Patent & R&D Operations — Overview
What the add-on is, its 4 sub-groups, who uses it, and the two-stage hybrid AI engine pattern used across the module.
Module Overview
The Patent & R&D Operations add-on — package id patent-rd-ops, domain PATENTRD — is a wired, purchasable package gated by a tenant entitlement and per-user Patent & R&D roles. It covers the full intellectual-property lifecycle for an R&D-heavy organisation: cataloguing a patent portfolio, screening prior art and freedom-to-operate risk before filing, synthesising academic literature to ground R&D specs, and tracking the grants, licensing and valuation side of IP once it exists.
At a glance
- Domain: PATENTRD
- URL prefix:
/patent-rd-ops/… - Governance screens: Roles, IP & R&D Team (staff)
- 16 operational screens across 4 sub-groups (see below)
- Default roles: Chief IP & R&D Counsel, Patent Registry Coordinator, FTO & Prior Art Analyst, R&D Literature Researcher, Grants & IP Valuation Officer
The 4 sub-groups
The add-on’s 16 operational screens are organised into four titled sub-groups, on top of the standard Administration & Access group (Roles, Staff).
| Group | Screens |
|---|---|
| Patent Registry & IP Setup | Patent Portfolio Catalog, Legal Jurisdiction Setup, Technology Taxonomy, Inventor Rights Config |
| Patent Prior Art & Freedom to Operate (FTO) | Patent Collision Scanner, Prior Art Search Engine, FTO Risk Analysis, Claims Matrix Comparer |
| Academic Literature & R&D Synthesis | Literature Synthesis, Research Paper Library, R&D Project Spec Audit, Invention Disclosure Review |
| R&D Grants & Compliance Tracking | Grant & Incentive Tracking, Tech Transfer & Licensing, R&D Tax Incentive Audit, IP Valuation & Royalty |
See Patent Registry & IP Setup, Patent Prior Art & FTO, Academic Literature & R&D Synthesis and R&D Grants & Compliance Tracking for the per-screen breakdown of each group.
Who uses it
Patent & R&D Operations targets in-house IP counsel, patent paralegals, R&D researchers and technology-transfer officers — the people who decide what gets filed, whether it’s safe to ship, and how existing IP gets licensed or monetised. A tenant activates the patent-rd-ops package, then assigns each team member one of the five default roles (or a custom role built from the same capability catalog) as a domain “hat” — see Roles & Capabilities.
Key concepts
Patents & jurisdictions
The Patent Portfolio Catalog is the master ledger of filed/pending/granted patents; Legal Jurisdiction Setup holds the statutory rules and fee schedules per patent office (TÜRKPATENT, EPO, USPTO, WIPO PCT) that govern them.
Prior art & FTO
Prior art search asks “has this been done before?” (novelty); Freedom-to-Operate asks “can we sell this without infringing someone else’s active patent, in this territory?” — two related but distinct legal questions.
Claim charts
A claim chart (Claims Matrix Comparer) maps a target patent’s claim limitations against a product’s features element-by-element — literal match, doctrine of equivalents, or non-infringing — the standard structure lawyers use for infringement analysis.
Literature synthesis
Turns a pile of academic papers into a state-of-the-art summary, key methodologies, benchmarks and technology gaps — grounding an R&D spec in what the literature actually supports before it’s written into a project plan.
Invention disclosures
The formal starting point of a new patent: a researcher discloses an invention, the system scores its novelty and commercial readiness, and an approved disclosure can be promoted directly into a real Patent Portfolio Catalog record.
Grants, licensing & valuation
Once IP exists, three screens track the money side: grant/incentive programs funding the R&D, tech-transfer & licensing deals monetising it, and IP valuation & royalty estimates that can sync straight back onto the patent record’s fair market value.
The two-stage hybrid engine
Nearly every screen in this add-on that does AI work follows the same two-stage pattern, and it’s worth understanding once rather than per-screen:
| Stage | What it does | Where it runs |
|---|---|---|
| Stage 1 — Embedding Engine | Vectorises the user’s query/spec text and a fixed corpus (competitor patents, prior art references, territory-specific blocking patents, papers, literature benchmarks), then ranks candidates by cosine similarity. | Always the user’s own local Ollama instance (src/services/patentEmbeddingService.ts) — there is no cloud embedding provider in this app, so this stage never proxies through the TotalApp server, which has no network path to a user’s 127.0.0.1:11434. |
| Stage 2 — AI Engine | Takes Stage 1’s top matches and reasons over them — risk classification, novelty scoring, legal claim-element analysis, or literature synthesis — producing the structured result that gets saved to the screen’s ledger. | Routes through whichever engine the user picked in Settings → Agentic (Ollama / Local LLM / Web LLM stay entirely on the user’s machine or browser; Local CLI / API route through the server’s /api/patent-engine/… endpoints). |
Why the split matters
If Stage 2 ignored the user’s Ollama/Local-LLM/Web-LLM selection and always called the server, a user running fully local would get an unrelated 500 error from the server trying to reach Anthropic/Cohere. Every AI-calling function in this add-on checks writerEngine before deciding whether to call the server at all — the “Full Writer Engine Support Rule”.
When the embedding model isn’t configured in Settings → Agentic (or Ollama isn’t reachable), Stage 1 degrades gracefully — it returns an empty candidate list and the screen shows a plain-language warning rather than crashing. Stage 2 is never called with candidates it doesn’t have.
Pages in this section
Frequently Asked Questions
patent-rd-ops package activated, and your user must hold a PATENTRD domain role. See Domain Roles & Capabilities.patent-rd-ops/ module folder on the server — one JSON file per screen (e.g. patent-portfolio.json, patent-collision-scans.json), plus roles/roles.json for role definitions.