The Search & Trends page is the primary interface for discovering, filtering, and monitoring patent filings. This guide covers all inputs, outputs, and workflows to help you make the most of the search and alert features.
Search & Trends combines three core capabilities:
The search interface provides multiple input fields to refine your patent search. Inputs no longer auto-submit; make your edits and click Apply to run the search so the trend graph and table stay perfectly in sync.
Enter a natural language description of the technology or concept you're interested in. Patent Scout uses AI embeddings to search for semantically similar patents and publications, which incorporates context, meaning, and other auxiliary information in order to return results that that are more robust, comprehensive, and relevant than traditional keyword searches.
Example: "autonomous vehicle perception systems using LIDAR and camera fusion"
Enter specific words or phrases that must appear in the patent title, abstract, or claims. This is a traditional keyword-based search using PostgreSQL full-text search.
Example: "neural network", "LIDAR", "convolutional"
Filter results to patents and publications assigned to a specific company or entity. Partial matching is supported.
Example: "Google", "Microsoft", "Tesla"
Filter by CPC code to narrow results to specific technology areas. Supports hierarchical matching.
Example: "G06N" (Computing arrangements based on specific computational models), "G06F17/00" (Digital computing or data processing equipment or methods, specially adapted for specific functions)
Filter patents and publications by grant/publication date. Both fields are optional. The default range spans the entire patent and publication data set (2023–present).
Example: From 01-31-2024, To 12-31-2024
Runs the search and trend fetch with the current filter state. Searches are only executed when you click Apply, ensuring both the Trend graph and Results table update together.
Clears all search inputs (semantic query, keywords, assignee, CPC, date range) and prepares for a fresh query. After resetting, click Apply to fetch results with the cleared filters.
Saves the current search configuration (all filters and semantic query) as a named alert. Email notifications are automatically sent when new patents or publications matching these search criteria are added.
Search results are displayed in a paginated list (20 results per page) with detailed metadata for each patent.
Use the Prev and Next buttons to navigate through pages. The current page and total page count are displayed between the buttons. Changing any filter resets pagination to page 1.
When results are available, two export buttons appear:
Both exports respect the current filter state, so you can refine your search before exporting.
The Trend chart visualizes patent filing patterns based on your current search filters. It updates automatically when filters change and respects all search inputs (keywords, semantic query, assignee, CPC, date range).
Displays patent/publication count over time, grouped by month. The chart is a line graph with time on the x-axis and count on the y-axis. Indicative of filing spikes, focus patterns, or growth trends.
Groups patents and publications by their CPC section and class (e.g., 'G06N', 'H04L'). Displays the top 10 CPC codes by count, with an 'Other' category aggregating the rest. Rendered as a horizontal bar chart. Indicative of the technological areas dominating the search results. (Note: patents and publications generally have multiple CPC codes, so counts may exceed total results.)
Groups patents and publications by assignee name. Displays the top 15 assignees by count as a horizontal bar chart. Indicative of most relevant entities in a technology space and their movements in different directions (e.g., competitive landscape analysis).
The trend chart is interactive and updates in real time as you refine your search. Key points:
Alerts enable specific search criteria to be monitored over time. USPTO publishes patent and publication data on a weekly basis. Search criteria is checked against the newly published data, and new matches to your criteria are sent via email. This section explains how to create, manage, and use alerts.
Use the search interface to define the query and/or filters of interest. This can include semantic queries, keywords, assignee filters, CPC codes, and/or date ranges. Run the search to verify it returns relevant results.
With the search parameters of interest still configured, click the 'Save Alert' button.
An input field will be displayed, prompting for an alert label. Alert labels must be unique. Descriptive names are recommended, but search parameters will be indicated in the alert details so comprehensive labeling is not necessary.
Click OK in the prompt. The alert is saved to the database with is_active=true, meaning it will be checked during the next weekly alerts run. A success message appears briefly confirming the save.
Click the "Alerts" button in the top navigation bar to open the alerts modal. This modal displays all your saved alerts with the following information:
Click the toggle switch to activate or deactivate an alert. Inactive alerts are not checked during the alerts run. This is useful if you want to pause monitoring without deleting the alert.
Click the 'Delete' button (usually styled as a red/danger button) to permanently remove the alert from the database. This action cannot be undone.
Some alert implementations may include a 'View Results' button that reconstructs the original search on the Search & Trends page. Accordingly, an aggregated and historical view of the alert's activity can be reviewed.
The alerts system runs via an automated backend process, which is scheduled to run on Friday at 07:00 AM PT (14:00 UTC), allowing a buffer period for the USPTO to make bulk data available. During each run, the alerts system performs the following steps for each active alert:
This ensures email alerts are only triggered once when search criteria is met, and that the emails provide timely notifications of new matches on the alert search criteria.
Example walkthrough of a Search & Trends session:
Log in via Auth0 using the login button in the navigation bar. The page displays a login overlay if unauthenticated.
Fill in one or more of the input fields (semantic query, keywords, assignee, CPC, date range). Inputs no longer auto-run searches; changes take effect only after you click Apply.
The Results section populates with up to 20 patents per page. Patent and publication number are clickable links to view the full document on Google Patents. Use Prev/Next to paginate through results.
After clicking Apply, the Trend chart reflects your current filters. Use the 'Group by' dropdown to switch between Month, CPC, and Assignee views. Observe filing patterns, spikes, or dominant players.
To save search results for later research and review, options to export the data set are provided for CSV and PDF. Preferred format is selected by clicking the corresponding button: 'Export CSV' or 'Export PDF'. Results exporting supports up to 1,000 results matching the current semantic search, keyword(s), and other filters.
If this is a search you want to monitor over time, click 'Save Alert', name the alert, and confirm. The automated backend process will then check the whether the search criteria match any new filings during the next scheduled alerts run, not an email notification will be triggered if necessary.
Click 'Reset' to clear all filters (no search runs until you hit Apply again), or refine your inputs and click 'Apply' to re-run the search.
AI semantic search and keyword search can be used together, e.g., to refine results, filter is concentrated within a specific context, or ensure that a search for keywords of interest is limited to a domain of interest. Thus, fine-grained control can be achieved by using both a semantic query (to capture conceptually similar patents and publications) and keywords (to ensure specific terms are present). This hybrid approach prevents the signal from being lost in the noise by leveraging the strengths of both methods.
To broadly explore technology areas, a CPC code (e.g., 'G06N' for Computing and Calculating; Counting) can be specified as a search criteria. The results can be further refined, e.g., by reference to the trend chart. CPC codes are hierarchical, so broader codes capture more patents and publications, while specific codes narrow the focus. (Note: patents and publications generally have multiple CPC codes, so counts may exceed total results.)
Create alerts for specific assignees (e.g., 'Google', 'Microsoft') to stay informed about their latest patents and applications. Combine with CPC and/or keyword filters to focus on technology areas of interest.
Trend Chart can provide a quick and convenient visual reference in a variety of contexts. For example, momentums, new products and services, shifts in technology focus, increases/decreases in R&D investment and allocation, etc. can be easily identified by tuning semantic and keyword searching and other filters to values consistent with a trend of interest. Accordingly, Trend Chart can provide a visual verification before exporting a results set.
Alerts provide a mechanism for continuous passive monitoring, avoiding the need for repetitive searches after each USPTO patent grant and application publication date.
Solution: Try broadening the search by removing filters or using a more general semantic query. Verify that the specified date range does not fall outside the searchable database (2023–present). Check that CPC codes are valid and assignee names are spelled correctly.
Solution: Usually caused by filters that are too narrow for any matches. If results are returned with no semantic search, keywords, or other filters applied, try resetting and incrementally narrowing the search focus.
Solution: Ensure you are authenticated (logged in via Auth0) and you have an active subscription. Check that at least one filter is active (an empty alert with no criteria cannot be saved). Verify the alert name is not empty.
Solution: Verify that the current search returns results before exporting. Check your browser's download settings to ensure the file is not being blocked. If the PDF fails, try CSV export as a fallback.
Solution: Very broad semantic queries or large date ranges can be slow. Try narrowing the date range or adding additional filters (assignee, CPC) to reduce the search space. Contact support if timeouts persist.