You can now configure ClaimMaster to run additional patent proofreading on the document using the configured LLM service/model (e.g., OpenAI, Anthropic, local LLM instances, etc.) and specific patent proofreading prompts. While proofreading with LLMs might not catch the same issues as ClaimMaster, they can still provide a second line of defense against many logical and grammatical issues in patent documents, as well as catch various typos. LLMs are also good for determining whether claims terms are structurally in the specification.
Note – this feature is currently complementary for all users with patent proofreading capabilities (i.e., from QA to Pro+Shells+Drafting subscriptions). You can use the default stateless LLM models set up by ClaimMaster with OpenAI, Azure, or Anthropic to perform proofreading or you can configure ClaimMaster to use your own private models. WhiIe LLM proofreading is complimentary, if you have QA, Pro, or Pro+Shells subscription, LLM proofreading capability is turned off by default and you’ll need to opt-in for this service on the admin level for your license to have this feature enabled. Please contact support[at]claimmaster.com to have this feature turned on. Pro+Drafting and Pro+Shells+Drafting users do not need to contact us to enable this feature, as their subscriptions already include LLM integration.
To enable LLM proofreading, follow these steps:
- Open ClaimMaster preferences
From the ClaimMaster Ribbon, click on the Preferences, Extra Tools, Help menu, then Preferences menu:
- Opt-in for LLM Proofreading (Optional for QA, Pro, and Pro+Shells subscriptions)
If you have QA, Pro, or Pro+Shells subscription, first contact us at support[at]claimmaster.com to have LLM patent proofreading enabled for your license. After this, you can turn LLM patent proofreading in ClaimMaster by clicking on the Enable LLM Proofreading button in the Patent Proofreading tab of Preferences:
- Enable LLM proofreading for all-in-one reports
If LLM proofreading is enabled for your license, you can turn it on for all-in-one reports and also configure the desired LLM service and customize LLM patent proofreading prompts. Click on “Use configured LLM service for proofreading” to actually enable/disable LLM proofreading for all-in-one reports. If enabled, the contents of your document will be sent to the configured LLM service with the specific proofreading prompt that you can configure.
- Customize LLM model (optional)
In addition, clicking to speech bubble button will let you configure the desired LLM service and settings. You can use the default stateless LLM models set up by ClaimMaster with OpenAI, Azure, or Anthropic to perform proofreading or you can configure ClaimMaster to use your own private models.
- Customize LLM prompts (Optional)
Here you also have an option to customize LLM patent proofreading prompt templates. Click on the template button to edit various patent proofreading prompts provided with ClaimMaster (you can also create new prompts, if you’d like)
The prompts provide specific instructions for the LMM input and output. They can also contain text replacement fields [CLAIMS_TEXT] and [SPECIFICATION_TEXT] that will insert claims/specification text into the prompts sent to the LLM.
- Run all-in-one reports
If LLM patent proofreading is enabled, you will see the LLM-identified Issues tab in all formats of the generated all-in-one reports, such as the one shown below.
For more information on this feature, check out the Online Manual