RocketDocs 2.46 Release Notes
Content
- Added wild card search capability to make it easier to find content that you are looking for without having to consider suffixes.
General
- When logging information on the ‘Attributes & Tags’ page of content records or projects, free form text fields will now adjust to the size of the text within them, for up to five lines. After 5 lines, a scroll bar will be added to enable the user to scroll through the text box.
Administration
- Ability to Choose System Fields & Custom Attributes to Include in Bulk Updates File – Content Administrators will now be able to choose which system fields and custom attributes will appear in the attributes table in the bulk updates file.
- Ability to Restore Projects Deleted within the Last 30 Days – Project Administrators will now have the ability to restore projects that have been deleted within the past 30 days.
- Ability to Sort and Filter on the User’s Table –Administrators will now have the ability to sort and filter in the Users Table.
Parsing
- Improved Question Recognition – We have added additional terms to our parsing function to be able to identify questions as precisely as possible.
Generative AI
- Ability to Upload Folders to be Used by Generative AI – Content Administrators will now be able to upload folders of documents (policies, proposals, etc.) that users can select for use by our Private Generative AI to generate responses.
Newly Released
Our Approach to Generative AI for Private Data
For customers wanting to make the most of their private data, at RocketDocs our next generation Generative AI will offer a secure in-house solution. It doesn’t make calls outside our system to third parties, so your data never leaves our environment. This means RocketDocs can answer more questions for you without the unnecessary additional risk of sharing your data with third party systems, like ChatGPT.
- It is private, using only your data
- Generates answers by stitching together a response based on facts contained in your content library
- Nothing leaves our servers
- Leaves fewer blanks when prepopulating responses
- Generated responses are highlighted for human review and once approved are by available for future use
- Unlike third party general models, ours will enable uncensored, fine-tuned models
Two Layers of AI are Better than One.
Two layers of AI. One generative, one non. This gives you the ability to have appropriate safeguards to minimize the chances of “hallucinations” accidentally making it into materials. The non-generative AI will always be run first, this AI will always pick directly from a preapproved subset of content library responses. With information from your library always being pulled first, our generative AI will not have the chance to create a backlog of duplicate responses, which is a problem many companies that use Generative AI will face.
Once the non-generative AI work is finished, the generative AI can be invoked to help to respond to remaining questions. The answers filled in by the generative AI are purposely treated as drafts and highlighted for the writer, until they have been reviewed and approved by a human. Once approved, the newly created responses become part of the RocketDocs libraires, to be available for use by the non-generative AI. This helps to ensure that reviewed and approved content always receives priority, before the generative AI is used.
By integrating these AI tools and features into RocketDocs, you can streamline the response management process, improve the quality of responses, and increase your organization’s overall success, accuracy and efficiency.