DCT
1:25-cv-01230
Certara USA Inc v. Edvise Inc
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: Certara USA, Inc. (Delaware)
- Defendant: Edvise Inc. (Delaware)
- Plaintiff’s Counsel: Morris James LLP
- Case Identification: 1:25-cv-01230, D. Del., 10/03/2025
- Venue Allegations: Venue is asserted on the basis that Defendant is a Delaware corporation and therefore resides in the District of Delaware.
- Core Dispute: Plaintiff alleges that Defendant’s AI-assisted spreadsheet plug-in, the Endex product, infringes a patent related to methods for automatically populating spreadsheet data using machine learning modules.
- Technical Context: The technology at issue involves using natural language processing and machine learning to automate the population of spreadsheet cells based on complex, patterned questions, aiming to supersede the limitations of traditional manual data entry and simple macros.
- Key Procedural History: The complaint does not reference any prior litigation, inter partes review proceedings, or licensing history related to the patent-in-suit.
Case Timeline
| Date | Event |
|---|---|
| 2019-10-07 | ’492 Patent Priority Date |
| 2021-09-28 | ’492 Patent Issue Date |
| 2022-09-19 | Defendant authorized to transact business in Delaware |
| 2025-10-03 | Complaint Filing Date |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 11,132,492 - "Methods for Automated Filling of Columns in Spreadsheets"
- Patent Identification: U.S. Patent No. 11,132,492, “Methods for Automated Filling of Columns in Spreadsheets,” issued September 28, 2021 (’492 Patent).
The Invention Explained
- Problem Addressed: The patent’s background section notes that populating spreadsheets has traditionally relied on either "laborious, piecemeal addition of data" or macros, which are "stored functions" limited to simple formulas like computing a total for a column of values (’492 Patent, col. 1:23-34). These methods are inadequate for automating more complex data-gathering tasks.
- The Patented Solution: The invention describes a method using "smart columns" that are populated by answers to a "template question." A user provides a template question containing a variable, such as "Who is the CEO of [Company Names]?" (’492 Patent, Fig. 1). The system takes each entry from a corresponding "variable column" (e.g., a list of company names), inserts it into the template question, and uses a machine learning module to find the answer and populate it in the smart column (’492 Patent, col. 2:1-19). The system also automatically updates the smart column if an entry in the variable column is changed (’492 Patent, col. 2:20-30).
- Technical Importance: The invention enables spreadsheets to automate sophisticated data acquisition by interpreting and answering a series of related questions, rather than merely performing mathematical calculations on existing data (Compl. ¶16).
Key Claims at a Glance
- The complaint asserts independent claims 1 (method) and 7 (system) (’492 Patent, col. 11:24, col. 12:15; Compl. ¶19). The analysis focuses on claim 1, which is detailed in the complaint.
- The essential elements of independent claim 1 are:
- Receiving a template question for populating a smart column, where the question has variables corresponding to a separate variable column.
- Automatically identifying alphanumeric responses to instantiations of the template question using a machine learning module.
- Populating a cell in each row of the smart column with the corresponding response.
- Automatically updating the smart column upon editing an entry in the variable column by using the processor to identify an updated response and populate the smart column accordingly.
- The complaint also asserts dependent claims 2-4, 8, and 11 and reserves the right to assert others (Compl. ¶19).
III. The Accused Instrumentality
Product Identification
- The accused instrumentality is the "Endex product," a Microsoft Excel-based, AI-assisted plug-in (Compl. ¶6).
Functionality and Market Context
- The Endex product is alleged to provide "AI-Powered Modeling Assistance" for building financial models (Compl. ¶26). According to the complaint, a user can provide a natural language prompt with key assumptions (e.g., "Create a P&L model for my B2B SaaS startup...Team of 8...") (Compl. ¶28). The product then allegedly uses machine learning to identify additional required assumptions (e.g., average salaries for the roles specified), pulls data from public sources or uploaded documents, and generates a complete financial model within an Excel spreadsheet (Compl. ¶¶ 26, 29, 31). The complaint includes a screenshot of a user prompt within the Endex for Excel interface requesting the creation of a P&L model with specific business assumptions (Compl. p. 8).
IV. Analysis of Infringement Allegations
’492 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| receiving, by a processor... a template question for populating a smart column... said template question comprising one or more variables... each of said one or more variables corresponding to a variable column... | The Endex product receives a user’s natural language prompt (e.g., to create a financial model) containing certain assumptions (e.g., "Team of 8"). This prompt is alleged to be the "template question," and the embedded assumptions are alleged to be the "variables." | ¶¶27-28 | col. 2:1-7 |
| automatically identifying, by the processor, one or more alphanumeric responses to each of a plurality of instantiations of said template question... wherein said identifying is performed using a machine learning module; | The Endex product's machine learning module allegedly identifies additional assumptions needed for the financial model, such as average salaries for the employee roles specified in the user’s prompt. These identified salaries are the alleged "alphanumeric responses." A screenshot shows the Endex product indicating it "Need[s] detailed salary expense data" (Compl. p. 11). | ¶¶29-31 | col. 2:8-14 |
| populating, by the processor, a cell in each row of the smart column with the one or more alphanumeric responses corresponding to the instantiation of the template question for that row; | The Endex product allegedly populates spreadsheet cells with both the user-provided assumptions and the machine-learning-identified assumptions (e.g., average salaries). The complaint provides a screenshot of a spreadsheet showing rows for "Role," "Headcount," and "Average Salary ($)" populated with data (Compl. p. 10). | ¶¶32-34 | col. 2:15-19 |
| and automatically updating said smart column upon editing of an entry in the variable column by identifying, by the processor, an updated response... | The complaint alleges on "information and belief" that because the inputs are described as "adjustable" and "dynamically linked," editing a user-provided assumption (e.g., headcount) will cause the machine learning module to identify an updated response (e.g., updated salary data) and automatically update the corresponding cells. | ¶36 | col. 2:20-30 |
Identified Points of Contention
- Scope Questions: The complaint alleges that a free-form natural language prompt ("Create a P&L model for...") functions as the claimed "template question" and that parameters embedded within it serve as the "variable column" (Compl. ¶¶27-28). This raises the question of whether the claim language requires a more structured relationship between a template and its variables, such as the patent’s exemplary "Who is the CEO of [Company Names]?" format (’492 Patent, Fig. 1). Further, the complaint alleges that data organized in horizontal rows can satisfy the "variable column" limitation because "Row and column are relative terms of orientation" (Compl. ¶34). The interpretation of "column" in the context of a spreadsheet may become a central issue.
- Technical Questions: For the "automatically updating" limitation, the complaint relies on marketing language from the Defendant's website ("adjustable inputs," "dynamically linked") rather than direct evidence of the accused product's operation (Compl. ¶36). A key technical question will be what evidence demonstrates that the machine learning module is used to re-identify a response upon editing, as required by the claim, versus a conventional spreadsheet function that simply recalculates a value based on a non-ML formula.
V. Key Claim Terms for Construction
The Term: "template question"
- Context and Importance: The definition of this term is critical because it dictates the required structure of the user input that initiates the patented method. The complaint's infringement theory rests on construing a general, natural-language prompt to be a "template question." Practitioners may focus on whether the term implies a query with explicit placeholders for variables, or if it can cover a more general instruction.
- Intrinsic Evidence for a Broader Interpretation: The claims do not specify the format of the template question, stating only that it "compris[es] one or more variables" (’492 Patent, col. 11:30-31).
- Intrinsic Evidence for a Narrower Interpretation: Every example provided in the patent specification depicts the template question as a distinct query with bracketed placeholders, such as "Who is the CEO of [Company Names]?" (’492 Patent, Fig. 1; col. 4:36-37). An argument could be made that this consistent usage limits the scope of the term.
The Term: "variable column"
- Context and Importance: This term defines the source of data used to instantiate the "template question". The infringement case depends on this term being flexible enough to read on either parameters embedded in a text prompt or data organized in horizontal rows (Compl. ¶¶27, 34). The dispute may center on whether "column" has its ordinary meaning in a spreadsheet context (a vertical series of cells) or a more functional meaning (a set of related variables).
- Intrinsic Evidence for a Broader Interpretation: The patent does not provide an explicit definition of "column," which could leave room for an argument that any organized set of variables, regardless of orientation, meets the limitation.
- Intrinsic Evidence for a Narrower Interpretation: The patent consistently uses "column" in a manner consistent with its conventional spreadsheet meaning. Figure 1 and Table 1 both depict a standard vertical "Company Names" column (’492 Patent, Fig. 1; col. 4, Table 1). The claim also distinguishes the "variable column" from the "smart column," suggesting two separate, conventionally oriented columns (’492 Patent, col. 11:33-36).
VI. Other Allegations
- Indirect Infringement: The complaint alleges induced infringement, stating that Edvise encourages end users to use the Endex product in an infringing manner (Compl. ¶20). It also pleads contributory infringement, alleging the product is especially made for infringement and is not a staple article of commerce suitable for substantial non-infringing use (Compl. ¶22).
- Willful Infringement: Willfulness is alleged based on knowledge of the ’492 patent as of the filing date of the complaint. The pleading asserts that Edvise is acting "without a reasonable basis for believing that it does not infringe" (Compl. ¶21).
VII. Analyst’s Conclusion: Key Questions for the Case
- A core issue will be one of structural equivalence: Can the patent’s structured framework of a "template question" instantiated by variables from a distinct "variable column" be construed to cover the accused product's functionality, which allegedly processes a single, free-form natural language prompt containing embedded assumptions?
- A second key issue will be one of definitional scope: Does the term "column", as used in the claims, require the conventional vertical orientation of data in a spreadsheet, or can it be interpreted functionally to read on data organized in horizontal rows, as alleged in the complaint?
- A central evidentiary question will likely concern the "automatic updating" element: What evidence will show that the accused product uses its "machine learning module" to re-identify and populate an updated response when an input is edited, as the claim requires, rather than using a standard, non-ML spreadsheet recalculation function?