FTO Search

This video provides an end-to-end walkthrough of the Eureka FTO Search Agent, showing how it takes users from technical input and search filtering through feature extraction, keyword and classification expansion, search optimization, screening, claim charting, and final report generation. It demonstrates how the agent helps automate much of the traditionally slow and expensive FTO process, making it easier for IP and legal teams to identify potential risks, review relevant patents, and share structured findings, while ensuring that final legal conclusions remain with qualified professionals.

  • Go to the Intellectual Properties section in Eureka.

    Select the Eureka FTO search agent to enter the freedom-to-operate workflow.

    Purpose: automate much of the traditionally slow, expensive, and hard-to-scale FTO process while keeping final legal judgment with qualified professionals.
  • Use the authority filter to include or exclude jurisdictions.

    Use the legal status filter to focus on:

    Active/enforced patents

    Pending patents

    Optional inactive patents for historical context

    PCT designated stage patents that may enter national phase soon

    Open More filters to:

    Include or exclude specific assignees

    Reduce self-noise by excluding your own company

    Narrow by application date if needed
  • Upload up to 10 images using the plus icon if helpful.

    Note: this is not image search.

    The agent converts images into textual descriptions using AI.

    Those descriptions are combined with your written technical solution to support the search.
  • Choose an example technical solution, such as the elevator door monitoring system.

    Click the blue run arrow to start the analysis.

    The agent begins by summarizing and extracting features from your input.
  • The agent rewrites your description into a more detailed summary.

    It may add implied details in blue.

    Review carefully and edit if needed:

    Remove incorrect implications

    Make minor wording changes

    Feature types include:

    Technical field: broad disciplinary area

    Core features: essential to the search strategy

    Essential features: important but editable

    Supporting features: helpful context

    You can hover over the info icon for definitions.

    Technical field and core features can be edited but not deleted.

    Essential and supporting features can be edited or deleted.

    Click Confirm and Run when ready.
  • For each extracted feature, the agent generates:

    Keywords

    Related terms

    IPC codes

    CPC codes

    These are grouped by feature to support the search strategy.

    You can edit the groups to:

    Add new keyword groups

    Add keywords to existing groups

    Add IPC/CPC codes

    Save changes to rerun the analysis if needed.
  • The agent merges keywords and classification codes into an initial query.

    This is an iterative, self-optimizing process.

    It measures performance using a recall score against an internal benchmark set of relevant patents.

    The agent adjusts the query over multiple iterations to improve recall.

    Target: reach about 90% recall.

    Once that threshold is reached, the agent performs a final semantic search to catch patents with different wording but similar meaning.
  • The search may return many references, so the agent narrows them to the 100 most relevant.

    It evaluates patents more deeply using:

    Text

    Abstract

    Especially the first claim

    You can switch between result tabs:

    All references

    Top 100 retained references

    You can also:

    Save references to a workspace

    Import references from a workspace

    Edit the retained/excluded set manually
  • The agent compares your invention’s extracted features against the claims of the retrieved patents.

    This is the key FTO step because claims define legal boundaries.

    It analyzes both:

    Independent claims

    Dependent claims

    For each patent, it identifies how many claims contain elements equivalent to your invention’s features.

    Use View details to see a side-by-side table of:

    Patent claim text

    Extracted claim features

    Your subject technology features

    AI equivalence determinations
  • Patents are rated high, medium, or low relevance.

    The rating depends on:

    Patent legal status

    Whether there is at least one matching independent claim

    Typical interpretation:

    High: active patent with strong claim overlap

    Medium: at least one matching independent claim, but patent is pending

    Low: little/no overlap, or patent is expired/lapsed

    High-relevance patents are added to the report by default, but you can add or remove any references manually.
  • You can remove a patent from the report by clicking Added.

    You can include medium- or low-relevance patents by toggling them to Added.

    You can also:

    Save to a workspace

    Add a missing reference manually

    Import references from a workspace

    The agent will run claim charting for any added reference before report generation.
  • Click Generate FTO report and then View report.

    The report starts with an overview table of:

    Assignees

    Patents captured

    Grouping by relevance

    The rest of the report provides a full summary of the workflow, including:

    Search scope and entered technology

    Claim charts for highly relevant patents

    AI-generated technical summary

    Extracted technical features

    Keyword list and IPC/CPC codes

    Final search strategy
  • Export the report to a Word document.

    Use the Share button to send a link to colleagues.

    This makes it easy to collaborate on the analysis inside or outside Eureka.
  • The Eureka FTO agent automates the most time-consuming parts of FTO searching:

    Feature extraction

    Keyword expansion

    Search strategy building

    Screening

    Claim charting

    Reporting

    It helps IP and legal teams move faster and work at scale.

    Final legal judgment still remains with qualified professionals.

Was this article helpful?

Have more questions? Submit a request