ArcPrime's AI features work best when you give them clear inputs and review their outputs with the right expectations. A few habits improve what you get back from prior art search, office action recommendations, portfolio search, and the AI assistant. This guide is for anyone on your team who uses ArcPrime's AI-powered capabilities.

When asking the AI assistant a question, include the patent number, claim number, technology area, or competitor name that is relevant. A specific question produces better results than a general one. Compare "What prior art exists for claim 3 of patent family X related to sensor fusion?" with "Find prior art for this patent." Open the assistant from the page you are asking about, so the answer is scoped to that patent, disclosure, or project.

The same principle applies to every AI feature. Before a prior art search, make sure the disclosure has a complete and specific description. Before an office action response, confirm the full office action document is on the patent record.

The Search... box in the navigation bar understands long, conceptual descriptions, not only numbers and titles. Describe the invention the way you would explain it to a colleague, and ArcPrime ranks families by meaning even when the patent uses different words. Use the Search only filter to limit matching to Claims, Specification, or Abstract when you need precision. See using global search.

Every AI output in ArcPrime is a recommendation, not a final answer. Prior art results map references to claim elements with a risk level; verify the top references against the source. Office action strategies include reasoning; read it before selecting an approach. The AI assistant lists its sources under each answer; open them to check the underlying record.

Build a review step into your workflow rather than treating AI outputs as ready to use. This takes minutes and catches the occasional result that needs adjustment.

Different AI features have different strengths. Prior art search is thorough and citation-heavy, so expect a comprehensive list that needs filtering. Office action response generation produces structured strategies and a full draft, so expect to steer and refine rather than copy. The AI assistant excels at specific, scoped questions; broad questions produce broad answers.

Spend a few sessions with each feature to understand its output style. This calibration helps you use each tool where it adds the most value.

Prior art results, Filing Recommendations, Keep Scores, and strategy recommendations are useful inputs for team discussions and committee reviews. Share them with colleagues to ground conversations in data rather than opinion. The AI provides a common baseline that speeds up decision-making.

If an AI output is off-target, check whether the input was complete and specific. Missing claim details, incomplete descriptions, or outdated documents are the most common causes of weaker results. Update the source data and rerun the analysis. If results still miss your expectations, contact support so the team can look into it.