When sharing a search case or collection, you can now leave a comment for the receivers to see. The comment will appear both in the notification centre and the email notification if the receiver has turned them on in Settings.

Sharing a search case:

How the comment looks in the notification center:

How the comment looks in the email notification:

Please note:

Image Search decodes the technical meaning of an image to find related technical concepts: Now it's possible to add an image and generate a search that finds documents similar to those represented by the image.

Minor enhancements to improve the use of the filters in collections: Now it is easy reset the filters by clicking the amount of Documents.

Ask AI drawer now has arrow buttons, which allow you to navigate between documents without closing the drawer. This can be helpful, e.g. when reviewing long Ask AI conversations after multi-patent Ask AI.

We added pagination for the Monitoring tab. This speeds up the opening of search cases with a large number of monitoring results and makes the navigation smoother.

Sometimes, the Ask AI component takes up unnecessary space in the result list. Now it is possible to hide the component, either from the View menu or from the View settings:

When leaving a comment, users can choose if the comment is private or public. Now you can change the default preference from Settings - Comment settings.

Users can now choose another family member and open the Ask AI drawer to ask questions from a selected family member. Earlier, you could have chosen the family member from the result list. Now, you can also change the family member directly from the Ask AI drawer.

A new, improved AI model was deployed, bringing on average a 10% improvement in search accuracy in terms of X citation recall.

Earlier the highlights you added to review your search results, were available just for that search case. Now it is possible to save the User highlights as preset highlights so that you can pull them for other search results later.

It is also possible to edit the existing preset highlights in the settings:

Earlier Ask AI could not see the contents of tables within the patent documents, and thus, answering questions that had the answer inside tables was not possible.

Now Ask AI also sees those document tables that are not images, and it is possible to ask questions about them. Tables in image format are not visible for the Ask AI yet.

You can now select multiple documents, and ask the same question from all of the documents with one click.

Select the documents and click Ask AI in the Selected menu. Write your question and click Ask AI. You also see how many credits you will spend and how many credits your company has left. We have given 1000 free credits to your organisation. The answers appear in the document table. Multi-document will require credits that you can buy more by contacting us. Single-patent questions will remain free just as they have been.

You can select the most critical search cases and collections with a pin, and they will appear at the top of your search case list. By re-clicking the pin, the search case or collection will return to the original place in the list.

It is possible to use Ask AI for any patent family member. You can click the "+ x more" button and the star icon next to the patent family member (and the copy icon). After the click the Ask AI drawer opens, and you can start asking questions.

It is now possible to clear the Ask AI chat from the whole page at once. If you choose Bulk Actions and Clear Ask AI chat, the chat is removed from all the documents on the page you are viewing. Please note that it's not possible to return the questions, except by re-asking them.

We constantly try to improve the workflow for our users. Renewing our result table is one of these attempts. The new result table doesn’t introduce any new features but it’s a big change for reviewing the results. We have kept the old information but tried to make the information more visual and easier to skim through, offering the most used actions easily available.

You can still adjust the view from the View Options, and expand and collapse each document or the whole result list.

We hope you enjoy the new view!

We are trying to make your life easier by providing a search that helps you to find old search cases and collections.

A search box is added above the "My search cases" and "My collections" lists. If you start typing, the list below shows those results, that have matching characters in the search case name or collection name. For example if the user types 'lmp' it will match 'lamp' and 'limp' and 'lampost'.

We have updated our boolean scoring: The score for the top-most boolean search result isn't always 100 anymore. Instead, the score reflects of how well the top result matches the TACD fields in the Boolean query.

For example, a search result that occurs for both
​TACD:device AND PBC:(US OR JP) AND PA_ALL:toshiba and
​TACD:device AND PBC:(EP OR JP), will have the same boolean score, because TACD fields (that are used for boolean score calculation) in the query are unchanged.

As we have nowadays boolean search up and running, we might end up having quite large result sets. We wanted to simplify the interactions you can do with the result sets.

The operations you can do for the whole dataset up to 50,000 documents can be found under the Bulk actions menu. The actions you can do for the documents you have selected from the viewed page can be found under the "x Selected" menu.

Previously we limited them to about 200 words. Now we raised the limit to about 1500 words. So if you are asking questions such as "Summarize each of the claims" on a patent document that has many claims, you will now get a better answer.

Knowing the amount of results is sometimes extremely important. Now, when you make a boolean search, you can see the amount of found families in the pagination. When the search includes wildcards, the number might be an estimate. In those cases, it's indicated with "est." at the end of the number.

You can find the pagination from the bottom of the result list for boolean search results.

The automatic highlights indicate what parts of the document best correspond with the search criteria. Now this feature is available also in the "Favorites" and "Monitoring" tabs. Hopefully, this gives you better insights into the document's relevancy!

Often the most recent documents are the most relevant ones, so we have adjusted our monitoring mails also so, that the newest results are on the top.

It is now easier to navigate boolean search results with a new pagination component. You can see the amount of documents if the information is available, select how many results you would like to see per page and move between pages.

For reviewing the results, there are more sorting options available in the new sorting menu. The biggest improvement is for the Classifier users, who can sort the documents by predicted tag confidence. You can find the new menu in the top right corner of your document lists.

It is possible to include special characters in the boolean search and boolean filters. Supported special characters are following: § ! # € $ % & / ° ´ ` - [ ] \ ^ _ { } | ~ @ , .

Ask AI improvements:

In addition, many small design and usability improvements all around the app.

Users can ask questions about a patent and get them answered by an AI. The chat is optimised for giving answers from specific, selected patent documents, and you can ask any questions you like. The feature uses Anthropic's Claude model.

Update on October 18: The conversations are now saved in the app.

A new, larger AI model was deployed, bringing on average a 15% improvement in search accuracy in terms of X citation recall.

In Boolean search & filtering, added support for wildcards inside phrases.

UI and UX improvements and bug fixes.

Users can now do pure boolean searches. The release is still in Beta mode but it's free for the users to test. There are some limitations in the release: Monitoring is not supported, export and add to the collections work only for the paginated documents, the pagination does not always indicate that there might be more pages, searching may take too long or timeout when there are wildcards in the query, wildcard inside a phrase is not supported and they need at least three consecutive characters on either side of the wildcard.

Users can narrow the search results by giving improved boolean operators as filters.

A major update to our core neural network architecture, which improves your patent search experience. With the update IPRally can better understand the context of the patent text, resulting in more relevant results being retrieved. You can read more here.

A new AI model brought a significant step change (up to 27%) in search accuracy in terms of X citation recall.

Various fixes and improvements in the Classifier use workflow

Up to 300-pixel thumbnails of images are now available in the result list.

In addition to XLSX and PDF Exports, the search results are now available in DOCX format.

You can now export Excel (XLSX) files without images up to 10 000 documents at once into a single file. XLSX export with images and PDF export are still capped at 500 documents for performance reasons.

In addition to Full specification and just the Independent claims, you have now more options to choose as the starting point for the search: All claims or a dependent claim of your choice. This makes claim-specific searches faster and easier.

You can now create classifiers just for your own use, without sharing them with with your colleagues.

The length of a visual tag has been capped, but you will see the whole tag name by hovering the mouse over it.

You can now import tens of thousands of documents to a collection at once, from a single spreadsheet.

The "Classifier box" has now a cleaner and more structured layout.

You can now switch automatic re-training on for any classifier. It means that every time your reject/confirm a tag predicted by that classifier, the corresponding document is added as part of the training corpus of the classifier.

When doing a publication number search, you can now choose between Invalidity search (applies a relevant date filter to results) or State of the art search (no date filter applied).

More drawings are now shown in the result list for quick review and available for further inspection in the smart image viewer.

The Custom AI Classifier module keeps growing: Accuracy metrics, quick sensitivity adjustment, automatic re-training on validating the predictions etc.

With the custom AI Classifier, your can categorize documents automatically according to your own taxonomy. The classifiers are trained with labeled (historical) data and can be applied to any fresh data (search results, monitoring results, portfolios, etc). Classifiers work together with Technology tags.

Quick dropdown menus make navigation inside IPRally easier.

Highlights are now more informative and interactive.

Several usability improvements, bugfixes and product infrastructure improvements.

The tags allow you to label patent families with your own technology labels.

In addition to the technical part names inside the images, now you have also a full list of the parts allowing for quickly scanning where a certain part is present.

The technical part names are now available in the images. In addition, there is automatic rotation of tilted images.

Export search cases and collections as handy PDF with summary of the result list and one-pager for each hit.

In-app and email notifications for when search cases and collections are shared with you, comments and documents are added, and other changes. You can change your Notification preferences in the Settings (Personal -> Notification settings)

Information about the current assignee, when available, was added to the bibliographic data of the hit and to the Excel export.

Improved database coverage with 20 new jurisdictions including Sweden, Brazil, Poland, Israel, Ireland, New Zealand.

Now you can share any search case or collection with other registered users of your company.

You can attach private and shared comments to publications. Comments are family-based: you'll see comments added to any family member at once for every other family member.

You can open documents as well as a whole search list in Orbit.

A scrollbar indicator shows where the AI and keyword highlights appear in the document or the result list.

The new Excel export includes images, abstracts, multiple sheets for search results, favorites, and better formatting.

A switch in the document browser allows you to open the images in a separate window. Switching between the documents and loading the images is quick.

When you open a hit in IPRally, you'll see:

New graph and AI model generation

See the first claim of a hit instead of the abstract in the expanded result list.

In addition to the Legal status, you can now also see the full Legal event history of a patent document.

We also added numerous possibilities to personalize IPRally to best fit your workflow and preferences. The new Settings panel can be accessed from the user profile menu at the top right corner of the screen.

With the personal settings you can:

Further, Company admin users have access to company-wide settings, where they can manage and invite the users, see the subscription status and and set company wide preferences, like the external links that the users see.​

Now you can start your search with a text describing the invention in languages different than English. The current languages supported are German, French, Spanish, Chinese, Japanese, Dutch, Danish, Finnish and Swedish. More languages to come soon!

With Collections you can:

This is the first step towards other smart features inside IPRally that allow you to add comments and other meta data to your findings, collaborate with your team, launch smart searches and more.

The current legal status of a patent and the family are now shown using a traffic-light-like system, allowing for quick scanning of the result list and with detailed information in the bibliographical data. A legal status filter helps focus the search on the requested type of data.

Improved functionality of the Keyword filter, corresponding to the NOT operator in boolean search.

The Statistics tab offers an overview about the result's CPC classes, Top assignees and inventors.

We improved the prioritization logic to show US / EP / WO / GB / AU family members in the result list instead of the translation!

We introduced an additional way of directing the AI: to search through the content of claims only, instead of the full specification. The claims search will provide more relevant results for the technology defined in the query graph from risk management or FTO perspective.

The examiner citations in invalidity search cases are shown in a separate tab, instead of at the end of the result list. If a citation is also found relevant by the AI, it appears among the search results and is indicated with the “Cited” label.

We launched an additional layer of security. All pieces of sensitive data, including full text input queries and graphs, are encrypted inside the database with encryption keys specific to your company.

We also imported the most comprehensive patent image database into IPRally and added image mosaics to the result list allowing for quick scanning of the results.

A new filter lets you limit the results to hits containing a specific keyword.

Now you can choose the maximum number of search results to be between 50 and 500.

Also, we added Video tutorials to the Help center.

We added a Keyword highlighting feature to help you scan results more quickly and find what you need.

CPC classes are the newest filter of the results and you'll see them also in the document view, provided a hit has them.

The Excel Export includes abstracts and all dates (priority, application & publication).

Filter the results by inventor's name.

Negative filters

The possibility to exclude fields in the different search filters (NOT filters) for IPC classes, Applicant, Inventor and jurisdiction.

Filtering based on priority date, application date, and publication date is now available.

On the features side, iterative AI searching is possible since a while, but now you can assist the AI even more efficiently by telling it the publications you already know as being relevant. This helps the AI to get on the right track and to improve the search results very quickly!

Now you can set up e-mail reports for your monitored search casesto go straight to your e-mail on weekly, monthly or quarterly basis! You can also now download these new results as a separate excel file.

We worked on the reported loading time issues and now loading the front page images is faster in the result list!

A new AI model was launched, increasing the search and AI highlighting accuracy once again.

New generation machine translations are now available in the advanced document browser, which will soon replace the Google patents link.

Last but not least, we added tooltips over various functionalities to make it easier for you to use the full capacity of IPRally! Just hover with the mouse of them to see what they do.

The drawings can now be found in the document view, behind the three stripes button. First, you'll see all the figures as thumbnails. Clicking enlarges a full figure viewer next to the specification.

The full document view is under construction and will soon replace the Google patents links as the primary way of opening the documents.

Front page figures are now available in the expanded result list view, just by the abstracts.

Downloading the original PDFs is also possible in the ACTIONS sections of the expanded view.

The Favorites heart symbol was moved to the left side of the result list, below the rhombus icon, which keeps track of tracks if the document has already been viewed.

Viewed documents handling: A blue rhombus symbol in front of the hit turns to white when you open a document for inspection. By clicking the symbol you can easily change the state of the document. You can also easily hide documents already viewed.

Graph panel hiding: To get more space for the result list, you can now hide the graph at the left hand side of the screen.

No-trace mode: For some companies, it is important not to leave any trace of which documents have popped up or viewed in searches. Admin users can find the setting "Allow saving search results" on the Company Admin page. Saving is allowed by default.

Espacenet links: You can find these in the expanded result view.
​
​Faster AI highlighting: The document view with relevant portions highlighted opens now quickly.

Global country coverage: The search index was extended with patent publications from almost 20 new countries. These include Chinese, Japanese, Korean, Australian, British, Canadian and Dutch documents, to mention some. About 75 million publications in total became searchable. You can see a full list of countries here.

Country filter: Now you can limit your search to only some of the countries using the filters. By default, all data sources are used.

Search accuracy: A new AI model was deployed, significantly improving overall search accuracy.

European language coverage: The search index was extended with native German, French and Finnish patent publications through machine translations. More information here.

Search accuracy: A new AI model was deployed, significantly improving overall search accuracy.

User interface: Fonts were modernized and other usability improvements were made.

Search accuracy: A new AI model was deployed, significantly improving overall search accuracy.
​
​Class coverage: Chemical patent classes were added to the search index. All patent classes are now covered.
​
​One-click monitoring: Any search case can be made a continuously monitored case. New documents published and matching the search profile will be collected to the Monitoring tab.

Development of the first version of the search engine.
​
​