One capability sets IP Copilot apart: prior-art search. It’s the engine under every feature. By modeling how ideas connect across papers, products, and people, we help teams see their landscape—and spot the next breakthrough.
This week we look at a recent PTAB decision. I pulled one at random from the Board’s docket. The case is Appeal 2024-003984 on Application 17/973,274. The decision’s notification date is Aug. 22, 2025; the real party in interest is Oracle International Corporation.
IP Copilot Invalidity Search
To start, we ran an Invalidity Chart (used here to analyze unpatentability of the pending claims). In IP Copilot: Search → Invalidity Chart, enter the application number, and select the claims.
IP Copilot estimates runtime, lets you pick which claims to analyze and how many references to display, then it runs. Setup takes seconds; this job estimated ~20 minutes, so I grabbed a coffee.
I received the completion email shortly after—run finished.
The results showed one very high-relevance reference, several high, and multiple medium hits. The top reference looks particularly strong.
The claim chart aligns each claim with the cited passages and shows specification-support counts (per claim and sub-claim) plus an overall tally.
Toggling citation view highlights the exact text in context; many mappings include multiple supporting citations.
Clicking a citation opens the reference with inline highlights, so you can confirm context and harvest details for a case or report.
Within ~20 minutes, the results suggested a strong §103 case against the pending claims of US application 17/973,274. Now, let’s compare to the actual PTAB decision.
PTAB Decision vs IP Copilot
First, let’s open up the decision. In this case, it’s affirming an examiner’s rejections. The PTAB affirmed all §103 rejections of claims 1–20. (Decision on Appeal; Decision Summary.) If you’d like to review the decision yourself, feel free to review and download it here.
Interesting! The Board did not rely on Zimmerman et al. The applied art was limited to Kirti, Scheidler, and Trenholm—the same set our results emphasized (with Trenholm appearing for one claim).
Nice, first let’s look for the Kirti reference, which was US 2015/0319185 A1 (application US 14/749,522). If we look at Google Patents the application was granted and is published under US 10,063,654 B2.
What do you know, one of the top results, although we surfaced several others as well for Kirti et al.
Next, let’s look for the Scheidler reference WO 2016/177437 A1, which if we look at Google Patents references US15/571,934 as the priority[2]. If we look at the application was granted and is published under US 10,681,060 B2.
Which we also found as our second result —
Overall, great results! The Board affirmed rejections based on three references: Kirti (US 2015/0319185 A1, later issued as US 10,063,654 B2), Scheidler (WO 2016/177437 A1, with a U.S. counterpart US 10,681,060 B2), and Trenholm (US 2016/0034809 A1).
Kirti and Scheidler were applied together against claims 1–16 and 18–20, while claim 17 required Kirti + Scheidler + Trenholm.
Our system surfaced Kirti and Scheidler correctly, but missed Trenholm in this instance. Importantly, Zimmerman et al. was not used in this case — the examiner and Board relied solely on the three references above.
Conclusion
A random pull from the PTAB docket and in minutes, IP Copilot’s Invalidity Chart surfaced the same primary art the examiner and Board relied on (Kirti, Scheidler—and Trenholm for claim 17), mapped claim-by-claim with inline evidence.
We missed one piece on the first pass, but with limitation-focused search it’s fast to close gaps and pressure-test any claim set. The takeaway isn’t that software replaces counsel—it’s that teams can triage faster, focus expert time where it matters, and walk into prosecution or PTAB review with data, not hunches. If you want to see this on your own portfolio (or a competitor’s), try an IP Copilot Invalidity Chart on your next case.
Leave a Reply
You must be logged in to post a comment.