Patent AI is easier to evaluate as judgment criteria and checklists than as feature tours. This page is the entrance to Patenty’s Market Intelligence series: how to choose a tool, how to compare prior art by element, what to freeze before an office-action draft, what tends to drop when you convert a specification into US or PCT form—and what still breaks when you extend the family to EP or CN.
Claim scope, response strategy, and the final wording of rights remain the attorney’s (or agent’s) call. The pieces below are practice frames that support that judgment. They do not promise patentability or allowance.
The dependencies are simple.
I1 builds the eye for source evidence, matter continuity, and human review. I2 applies that eye to prior-art cells. I3 carries the same matrix and claim history into office actions. I4 checks whether limitations survive jurisdiction conversion. I6 adds EP/CN form and practice checkpoints after I4. If you already have a question, jump from the AEO question map (I5).
Problem. General LLMs, search databases, drafting/intelligence AI, and matter workspaces land in the same shortlist.
You get. A checklist—no brand scorecard—that separates tools by passage-level evidence, matter records, OA continuity, jurisdiction conversion, data policy, and attorney final review.
Link. 7 criteria patent attorneys use when choosing a patent AI workspace
Problem. “AI says similar” is not enough to fix differentiation or claim scope.
You get. Rows = invention elements, columns = references, cells = mapping plus a clickable source passage—and a clear split from dispute/licensing claim charts.
Link. Element-by-element prior-art matrix
Problem. When filing context, cited passages, and claim versions live elsewhere, the OA draft becomes another orphan chat.
You get. Freeze facts → open cited passages → align claim history → choose strategy options → then generate; then human review and a record.
Link. Office-action response checklist
Problem. Fluent English can still swallow numbers, optional vs mandatory language, figure references, and dependency chains.
You get. An omission-type checklist and a short post-conversion review protocol. EP/CN extension is I6.
Link. Patent translation US/PCT omission checklist
Problem. Limitations that survived US/PCT can still drop when the family extends to EP or CN—form, language, and claim habits add another layer.
You get. EP/CN form and practice checkpoints on top of I4 limitation survival (not a paste of I4).
Link. After US/PCT — what still breaks in EP and CN conversions
Editor note: EN site category is Market Intelligence (alt: IP Insights). There is no Naver path for EN.
Use the stuck point as the entry.
| Question | Go to |
|---|---|
| General LLM vs search platform vs workspace—what should I evaluate? | I1 buying criteria |
| Are prior-art hits stored as elements + passages? | I2 matrix |
| Does OA drafting share the same matter context? | I3 OA checklist |
| After conversion, do limitations and numbers still match the source? | I4 omission checklist |
| Adding EP/CN or another family jurisdiction—form or practice gaps? | I6 EP/CN |
For short, citable answers first, open the question map (I5).
If the job is vendor evaluation, start with I1. On the translation axis, continue from I4 into I6 when useful. If you want to see how prior-art compare, specification, review, and jurisdiction convert sit in one matter, request a Patenty demo or plans walkthrough. Filing and claim scope still belong to the attorney.
When a child guide is revised or published, note its last modified (or publish) date briefly on this hub. Do not rewrite the hub at length—keep the two-to-three-line cards and links accurate. Do not invent search volume, rankings, or AI-overview share.