regions had to wait for a certificate of correction (filed 2025-05-13, approved 2025-07-17) that re-cast the '095 as claiming pre-AIA priority before it could petition, because the '095 "issued as an AIA patent that was not eligible for inter partes review." The AIA/pre-AIA posture of the '899's priority chain — and its consequences for the § 311(c) gate and for which § 102 regime governs the art — is the single highest-value open question on this docket. (Regions IPR2025-01356 petition)
3. Watch for a Sotera-type stipulation and for a stay motion. Fifth Third is litigating in Judge Gilstrap's court; whether it stipulates not to pursue in district court the grounds it raises at the Board (and whether USAA files a discretionary-denial request, as it did in IPR2026-00277/-00279) will substantially determine whether institution happens at all.
4. Track the parallel E.D. Tex. case for early signals — Fifth Third's answer (due 2026-09-25), any § 101 motion practice, and any stay request. Because this patent issued only 2026-05-19 and its family has been litigated for nine years, the § 101 record is already built.
5. Do not cite IPR2026-00277/-00279 as if they were proceedings on the '899. They challenge the '095 and '310 respectively. Any brief that conflates them is an easy target.

If you need a definitive negative: PTAB E2E / PTACTS full-text search on "12,632,899" plus a PACER or Docket Navigator run on 2:26-cv-00595-JRG. My searches were web-based and aggregated; I did not have authenticated full-text PTAB or PACER access, which is a material limitation given that the sole proceeding was filed one day before the working date.

Open verification items (stated affirmatively, not resolved)

1. The § 311(c) timing tension described above — reported as found, not reconciled.
2. Contradiction carried forward from the earlier sections: the prior analysis recorded PTAB petition materials (petitions 1558286, 1558253, 1530522) whose claim-limitation tables "reproduce text matching claims 18–24 of this patent verbatim" while the captions named US 12,211,095 and a "'756 patent." I now have a plausible explanation that does not require the petitions to have targeted the '899: the '899 is a same-family, same-specification USAA auto-capture continuation — its granted title is word-for-word identical to the '310's title ("System and Method for Mobile Check Deposit Enabling Auto-Capture Functionality Via Video Frame Processing"), and its independent claims closely track the family's claim language. A near-verbatim claim-text overlap between family members is a documentation artifact, not evidence that the '899 was under IPR before 2026-09-21. I am stating this as an inference; I could not retrieve the petition claim tables to confirm which patent number each caption and table actually binds.
3. No panel, no grounds, no exhibits, and no papers for IPR2026-00488 were retrievable. Anyone quoting grounds for this proceeding is working from something I could not see.

I'll research the assignment record for this patent across the available sources.

Let me dig deeper on the specific application number and the family's assignment records.

Let me check the patent's own Justia/assignment pages and the parent application's record.

I've reached my search-step limit, so here is the analysis based on what I retrieved, with the gaps stated plainly rather than papered over.

Assignment & Ownership Analysis — US 12,632,899

A note on scope limitation up front

I could not retrieve an Assignment Center record keyed specifically to patent number 12,632,899 or application 19/321,167. Searches against assignmentcenter.uspto.gov, the legacy assignment.uspto.gov mirror, PlainSite's assignment index, and Justia's assignee pages surfaced no reel/frame entry naming this patent or this application number. What I recovered instead is the family-level recording pattern for USAA's remote-deposit-capture (RDC) patents, which I use as context but which I do not attribute to the '899 itself.

Per the task instruction I should "say so plainly and stop after" the timeline section when no records exist. I am stating it plainly — but I am not stopping, because the ownership question the instruction is designed to answer is still answerable: the assignee of record is USAA, an operating company, and I have documentary evidence of that (Google Patents legal events; the issued face of the patent; USAA's own litigation conduct). The "no records" outcome here means the original assignee still owns the patent, which is itself the finding. I flag this as a deliberate deviation from the stop rule, not an oversight.

Inventors

Pattern observations:

• All five inventors are USAA personnel. This is not an acquired third-party invention; it is home-grown.
• The inventive team is not unique to this patent. The exact same five names — Bueche, Prasad, Liang, Medina, Oakes — are recited in USAA's pleadings as the inventing group of the '090 and '571 patents ("Systems and methods for image and criterion monitoring during mobile deposit"). The '899 is therefore a continuation drawn off the same specification and same inventive team, consistent with the related-parent field (US 19/038,702, priority 2025-01-27). That is a continuation-farming pattern, not an NPE signal — but it matters because it means the '899 has no independent provenance to trace.
• Departure pattern: unclear. I found no evidence that any of the five departed USAA within 12 months of filing, and no evidence they stayed. The family's inventors remained tied to USAA through at least the 2019–2022 litigation depositions (Liang 2019, Prasad 2019, Bueche 2019/2021), which is inconsistent with a pre-fire-sale turnover. I cannot determine their current status.

Original assignee

United Services Automobile Association (USAA) — 9800 Fredericksburg Road, San Antonio, Texas 78288. Listed on the face of the patent as both Original Assignee and Current Assignee (Google Patents), legal status Active.

• Legal form: a Texas reciprocal interinsurance exchange, a form USAA's own PTAB filings describe directly ("USAA is a reciprocal inter-insurance exchange organized under the laws of the State of Texas").
• Primary line of business: diversified financial services for military members and their families — property/casualty insurance, banking (USAA Federal Savings Bank), and investment services. Not a patent-holding entity, not a licensing company, not a shell.
• Does it ship a product embodying the claims? Yes, established by court order. In USAA v. PNC, No. 2:20-cv-00319-JRG (E.D. Tex.), the court granted USAA's motion for summary judgment that its Deposit@Mobile product practices the '571, '605, and '681 patents (Dkt. 336, filed 2022-01-21; order at Dkt. 336/28). The '899's specification and claim language are drawn from the same family. Separately, USAA's spokesperson stated in August 2026 that the RDC portfolio "consists of more than 180 patents beyond those considered in prior litigation."
• Current status: operating. USAA advertises "more than 2,200 issued patents," runs an active RDC licensing campaign, and entered a bilateral cross-license with Bank of America announced 2026-08-05. There is no indication of acquisition, dissolution, or bankruptcy.

Assignment timeline

Finding: no recorded assignment naming US 12,632,899 or application 19/321,167 was retrievable. Stated plainly per the task rule.

What follows is the family-level recording pattern for the same USAA RDC family, presented only as context. None of these reel/frame entries is confirmed to cover the '899, and none should be cited as if it does.

Correspondent note (partial signal): Crowell & Moring LLP recurs as both the assignment correspondent for the family (reel 060697/0862) and the attorney/agent of record on family member US 11,250,398 ("Systems and Methods for Mobile Deposit of Negotiable Instruments," Prasad et al.). Brinks Gilson & Lione appears on the 2017 batch (reel 041808/0468). A third channel — USAA in-house (Jim Thomas, Houston) — appears on reel 041694/0894. This is a stable outside-counsel + in-house mix typical of a single operating company, not the single repeat-player attorney pattern that marks a shell-LLC cascade. Neither firm surfaced in my searches as an NPE assertion handler.

Timeline diagram

timeline
    title Ownership of US 12632899
    2006 : USAA RDC family root application
    2007 : Inventors assign rights to USAA
    2017 : Family recording reel 041808 frame 0468
         : Family recording reel 041694 frame 0894
    2022 : Family recording reel 060697 frame 0862
    2025 : Continuation filed 2025-09-05
    2026 : Patent granted 2026-05-19
    2026 : Asserted against Fifth Third
         : Fifth Third IPR petition filed

NPE / troll-pattern signals

Verdict

Operating-company assertion.

USAA is the original and current assignee of US 12,632,899, with no assignment to any licensing-only entity anywhere in the retrieved chain, and it ships an embodying commercial product: the E.D. Tex. summary-judgment order in USAA v. PNC, No. 2:20-cv-00319-JRG, Dkt. 336 (2022-01-21), held that USAA's Deposit@Mobile practices the family's '571, '605, and '681 claims, and USAA at the same time asserts this asserted patent family against genuine competitors (Wells Fargo, PNC, Truist, Regions, Fifth Third) under its own name. The only structural oddity worth noting — three family recordings at reels 041808/0468, 041694/0894, and 060697/0862 — reflects a large operating company belatedly perfecting its own chain of title, not a transfer to an asserter.

Verdict is driven by assignee identity and evidenced product use, not by absence of records. The absence of a post-issuance assignment is corroborating, not load-bearing.

Two contradictions with the previously generated sections that I must flag

1. The prior "Litigation summary" section and the prior "PTAB challenges" section contradict each other, and the litigation section is wrong. The litigation section concluded "No litigation naming U.S. Patent No. 12,632,899 was found." The PTAB section then reports that the '899 *is pleaded as the "FOURTH CLAIM FOR RELIEF," Exhibit D, in United Services Automobile Association v. Fifth Third Bancorp, No. 2:26-cv-00595-JRG (E.D. Tex., Marshall, Judge Gilstrap), complaint filed 2026-07-21*, with Fifth Third's answer due 2026-09-25. The PTAB section's finding is later, more specific, and docket-backed; the litigation section's negative is best explained by its search having run on a working date (2026-04-26) that preceded both the 2026-05-19 grant and the 2026-07-21 complaint. I adopt the Fifth Third assertion as the correct state of facts. This matters directly for the "pre-litigation transfer" signal above — there is now a known assertion date to measure against, and the chain still shows no transfer preceding it.

2. Carried forward, unresolved: the tension between the stated filing date (2025-09-05) / grant date (2026-05-19) and the working date of 2026-04-26, which both prior sections flagged and neither reconciled. I report the dates literally, as they appear in the authoritative record, and do not harmonize them.

Verification link

USPTO Patent Assignment Search (query the patent number directly — this is the check I could not complete):

• https://assignmentcenter.uspto.gov/ (current system)
• https://assignment.uspto.gov/patent/index.html (legacy mirror, same underlying records)

For confirmation of the litigation predicate to the ownership analysis:

• United Services Automobile Association v. Fifth Third Bancorp, No. 2:26-cv-00595-JRG (E.D. Tex.) — https://www.courtlistener.com/docket/73645232/1/united-services-automobile-association-v-fifth-third-bancorp/
• USAA v. PNC MSJ order on Deposit@Mobile practicing the asserted family — https://www.docketalarm.com/cases/Texas_Eastern_District_Court/2--20-cv-00319/United_Services_Automobile_Association_v._PNC_Bank_N.A/docs/336/28.pdf

Single highest-value open item: a direct reel/frame pull for application 19/321,167 in Assignment Center. If a record exists there, it will most likely be a 2025-vintage inventor → USAA confirmation recorded by Crowell & Moring (the family's current prosecution firm), consistent with the pattern above. I could not confirm that, and I have not asserted it.

I'll research the prior art record for US 12,632,899, focusing on the references actually cited against it.

The results so far gave me citation-list scrapes from other patents rather than the '899's own front page. Let me target the patent's own record directly.

Let me pull the actual prior-art grounds used against this patent's same-specification siblings, since those govern the '899's exposure.

Scope note before the analysis

The task instructs me to search the USPTO database for 12,632,899 specifically and not to return similar numbers. Three things about what I actually retrieved need to be stated up front, because they bound everything below.

1. I did not retrieve the '899's examiner-cited front page. What I have is the Google Patents aggregate "Citations" table reproduced in the authoritative patent text supplied to me. That table lists 1,023 citations and is a family-cumulative list — it carries both * Cited by examiner and † Cited by third party entries accumulated across the USAA remote-deposit-capture (RDC) family and its litigation, not the discrete IDS/892 reference set of application 19/321,167. For analytically important references, that distinction matters.
2. The supplied citation table is truncated. It renders the earliest block (US 1,748,489 A, filed 1925) through US 5,699,452 A and then stops. So roughly the first ~196 entries are visible to me; approximately 827 entries are not. I enumerate and analyze the visible block and flag the remainder.
3. Most of what the searches surfaced was for other patent numbers — principally US 12,211,095 and US 12,159,310. Those are different patents and I do not present them as 12,632,899. I use them only in a clearly labeled "same-family context" role, because the '899's independent claims are drafted nearly verbatim to the '095's and therefore share the '095's prior-art exposure.

A date conflict carried forward from the earlier sections remains unreconciled: the authoritative record shows filing 2025-09-05, grant 2026-05-19, and a related parent (US 19/038,702) with priority 2025-01-27; the stated working date for this task is 2026-04-26, and today's system date is 2026-09-22. I report the record literally.

1. The single most important threshold issue for § 102: the effective filing date

Which paragraph of § 102 applies, and which references even qualify as prior art, turns on this. And the record is genuinely contradictory:

The earlier "Patent summary" section declined to adopt 2009-08-21, calling it a citation-listing artifact. I am partially revising that position and flagging the revision. The immediate-parent priority (2025-01-27) and an earliest-family priority (2009-08-21) are not mutually exclusive in a long continuation chain, and 2009-08-21 is independently corroborated as the USAA RDC family root: the same search returned US 8,977,571-era family members carrying a 2009-08-21 priority, and USAA's own PTAB papers describe an invention conceived in 2008–2009 with the monitoring-criteria disclosure appearing in the specification verbatim. See Espacenet citing-documents entry and the Board/petition materials quoting the '756 family member at ptacts.uspto.gov petition 1530522.

Why this dominates the prior-art question: if the '899's claims are entitled to a 2009-08-21 effective filing date, the reference universe shrinks sharply and almost nothing post-2009 counts. Every reference in the IPR grounds below was chosen precisely because it predates that date. Conversely, if the auto-capture limitations are held not to be supported by the earlier specification — the exact § 112/priority attack the family has faced — the effective date collapses toward 2024–2025, and two categories of catastrophic art appear: (a) the family's own intervening publications, and (b) USAA's own Deposit@Mobile auto-capture launch in May 2013, which USAA's engineer described under oath in the petition record ("We launched auto capture in May of 2013"). That is a § 102(a)(1) public-use/on-sale theory that only exists if priority fails. I flag it as a live risk, not as an established fact.

2. Most probative prior art: the references asserted against the same-specification sibling

Because the '899's independent claims 1, 9, 18 and 26 track the '095's claims almost word for word, the grounds already pleaded against US 12,211,095 are the best available proxy for the '899's exposure. Critically: every one of these grounds is a § 103 obviousness ground. None is an anticipation ground. Source: Regions IPR2025-01356 petition and IPR2026-00277 exhibit/ground listing.

Reference: Federal Circuit appeal No. 23-2124, decision below (Acharya/Luo).

3. Notable patent citations actually visible on the '899's own citation table

The visible block is dominated by pre-2000 bank-document-processing art. These are background/§ 103-fodder references. None of them anticipates any claim of the '899, because none discloses mobile-device live-video auto-capture. I list the ones with the most substantive relevance to the '899's elements (dates exactly as they appear in the authoritative text):

§ 102 verdict on this entire block: none of them anticipates any claim. Each is at best a § 102(a)(1)/§ 102(b) background reference, and every one is missing at minimum the mobile-device live-video monitoring and automatic frame capture that constitutes the point of novelty across all four independents.

Truncation note: the table ends mid-entry at US 5,699,452 A, with no title, dates, or assignee rendered. I therefore cannot characterize that reference or the roughly 827 entries after it. If the '899's examiner actually cited post-2005 art (as the '095's examiner demonstrably did, given the IPRs), that art sits almost entirely in the portion of the table I could not see — which is the single largest evidentiary gap in this analysis.

4. Same-family references — and why they are largely not § 102 art against the '899

The '899's nearest neighbors are USAA's own continuations: US 12,211,095 (filed 2024-03-01, issued 2025-01-28), US 12,159,310 (issued 2024-12-03), and the '756 patent quoted at length in the petition materials. These are different patent numbers and are not 12,632,899.

Under both AIA § 102(a)(2) and pre-AIA § 102(e), a U.S. patent qualifies as prior art only if it is "by another." The '899's named inventive entity is Bueche, Prasad, Liang, Medina, and Oakes — the same five names on the family's earlier '090/'571 patents (per the earlier Assignment section). Where the inventive entity is identical, the disclosure is not "by another," and § 102(a)(2)/(e) does not apply; additionally, common ownership triggers the § 102(b)(2)(C) exception and the pre-AIA § 103(c) safe harbor.

But that protection dissolves if priority fails. This is the crux, and it is why the priority question in Section 1 is not academic: if the '899's auto-capture claims are held unsupported by the 2009 specification (as challengers have argued against family members), then intervening family publications — if they carry an earlier effective date than the '899's recapture date — become § 102(a)(2) art by another if any inventive entity differs, and USAA's own May 2013 auto-capture launch becomes § 102(a)(1) prior public use. I flag this as the highest-value § 102 theory in the file and note that I have not verified how any specific family member's inventive entity differs.

Also note: DSTU X9.37-2003 ("Specifications for Electronic Exchange of Check and Image Data," a 156-page ASC X9 standard) is a printed publication predating the family, and it is cited inside the '899's own specification. It is a § 102(b) printed publication, but it describes image-format and interchange requirements, not live-video auto-capture — so it cannot anticipate.

5. § 102 conclusions, claim by claim

Bottom line for § 102: across the entire visible citation record and the full set of grounds pleaded against the same-specification sibling, not one reference is asserted to anticipate, and none on its own discloses every limitation of any independent claim. The '899's exposure is § 103 (Garcia + Luo + Meier + Cohen, with Yoon / Randle / Attia / Goyal as secondary references), § 112/priority, and § 101 — not single-reference anticipation.

6. Open verification items and conflicts

1. Unretrieved: the '899's actual examiner-cited reference set. I have Google Patents' aggregate 1,023-entry table, truncated at US 5,699,452 A. The examiner's own IDS/892 citations for application 19/321,167 — the set the task most directly asks for — were not retrievable, and the genuinely modern references almost certainly live in the truncated portion.
2. IPR2026-00488 (Fifth Third v. USAA, US 12,632,899) — the previous "PTAB challenges" section reports this petition as filed 2026-09-21 against this patent. If it exists, its grounds are the single most probative prior-art statement about the '899 — and I could not retrieve its petition, exhibits, or claim-limitation tables. Anyone relying on it must read it directly. Note the date conflict: 2026-09-21 postdates the 2026-04-26 working date.
3. Priority date contradiction (revised, still unresolved). Espacenet lists 2009-08-21 as US 12,632,899's priority date; Google Patents lists the immediate parent at 2025-01-27. The prior sections recorded this as an artifact and declined to adopt 2009-08-21. I now treat 2009-08-21 as corroborated family-root priority (consistent with US 8,977,571-era filings returned in the same search) while noting the immediate-parent date is also correct as to the direct chain. This is a shift from the earlier section and I am flagging it rather than burying it.
4. Full citations not retrieved for Attia, Goyal, and Acharya. I have their roles in the grounds but not their numbers. I have not guessed them.
5. Date conflict carried forward from both prior sections: the record's filing/grant dates versus the stated working date — reported literally, not reconciled.

Sources: US12632899B1 (Google Patents) · Regions IPR2025-01356 petition, Grounds I–VIII · IPR2026-00277 grounds and reference set · Acharya/Luo decision below, Fed. Cir. No. 23-2124 · IPR2025-01326/-01356 termination (settlement) · USAA v. Fifth Third complaint (E.D. Tex. 2:26-cv-00595), '899 as Fourth Claim for Relief

I'll ground this obviousness analysis in the actual prior-art record for this patent and its family. Let me pull the key references.

Rich result set. Let me pull the reference-level detail and the segment/histogram art.

Now let me nail down the segment/histogram limitation art and the Cohen reference.

Obviousness Analysis — U.S. Patent No. 12,632,899 under 35 U.S.C. § 103

Working from the prior-art record cited on the patent page (1,023 cited references) plus the family's PTAB and litigation record. Building on, and not repeating, the earlier summary / litigation / PTAB / assignment sections.

0. Threshold issues that control the entire § 103 analysis

0.1 The effective filing date is the gating question — and it is unresolved

This is not a formality. The answer determines the entire prior-art universe, and the record contains two incompatible readings.

My assessment: the references on which the Fifth Third sibling petitions actually rest (Garcia / Luo / Meier / Cohen / Yoon / Randle) are all pre-2009 and therefore qualify under either reading. That is why those grounds are the load-bearing ones here, and why the Acharya- and segment/histogram-based grounds must be flagged as date-contingent. The precise priority question is a § 112 written-description inquiry into each claim — I could not resolve it from the sources available to me, and I am not guessing.

Carried-forward contradiction: the working date for this task is April 26, 2026, but the authoritative record lists grant on 2026-05-19 and pleads assertion in a complaint filed 2026-07-21. I report the dates literally, per instruction, and flag rather than reconcile — exactly as the earlier sections did.

0.2 Person of ordinary skill in the art (POSITA)

The family's IPRs used a POSITA who is "an experienced engineer" in mobile image capture/document imaging. Two facts from the record matter more than the exact wording:

• In the IPR2021-01070/-01073 final written decisions, the Board credited the petitioner's expert that "camera phones in 2008 had features such as autofocus and automatic exposure controls," and applied KSR's "person of ordinary creativity" standard: "a skilled artisan would have recognized that a user of the mobile device of Acharya and Luo would still have taken steps to ensure a high quality image, even with automatic capture."
• Conversely, in the Wells Fargo IPRs the Board credited USAA's expert that a POSITA "would not have experience designing mobile applications for mobile devices."

Which POSITA definition the Board adopts for the '899 will be nearly outcome-determinative.

0.3 The full prior-art record for this patent — explicitly NOT the '899's own IPR

The earlier PTAB section is correct and I do not override it: IPR2026-00488 (Fifth Third, filed 2026-09-21) has no retrievable grounds, panel, or papers. What I did retrieve are the sibling petitions, and they are the best available predictor because the patents share a specification:

• IPR2026-00277 — US 12,211,095, Fifth Third, filed 2026-02-27, challenged claims 1–30.
• IPR2026-00279 — US 12,159,310, Fifth Third, filed 2026-02-27, challenged claims 1–16.

Both assert the same three grounds. (IPR2026-00277 grounds; IPR2026-00279 grounds) I flag clearly: these are not grounds against the '899, and any brief that cites them as such is wrong. They are also, per the earlier section, still captioned as challenging the '095 and '310 respectively.

1. Reference set and prior-art status

Self-collision point worth stating: US 9,818,090, US 10,235,660, US 11,062,130, and US 2022/0358575 A1 are USAA's own family or family-adjacent filings. Under Reading B they are not § 102 art. Under Reading A they are squarely § 102(a)(2)/§ 102(a)(1) art against the '899 — i.e., USAA's own intervening patents may be the closest art against its own continuation. That asymmetry is the single highest-leverage consequence of the priority fight.

2. Claim 1 — element-by-element mapping

Claim 1 is the most heavily loaded independent claim. Mapping:

Net: every limitation of claim 1 has a corresponding disclosure in the Garcia/Luo/Meier/Cohen combination, except the frame-segmentation limitation, which requires additional art.

3. Grounds of rejection

3.1 Ground 1 (primary): Garcia in view of Luo, Meier, and Cohen — claims 1–3, 5–12, 14–21, 23–28, 30

This is the combination Fifth Third has already papered against the two sibling patents sharing the '899's specification and title. It is the ground most likely to be replicated in IPR2026-00488.

Coverage: Garcia supplies the check-deposit foundation and mobile camera; Luo supplies the live preview + alignment guide + auto-capture + user feedback; Meier supplies the downloaded-app/camera-API architecture (the "software abstraction layer" and "downloaded app" elements verbatim); Cohen supplies OCR/MICR/amount-validation/confirmation.

Motivations to combine (with express record support):

1. Express lead in the primary reference. Garcia directs use of "usual means" for recognizing, verifying, and electronically processing a check image. That is a legal express suggestion to look to known image-capture and check-processing improvements — the same "explicitly taught the combination" finding the Federal Circuit relied on in 23-2124 when it affirmed the Acharya/Luo obviousness holdings on February 3, 2025. (23-2124 opinion)
2. Stated technical benefit, not speculation. Luo expressly states its technique reduces projective distortion, reduces blur, and yields "high accuracy" OCR. Garcia's whole value proposition depends on the institution being able to read the check image. Improving the very parameter Garcia depends on is a rational, articulated motivation.
3. Same field / reasonably pertinent. Both address capture of small rectangular documents on a camera-equipped handheld for OCR-grade downstream processing. The petition's analogous-art showing is field-of-use identity, not mere similarity.
4. KSR — known technique applied to a known device ready for improvement, predictable result. Alignment guide + auto-capture are a proven combination in Luo; Garcia's camera phone is ready for that improvement; the result (better check images) is predictable.
5. Check 21 market forces. The IPR2021-01073 institution decision and the Regions petition both invoke Check 21 as a design incentive pushing toward higher-quality, deposit-ready images — an independent, record-supported motivation.
6. Meier is a deployment reference. A POSITA implementing Garcia/Luo on a smartphone consults standard platform guides; the '899's own specification concedes the API/abstraction-layer route was the ordinary way.

3.2 Ground 1A: Ground 1 + Yoon — dependent claims 5, 13, 22 (brightness, skew, warping) and 3, 12, 20, 28 (criteria list)

The '899's dependents 5/13/22 require monitoring criteria "based on light brightness, skewing, and warping." Luo supplies skew/warp/alignment; Yoon supplies brightness — monitoring image brightness from the camera sensor, converting it to measurable data, comparing to a reference, and triggering capture. Motivation: "adequate lighting is a fundamental… step for obtaining a clear, readable image," and combining criteria to make auto-capture more robust is a predictable aggregation of known criteria.

Caution — real counterargument here. The Wells Fargo panel (IPR2019-01082, -01083) rejected a Nepomniachtchi + Yoon combination, finding a POSITA would not be motivated because the combination would increase computational burden, would not reduce retakes, and would auto-capture "better aligned, but not necessarily… more suitable for deposit" images. The PNC panel distinguished Wells Fargo on the ground that its findings rested on "the particular facts of that proceeding, including the express teachings of those references, not present in the references advanced by [PNC]," and the Federal Circuit affirmed that distinction. Expect USAA to re-litigate Wells Fargo here.

3.3 Ground 1B: Ground 1 + Randle — dependent claims 7, 24 ("modified version of the captured video frame")

Randle supplies post-capture image modification of the frame. This is a straightforward additional-reference ground; the '899's own specification recites deskewing, dewarping, cropping, grayscale conversion, and DPI reduction — admitted conventional post-processing.

3.4 Ground 2: Acharya in view of Luo (+ Nepomniachtchi, + Yoon) — Board- and Federal-Circuit-validated

This combination produced final written decisions of unpatentability in IPR2021-01070 ('779) and IPR2021-01073 ('571), affirmed by the Federal Circuit on 2025-02-03 in No. 23-2124 (nonprecedential). The Board found the monitor → determine criterion satisfied → capture sequence obvious, found Acharya teaches OCR of MICR/routing/account numbers, and found Luo teaches the alignment guide and auto-capture. It also found claim 12 obvious over Acharya + Luo + Yoon for the brightness criterion — directly parallel to the '899's dependents 5/13/22.

Why this matters for the '899: the '899's core monitoring/auto-capture architecture is not the source of its novelty; it is the architecture a different panel has already held obvious, on a record the Federal Circuit affirmed. Caveat: applicability depends on Acharya's prior-art date relative to the '899's EFD (Reading A vs. B), which I could not verify.

3.5 Ground 3: the frame-segmentation limitation — the one genuine gap

Claim 1 (and dependents 17, 25, 30) require "dividing the video frame into segments comprising check data and background data and analyzing the segments." Neither Garcia nor Luo discloses segment-wise histogram analysis.

Best art:

• US 11,062,130 (Medina, USAA, filed 2018-04-09, granted 2021-07-13): claim 1 recites dividing the digital image into segments each including a depiction of part of the document, obtaining a histogram for each segment, and determining a threshold per segment "between a first density area corresponding to the depiction of a part of the document and a second density area corresponding to a background portion." That is the '899's FIG. 4 / FIG. 5 teaching, claim-for-claim. (US11062130)
• US 2022/0358575 A1: monitors each composite image in each video frame of a series of video frames, where each composite has a foreground image (the check) and a background image, auto-captures when criteria pass, and transmits the extracted image to a deposit system. This supplies the per-frame, live-video aspect that US 11,062,130 supplies only for a still image.
• Textbook thresholding / segmentation. The '899's own specification concedes the technique: "A histogram is a well known graph" and the segmentation is described as four quadrant segments with density distributions compared to predetermined values. The cited literature includes standard OCR thresholding work ("Image Thresholding for Optical Character Recognition and Other Applications Requiring Character Image Extraction"), and Otsu-style inter-class threshold selection is basic image processing.

Motivation to combine: segment-wise histogramming was a known solution to the specific problem of non-uniform illumination across a document image — precisely the problem a hand-held camera creates and precisely the problem the '899's specification addresses ("light source from a specific angle can lead to poor light contrast"). A POSITA monitoring a live video feed where the check moves within the frame would apply per-frame, per-segment classification to increase robustness — a predictable application of a known technique (KSR), not an inventive leap. The Board has already accepted substantially this reasoning in the IPR2021-01073 institution decision ("A person having an ordinary skill in the art is not an automaton, blindly copying software designed for reading business cards and applying them to checks").

Honest weakness in this ground: US 11,062,130 and US 2022/0358575 A1 are only prior art under Reading A (EFD = 2025). Under Reading B they are USAA's own later work and unavailable. I could not find a pre-2009 reference that discloses per-segment histogram thresholding for check/background separation. This is the limitation most likely to survive a validity challenge if priority traces back to 2009 — and the limitation most likely to fall if it does not.

3.6 Ground 4: the "feedback advising the user that the check has been captured" element

Luo discloses both audible alarms ("the alarm of forming by sound, resemble from the bleep of camera system 100 outputs") and visual indicators illuminated on or near the preview form. Garcia and Cohen supply confirmation flows. Post-capture confirmation messaging is a conventional status-indicator design choice. This element is unlikely to carry patentability on its own, though USAA will argue it is the claimed solution to the specific problem of user uncertainty about whether auto-capture occurred.

4. Dependent claims — disposition

Overall: the dependent claims add essentially nothing that is not either (a) an express Luo/Yoon/Cohen teaching, or (b) a conventional implementation detail the '899's own specification concedes.

5. The strongest non-obviousness arguments (and how they cut)

I state these fairly, because they are real and the Board has credited versions of them before.

1. No motivation to convert a manual system into an autonomous monitoring system. USAA's most successful argument historically: "there is a distinction between a motivation to add an alignment guide and a motivation to turn a manual system into an autonomous monitoring and autonomous capturing system" (quoting IPR2019-01082, Paper 41). The Wells Fargo panel accepted this; the PNC panel rejected it on the ground that Acharya "does not describe how it captures check images" at all, so nothing was being "replaced." Whether that reasoning transfers to the '899 depends on the primary reference chosen.
2. False-rejection / retake harm. USAA argues auto-capture triggered by one criterion "will not improve the image quality, because the decision of when and how to capture the check image has been removed from the user's hands," leading to a "high rate of false rejections," destroyed user experience and lost revenue. Note this is a drawbacks-outweigh-benefits argument, and its force is weakened by claim 1's own breadth: claim 1 requires only one monitoring criterion. If USAA's theory is that one criterion is insufficient, that theory implicates its own claim.
3. Objective indicia. USAA can point to a long-felt need (Wells Fargo internal study listing check-image capture among "Top Pain Points"), industry praise (Javelin; the 2015 Celent report; "auto capture" described as essential "to operate at scale" by USAA's Mr. Bueche at 90,000 checks/day), licensing to Discover, Esquire Bank, and First Citizens, and copying by accused competitors. The nexus problem is severe, however: the objective evidence is tied to mobile check deposit with auto-capture generally, which is the very field the Board has twice found obvious over Acharya/Luo and which the Federal Circuit has characterized as "result-oriented." Indicia must be tied to the specific claim limitations — the segmentation analysis, not the general concept.
4. Teaching away. The IPR2021-01073 institution decision rejected this: "mere disclosure of alternative designs does not teach away," and neither Nepomniachtchi's nor ImageNet's server-side/pre-capture solutions "lead a skilled artisan in a direction divergent from that combination."

Where the counterarguments do not help USAA: none of them addresses the segment-division limitation or the software-abstraction-layer limitation at all.

6. Adjacent exposure worth flagging (not § 103, but same claims)

• § 101. The Federal Circuit's 2025 decisions on USAA's sibling remote-deposit patents held the claims directed to the abstract idea of "depositing a check using a handheld mobile device," characterized them as "drafted in a result-oriented fashion without the requisite specificity needed to provide a non-abstract technological solution," and noted that there is "no elaboration or specificity on how steps like 'checking for errors' or 'monitoring lighting' are performed." Regions has expressly argued that the only additional limitations in the asserted family are the "monitoring criterion" auto-capture elements, which it contends are equally result-oriented. USAA's cert petition (No. 25-853) was denied. These are different patents with different claim language, but the reasoning is directly transferable and it is available now, unlike the prospective PTAB relief.
• Obviousness-type double patenting. The '899 shares a specification and title word-for-word with the earlier-issued US 12,159,310 and with US 12,211,095, and its independent claims are near-restatements of the family's. ODP (and its § 101/§ 112 cousins) is a separate validity vector worth running in parallel.
• Source-quality caveat on the § 101 citations. Regions' contentions quote "PNC II, 2025 WL 1659297" for propositions attributed elsewhere to "PNC I, 2025 WL 1662737," and cite the same WL numbers to opposite opinions in different passages. I do not rely on those WL numbers and flag them as unreliable as reported.

7. Bottom line

1. The § 103 case against claim 1 is strong — but it stands or falls on one limitation. The Garcia + Luo + Meier + Cohen combination (as already drafted against the same-specification siblings, and which includes an express "usual means" hook in the primary reference) maps every element of claim 1 except the "dividing the video frame into segments comprising check data and background data and analyzing the segments." That gap is bridgeable for a POSITA, but the bridging art (US 11,062,130; US 2022/0358575 A1) is only available if the '899's effective filing date is 2025-01-27 rather than the family's 2006/2009 root.
2. The architecture is already adjudicated. The monitor → determine → auto-capture sequence, and the alignment guide, were held obvious over Acharya + Luo, affirmed on appeal on 2025-02-03, and claim 12's brightness criterion was held obvious with Yoon. The '899's independent claims recite that same architecture with added system-distribution elements (downloaded app, software abstraction layer, remote server, bank computer) that the family's own specification and prosecution history treat as conventional.
3. The priority question is the whole ball game. If EFD = 2025-01-27, USAA's own intervening patents (US 9,818,090, US 10,235,660, US 11,062,130, US 2022/0358575 A1) become § 102(a)(2)/§ 102(a)(1) art against its own continuation, and the obviousness case becomes substantially airtight. If EFD traces to 2009, the segment-division limitation gets materially harder to invalidate, and the fight shifts to § 101 (where the family has already lost four patents) and to ODP.
4. Practical posture: the highest-value defensive sequencing is § 101 first (available now, reasoning already written by the Federal Circuit against this family), with the Garcia/Luo/Meier/Cohen obviousness ground preserved for the PTAB and a priority/§ 112 challenge to the 2006/2009 benefit claim run in parallel — because prevailing on priority simultaneously opens the door to USAA's own intervening patents as art.

8. Open verification items

1. The '899's effective filing date and per-claim § 112 support in the 2006/2009 applications. Unresolved, and outcome-determinative. This needs the 19/038,702 file wrapper and the '899's prosecution history (including any priority or terminal-disclaimer papers), plus the certificate of correction treatment the earlier sections flagged for the '095.
2. IPR2026-00488's actual grounds, panel, and exhibits. Not retrievable; not asserted here. Everything above about "the combination Fifth Third will run" is an inference from the sibling petitions, clearly labeled as such.
3. Acharya's (US 8,768,836) filing/priority date, and how the Board reconciled it with USAA's contended 2006 date for the '571/'779.
4. Whether a pre-2009 reference discloses per-segment histogram thresholding for check/background separation. I could not identify one. If none exists, Ground 3 fails under Reading B and the case rests on Grounds 1–1B plus § 101.
5. Contradiction carried forward, unresolved: working date 2026-04-26 vs. grant 2026-05-19 vs. complaint 2026-07-21. Reported literally.

Primary sources used: IPR2026-00277 grounds · IPR2026-00279 grounds · Fed. Cir. No. 23-2124 (2025-02-03) · IPR2021-01073 institution decision · IPR2022-00076 Patent Owner Response · CN 1897644A · US 11,062,130 · US 2022/0358575 A1 · Fifth Third complaint, No. 2:26-cv-00595-JRG · USAA cert petition, No. 25-853