Methodology & transparency

A process built for defensible conclusions.

An honest breakdown of what technology does in our process, where human judgment takes over, and how the complete research record is produced.

Analyst drawing and documenting a technical research process

Accountability

Named analyst at every stage

01The four stages

Efficiency where it helps. Judgment where it matters.

Every search moves through the same accountable sequence, while the actual research plan adapts to the claims, technology, jurisdictions and intended use.

Human-Built Strategy, Analyst-Run Search

A named analyst reads the invention and builds the search strategy: concepts, CPC/IPC classifications, keywords and technical equivalents. The analyst then runs and refines the search across patent databases and non-patent literature. AI tools help with routine tasks such as de-duplication, family grouping and translation; they do not decide what is searched or what is relevant. The client’s confidential disclosure is never submitted to any third-party AI tool.

Human Expert Review

A named analyst assesses every retrieved reference against the question the research must answer. Where the engagement is claim-focused — such as FTO, evidence-of-use or infringement analysis — the analyst maps the relevant claim elements against the technology or evidence under review. Hallucinated, unsupported or misapplied citations are flagged in every service.

Documented Reasoning Trail

Full audit trail: search queries run, databases accessed, relevance rationale for every included and excluded reference. Reproducible and defensible.

Named Sign-off

A credentialed analyst and senior reviewer co-sign every deliverable. The opinion is ours, not the algorithm’s. Every methodology decision is documented and reproducible.

02Inside the methodology

01What AI does in our processA named analyst reads the invention, builds the search strategy and runs the search across patent databases and literature. AI tools help only with routine tasks — removing duplicates, grouping patent families, translating foreign-language documents and formatting — and never receive the client’s confidential disclosure. AI does not choose what is searched, determine relevance or construe claims.
02What a human analyst doesA named analyst reads every potentially relevant reference, assesses relevance against the specific claims under analysis, maps claim elements to disclosed features, identifies claim-construction issues, and, in AI Output Validation work, flags hallucinated or misapplied citations in the client’s AI-generated results.
03How the audit trail is producedSearch queries, databases accessed, date and time of search runs, relevance rationale for every included reference, and reasoning for excluded references are logged in a structured search summary that accompanies every deliverable. The methodology is reproducible.
04Who signs offA senior analyst co-signs every final deliverable with the primary analyst. Both are named, credentialed analysts who stand behind the work. No deliverable leaves under an anonymous or organizational signature.

03Questions counsel ask

How the process works in practice.

Why not just use our AI search tool ourselves?

You can — and many clients do. The question is what you’re relying on the output for. AI tools are useful for broad retrieval across indexed databases, but they remain inconsistent with older scanned records, standards, conference material, shifting terminology, claim-construction judgment and hallucinated or misapplied citations. If the search supports an FTO opinion, an IPR petition or a litigation position, you need a named professional who can defend the methodology. An AI tool cannot be held accountable for its conclusions.

What about confidentiality for pre-filing invention disclosures?

Every engagement begins with a mutual NDA before any project materials are shared — no exceptions. Invention disclosures and unfiled applications are not transmitted through any third-party AI system’s servers, not stored in any shared platform, and accessible only to the assigned project team. Client materials are never used to train any AI model, internal or third-party. This is contractually committed in every engagement agreement.

Can you validate search results our AI tool already produced?

Yes. You send us the output produced by your AI tool, internal LLM, or prior art platform. Our analysts independently test the cited sources, assess whether the conclusions are supported, identify what is strong or weak, and flag material gaps or missing search coverage. You receive a written AI Output Validation report explaining how reliable and complete the output is. If you then commission 3AIP to conduct a comprehensive manual search on the same matter, the validation fee you have already paid is credited in full against the total cost of that search.

Do you use AI in your own research process?

Yes, but only for routine support tasks. A named analyst understands the research question, builds the search strategy and runs the search. AI tools may help remove duplicates, group patent families, translate foreign-language documents and format material. The client’s confidential disclosure is never entered into a third-party AI tool. Every result is reviewed by a human analyst against the purpose of the engagement; for claim-focused work such as FTO, evidence-of-use or infringement analysis, that includes claim construction and element-by-element mapping. AI does not make the relevance judgment or write the final opinion — the named analyst and reviewer take responsibility for the deliverable.

Begin with the question

Discuss the methodology for your matter.

A 20-minute scoping call, with an NDA in place before any project materials are shared.

Get in touch