What Are Smart Filters?
Smart Filters help you narrow down patent datasets intuitively by asking specific questions about them, in pure natural language. They make our Multi-Patent Ask AI feature directly actionable by:
Detecting if the user asks a Yes/No question.
Answering the question based on each document, providing both reasoning and a final answer: YES / NO / MAYBE.
Allowing users to apply these answers as filters.
Smart Filters utilize a large language model (LLM), currently Anthropic's Haiku 3, in "patent professional mode." This means it processes each document individually and in parallel rather than aggregating responses.
When to Use Smart Filters
This feature is particularly useful for:
Feature-level analysis in novelty, invalidity, and state-of-the-art searches.
Patent portfolio understanding and in-depth analysis.
Office action responses, as well as defensive invalidity and opposition work.
How to Use Smart Filters
Smart Filters are available on all patent list views in IPRally: Search results, Monitoring results, Collections.
Smart filters are initiated using Multi-patent Ask AI:
1. Select all documents in your dataset.
2. Click on Ask AI button at the top of the list to open the Multi-patent Ask AI panel
3. Ask a Yes/No question relating to the contents of the patents
4. The system will analyze each document and return answers in a YES/NO/MAYBE format
5. The system will suggest filtering out the "No" documents if found
Note: Smart Filters primarily provide binary answers, i.e. YES/NO. However, IPRally may return a MAYBE response when the AI cannot determine a definitive answer, ensuring non-hallucinogenic responses aligned with Ask AI’s behavior.
To continue, simply ask another YES/NO question to continue refining the result set:
Finally, complete your line of questioning with a more specific 'binary' focused question:
You are able to view your 'Smart Filters' by clicking on the icon, and if necessary 'Re-run' to ensure complete coverage:
Example Questions for Smart Filters
Exemplary single-aspect Smart filter questions
The typical use of Smart filters consists of single-aspect questions, or a cascade of single-aspect questions. Examples:
"Does the document disclose a piezoelectric transducer suitable for pressure sensing?"
"Is there an active temperature drift compensation mechanism?"
"Does the document disclose kaolin as filler in the main layer of the sheet?"
"Does it discuss the relationship between the temperature of the lubricant and energy efficiency of the apparatus?"
"Is it a multi-stage drying process?"
"Does it address the problem of reducing vibrations at low rotation speeds?"
Single-aspect questions focus on one feature at a time, but as you can see, the aspect can be a complex technical concept, with structural, functional or purpose-related sub-aspects.
Exemplary multi-aspect Smart filter questions
Smart Filters can also handle multi-aspect questions, provided that they are formulated in a way that allows binary YES/NO answering. In that case, the AI analyzes multiple features simultaneously. Example:
“Does the patent disclose all of the following features:
A) A material suitable for use in a three-dimensional printing system;
B) A carrier ink containing pigment particles;
C) The pigment particles comprising mica; and
D) A colorant that is different from the pigment particles.
Analyze feature by feature, indicate potential gaps, and cite relevant passages.”
FAQs
1. How reliable are Smart Filter results?
Smart Filters use LLMs, which are not perfect. Always double-check the answers, as AI may misinterpret details, misunderstand the question, or miss key information.
2. Are results more reliable for certain patents?
Yes, analysis of short patents tends to be more reliable. Longer patents, such as pharmaceutical ones, have a higher risk of missing information or misreferencing paragraphs.
3. Can Smart Filters analyze image-based data?
No. Smart Filters do not analyze images. They rely on text-based content only. Tables and chemical formulas are considered only if they are in text format rather than images.
For additional support, visit IPRally Support or check out our video walkthrough of Smart Filters in action! 🎥