No. The system prompt forbids patentability, infringement, or freedom-to-operate opinions, and the bot routes those requests to the responsible attorney. The bot describes search scope, methodology, deliverable format, and SLAs; the legal conclusion is the attorney's call, on the formal record. This boundary matters as much for the firm's professional posture as for the buyer's expectations.

Yes, when that data lives in ACF fields on the service. Most firms have a deep core (USPTO, EPO, JPO) and varying depth in CNIPA, KIPO, and other offices. A boolean or rank field per jurisdiction lets the bot say honestly 'we search China at full depth' or 'China coverage is surface-only via WIPO; deep CNIPA requires a partner firm.' Honest scoping prevents the disappointed-delivery problem at handoff.

Yes, for published rush tiers. Standard turnaround at standard price, expedited at the uplift percentage, and rush at the larger uplift can all live in postmeta. The bot quotes the published structure but routes any negotiated multi-search portfolio pricing to the PM under NDA. In-house counsel get fast answers on standard rates, you keep control of negotiated deals.

Yes. Each search type's NPL inclusion (default, optional, excluded) lives in an ACF field. A buyer asking 'does the invalidity search include IEEE and ACM literature' gets a real yes or no plus the supplemental cost if applicable. For litigation prep where NPL is often the decisive evidence, this scoping question is the first thing counsel asks; the bot answering it in chat saves an email round trip.

Bot display can be scoped via display conditions to specific URLs or post types, but you cannot reliably block competitor visitors based on IP or company name from chat. What you can do is configure the system prompt to refuse questions that look like competitive intelligence gathering (e.g., 'how do you compare to firm X', 'what is your win rate against firm Y') and route those firmly back to your services. The guideline filter helps enforce this.

Yes. WPML or Polylang content is honored, and the system prompt can be tuned per language. A Japanese in-house counsel asking about JPO depth in Japanese gets a Japanese-language answer drawn from your Japanese content. For firms serving global IP teams, this preserves the relationship and signals professionalism before PM ever picks up the call.

Inside your WordPress database. Each log row stores the model, token usage, the page URL, and the full conversation. For patent search firms with confidentiality obligations to clients (especially in active litigation), in-database logs you control are categorically different from logs on a hosted chatbot service. You set retention, you grant access, and you can purge or export under your own rules.

Yes, with strict scoping. If your firm also offers expert witness services, the bot can describe technology area expertise, deposition experience, and engagement structure, but never opine on case-specific facts. The system prompt routes substantive expert-witness scoping to a named senior analyst with the appropriate conflict-check workflow, rather than letting the bot drift into territory that could conflict-out the firm.