DCT
3:24-cv-04388
Dialect LLC v. Google LLC
Key Events
Amended Complaint
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: Dialect, LLC (Texas)
- Defendant: Google LLC (Delaware)
- Plaintiff’s Counsel: BLUE PEAK LAW GROUP Group
- Case Identification: 3:24-cv-04388, N.D. Cal., 11/13/2025
- Venue Allegations: Plaintiff alleges venue is proper as Defendant is domiciled and headquartered in the Northern District of California.
- Core Dispute: Plaintiff alleges that Defendant’s Google Assistant and Google Gemini platforms, and the products that use them, infringe a portfolio of eleven patents related to conversational voice recognition and natural language understanding (NLU) technology.
- Technical Context: The technology at issue involves systems and methods that enable computers to understand and respond to natural, conversational human speech, a foundational technology for the modern market of smart devices and digital assistants.
- Key Procedural History: The complaint alleges that VoiceBox Technologies, the original inventor of the Asserted Patents, engaged in extensive pre-suit discussions with Google between 2007 and 2012. These discussions allegedly included multiple meetings with Google executives and patent counsel, during which VoiceBox specifically disclosed the patents-in-suit and related patent applications, years before Google launched its accused Google Assistant platform in 2016. The complaint also notes that Google cited several of the Asserted Patents as prior art during the prosecution of its own patent applications.
Case Timeline
| Date | Event |
|---|---|
| 2001-01-01 | VoiceBox founded to develop NLU technology. |
| 2002-06-03 | Priority Date for ’209, ’738, ’549, ’160, ’720, ’006, ’468, and ’224 Patents. |
| 2007-01-01 | VoiceBox approached Google for collaboration on an NLU product. |
| 2007-01-01 | VoiceBox met with Mike Cohen, Google's Head of Speech. |
| 2008-01-01 | VoiceBox met again with Google's speech and business development teams. |
| 2008-07-08 | U.S. Patent No. 7,398,209 Issued. |
| 2009-03-10 | U.S. Patent No. 7,502,738 Issued. |
| 2009-10-01 | Priority Date for ’607 and ’652 Patents. |
| 2009-11-17 | U.S. Patent No. 7,620,549 Issued. |
| 2009-12-29 | U.S. Patent No. 7,640,160 Issued. |
| 2010-01-01 | VoiceBox began corresponding with Steve Woods, Google’s Senior Director of Engineering, and disclosed the ’209, ’738, and ’160 Patents. |
| 2010-04-06 | U.S. Patent No. 7,693,720 Issued. |
| 2011-02-08 | Priority Date for ’659 Patent. |
| 2011-09-06 | U.S. Patent No. 8,015,006 Issued. |
| 2012-01-01 | In mid-2012, VoiceBox provided a list identifying the ’209, ’738, ’549, ’160, ’720, ’006, and ’468 patents to Google's patent attorney. |
| 2012-06-05 | U.S. Patent No. 8,195,468 Issued. |
| 2012-12-11 | U.S. Patent No. 8,332,224 Issued. |
| 2013-05-21 | U.S. Patent No. 8,447,607 Issued. |
| 2013-12-31 | U.S. Patent No. 8,620,659 Issued. |
| 2014-09-30 | U.S. Patent No. 8,849,652 Issued. |
| 2016-01-01 | Google allegedly launched the accused products. |
| 2023-04-03 | Plaintiff filed its initial Complaint in this action. |
| 2025-03-18 | Plaintiff provided notice to Google regarding Gemini's infringement of the ’549, ’468, ’224, and ’659 Patents. |
| 2025-11-13 | Plaintiff’s Third Amended Complaint filed. |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 7,398,209 - “Systems And Methods For Responding To Natural Language Speech Utterance,” issued July 8, 2008
The Invention Explained
- Problem Addressed: The patent’s background section describes the “fundamental incompatibility” between natural human speech, which relies heavily on context, and machine-based queries, which are “highly structured and are not inherently natural to the human user” (’209 Patent, col. 1:27-35).
- The Patented Solution: The invention proposes a system architecture that uses “domain agents” to manage information and behavior specific to a particular topic (e.g., weather, music) (’209 Patent, col. 2:48-59). When a user speaks, the system parses the utterance to determine its meaning and context, selects the appropriate domain agent, formulates a structured request for that agent using a corresponding grammar, and invokes the agent to process the request and generate a response (’209 Patent, col. 4:46-54; Compl. ¶21). This process is illustrated in the flowchart of Figure 6 (’209 Patent, Fig. 6).
- Technical Importance: This approach provided a framework for creating more flexible and powerful conversational interfaces that could understand user intent across various topics, moving beyond the rigid, menu-driven "Command and Control" systems of the prior art (Compl. ¶11).
Key Claims at a Glance
- The complaint asserts independent Claim 1 (’209 Patent, col. 22:46-24:8; Compl. ¶22).
- Claim 1 recites a method comprising the following essential elements:
- Receiving a user-generated natural language speech utterance containing at least one request;
- Maintaining a dynamic set of prior probabilities or fuzzy possibilities usable at each stage of processing;
- Recognizing words and phrases in the utterance using information in one or more dictionary and phrase tables;
- Parsing the recognized words and phrases to determine a meaning and a context for the request based on keywords;
- Selecting at least one domain agent based on the determined meaning, where the domain agent is an autonomous executable that processes requests associated with the determined context;
- Formulating the request in accordance with a grammar used by the selected domain agent;
- Invoking the selected domain agent to process the formulated request; and
- Presenting the results generated by the invoked domain agent.
- The complaint does not explicitly reserve the right to assert dependent claims.
U.S. Patent No. 7,502,738 - “Systems And Methods For Responding To Natural Language Speech Utterance,” issued March 10, 2009
The Invention Explained
- Problem Addressed: Similar to the ’209 Patent, the technology addresses the difficulty of enabling machines to communicate with humans in a natural, context-aware manner, which is described as a “difficult technical problem” (’738 Patent, col. 1:26-37; Compl. ¶27).
- The Patented Solution: The patent describes a system architecture composed of several key modules: a plurality of autonomous “domain agents” for handling specific contexts, a “parser” to determine context and select the appropriate agent, and an “event manager” to coordinate interactions (’738 Patent, col. 2:47-56; Compl. ¶28). A distinguishing feature is the inclusion of an “update manager” that “enables the user to purchase one or more domain agents from a third party on a one-time or subscription basis,” allowing for the system’s functionality to be expanded (’738 Patent, col. 2:63-67; ’738 Patent, Claim 1). This architecture is depicted in Figure 2 of the patent (’738 Patent, Fig. 2).
- Technical Importance: The invention outlines an extensible framework for conversational AI, presaging the "app store" or "skills" model where third-party developers can add new capabilities to a central voice assistant platform (Compl. ¶28, ¶30).
Key Claims at a Glance
- The complaint asserts independent Claim 1 (’738 Patent, col. 25:5-24; Compl. ¶28).
- Claim 1 recites a system comprising the following essential elements:
- An agent architecture that includes a plurality of domain agents, each being an autonomous executable configured for a respective context;
- A parser configured to determine a context and meaning from an utterance and select at least one of the domain agents based on that meaning;
- An event manager configured to coordinate interaction between the parser and the agent architecture; and
- An update manager that enables a user to purchase one or more domain agents from a third party on a one-time or subscription basis.
- The complaint does not explicitly reserve the right to assert dependent claims.
U.S. Patent No. 7,620,549 - “System And Method Of Supporting Adaptive Misrecognition In Conversational Speech,” issued November 17, 2009
- Patent Identification: U.S. Patent No. 7,620,549, “System And Method Of Supporting Adaptive Misrecognition In Conversational Speech,” issued November 17, 2009 (Compl. ¶31).
- Technology Synopsis: The patent describes methods for processing natural language utterances that may be incomplete, ambiguous, or subjective (Compl. ¶33). The invention focuses on a multimodal device that receives a natural language utterance and subsequently receives a "follow-up multimodal input," then determines that the initial interpretation was incorrect if the follow-up input is presented "proximate in time" to the original utterance (Compl. ¶34).
- Asserted Claims: Claim 4 is asserted (Compl. ¶34).
- Accused Features: The complaint alleges that Google's Gemini platform, which operates on multimodal devices and can be interrupted by follow-up voice or touch inputs that cause it to cancel or correct a prior action, infringes this patent (Compl. ¶174, ¶179-181, ¶193-195).
U.S. Patent No. 7,640,160 - “Systems And Methods For Responding To Natural Language Speech Utterance,” issued December 29, 2009
- Patent Identification: U.S. Patent No. 7,640,160, “Systems And Methods For Responding To Natural Language Speech Utterance,” issued December 29, 2009 (Compl. ¶35).
- Technology Synopsis: The patent describes a system for processing multi-modal inputs that include both a natural language utterance and a non-speech input (Compl. ¶37-38). The system merges transcriptions of both inputs, uses a "knowledge-enhanced speech recognition engine" to determine the most likely context by comparing the merged text against grammar entries and a "context stack," and then identifies a domain agent to process a request in that context (Compl. ¶38). The system is also configured to handle follow-up multi-modal inputs (Compl. ¶40).
- Asserted Claims: Claims 1, 2, and 3 are asserted (Compl. ¶38-40, ¶219).
- Accused Features: The complaint alleges that Google's Assistant and Gemini platforms, which process combined speech and non-speech (e.g., touch) inputs and handle conversational follow-ups, infringe this patent (Compl. ¶219, ¶224-228, ¶259).
U.S. Patent No. 7,693,720 - “Mobile Systems And Methods For Responding To Natural Language Speech Utterance,” issued April 6, 2010
- Patent Identification: U.S. Patent No. 7,693,720, “Mobile Systems And Methods For Responding To Natural Language Speech Utterance,” issued April 6, 2010 (Compl. ¶41).
- Technology Synopsis: The patent addresses the challenge of building natural-language interfaces for noisy vehicular environments (Compl. ¶43-44). The claimed mobile system features a speech recognition engine that uses dictionary and phrase entries "dynamically updated based on at least a history of a current dialog and prior dialogs," a parser that selects a domain agent based on context, and an agent architecture that couples various agents together (Compl. ¶46).
- Asserted Claims: Claim 1 is asserted (Compl. ¶46, ¶280).
- Accused Features: The complaint alleges that the Google Assistant Platform operating in automotive environments (e.g., Android Auto, Android Automotive OS) infringes this patent by using conversational history to interpret user requests (Compl. ¶279, ¶287).
U.S. Patent No. 8,015,006 - “Systems And Methods For Processing Natural Language Speech Utterances With Context-Specific Domain Agents,” issued September 6, 2011
- Patent Identification: U.S. Patent No. 8,015,006, “Systems And Methods For Processing Natural Language Speech Utterances With Context-Specific Domain Agents,” issued September 6, 2011 (Compl. ¶50).
- Technology Synopsis: The patent describes a method of processing speech that involves parsing an utterance to determine its meaning and context, and then formulating a request for a domain agent using a specific grammar (Compl. ¶55). The formulation step includes extracting and inferring criteria and parameters and transforming them into "tokens" compatible with the agent's grammar (Compl. ¶55).
- Asserted Claims: Claim 5 is asserted (Compl. ¶55, ¶312).
- Accused Features: The complaint alleges that Google Assistant infringes by parsing user speech to determine context, extracting parameters (e.g., "SFO" for a location), inferring context from prior interactions, and formulating a request for a domain agent (e.g., an app) (Compl. ¶313, ¶319-322, ¶325-328).
U.S. Patent No. 8,195,468 - “Mobile Systems And Methods Of Supporting Natural Language Human-Machine Interactions,” issued June 5, 2012
- Patent Identification: U.S. Patent No. 8,195,468, “Mobile Systems And Methods Of Supporting Natural Language Human-Machine Interactions,” issued June 5, 2012 (Compl. ¶59).
- Technology Synopsis: The patent claims a mobile device that processes multi-modal inputs by using a "semantic knowledge-based model" to create a speech transcription (Compl. ¶62). This model includes a "personalized cognitive model" from the user's prior interactions, a "general cognitive model" from a plurality of users' interactions, and an "environmental model" derived from the user's environment (Compl. ¶62).
- Asserted Claims: Claim 1 is asserted (Compl. ¶62, ¶352).
- Accused Features: The complaint alleges infringement by Google's Gemini platform, which runs on mobile devices and allegedly uses personalized data from user interactions, generalized data from multiple users, and environmental data (e.g., location) to process requests (Compl. ¶352, ¶354, ¶357-365).
U.S. Patent No. 8,332,224 - “System And Method Of Supporting Adaptive Misrecognition Conversational Speech,” issued December 11, 2012
- Patent Identification: U.S. Patent No. 8,332,224, “System And Method Of Supporting Adaptive Misrecognition Conversational Speech,” issued December 11, 2012 (Compl. ¶66).
- Technology Synopsis: The patent describes a method for adaptively improving speech recognition by monitoring user actions after a request is processed (Compl. ¶69). The system determines if an interpretation was incorrect, tracks the user's interaction pattern over time to generate a "personalized cognitive model," uses that model to predict future actions, and updates the model based on the frequency of incorrect interpretations (Compl. ¶69).
- Asserted Claims: Claim 26 is asserted (Compl. ¶69, ¶394).
- Accused Features: The complaint alleges infringement by Google's Gemini platform, which allegedly tracks user interactions (such as corrections and feedback) over time to create personalized models that are updated based on incorrect interpretations to improve future responses (Compl. ¶394, ¶407-408, ¶414-415, ¶424-426).
U.S. Patent No. 8,447,607 - “Mobile Systems And Methods Of Supporting Natural Language Human-Machine Interactions,” issued May 21, 2013
- Patent Identification: U.S. Patent No. 8,447,607, “Mobile Systems And Methods Of Supporting Natural Language Human-Machine Interactions,” issued May 21, 2013 (Compl. ¶70).
- Technology Synopsis: The patent claims a device that processes natural language inputs by identifying the user, generating a speech transcription based on a "personal cognitive model" (from the user's prior interactions) and a "general cognitive model" (from a plurality of users' interactions), and using a "context stack" to identify a matching entry and its associated domain agent (Compl. ¶73).
- Asserted Claims: Claim 14 is asserted (Compl. ¶73, ¶449).
- Accused Features: The complaint alleges that Google Assistant infringes by identifying users (e.g., via "Voice Match"), using personal and general interaction history to generate transcriptions, and using conversational context to select a domain agent to handle a request (Compl. ¶450, ¶454-458).
U.S. Patent No. 8,620,659 - “Systems And Methods Of Supporting Adaptive Misrecognition in Conversational Speech,” issued December 31, 2013
- Patent Identification: U.S. Patent No. 8,620,659, “Systems And Methods Of Supporting Adaptive Misrecognition in Conversational Speech,” issued December 31, 2013 (Compl. ¶74).
- Technology Synopsis: The patent describes a method of processing a natural language utterance by using a "personalized cognitive model" (based on the user's interaction patterns) to predict subsequent user actions (Compl. ¶76). If the personalized model lacks sufficient information, the system predicts actions based on a "generalized cognitive model" (based on patterns from a plurality of users) instead (Compl. ¶76).
- Asserted Claims: Claim 42 is asserted (Compl. ¶76, ¶485).
- Accused Features: The complaint alleges infringement by Google's Gemini platform, which allegedly uses personalized models from user history to predict and process requests, and falls back on generalized models derived from multiple users when sufficient personal data is unavailable (Compl. ¶485, ¶498-504).
U.S. Patent No. 8,849,652 - “Mobile Systems And Methods Of Supporting Natural Language Human-Machine Interactions,” issued September 30, 2014
- Patent Identification: U.S. Patent No. 8,849,652, “Mobile Systems And Methods Of Supporting Natural Language Human-Machine Interactions,” issued September 30, 2014 (Compl. ¶77).
- Technology Synopsis: The patent describes a method for processing natural language utterances where recognized words alone are insufficient to determine the request (Compl. ¶80). The method relies on "context information" from prior utterances, which is stored in a "context set" that can be synchronized across multiple devices, to determine the command or request (Compl. ¶80).
- Asserted Claims: Claims 24 and 27 are asserted (Compl. ¶80-81, ¶526).
- Accused Features: The complaint alleges that Google Assistant infringes by using a user's prior conversational history (context), which is synchronized across multiple devices (e.g., phone and smart speaker), to interpret ambiguous commands (Compl. ¶527, ¶529-532, ¶540).
III. The Accused Instrumentality
Product Identification
The complaint identifies the "Accused Google Products and Services," which primarily consist of the Google Assistant Platform and the Google Gemini Platform (Compl. ¶98). This includes related services such as Conversational Actions, App Actions, and Dialogflow virtual agents, as well as the hardware and software that access these platforms, such as smartphones, tablets, smart displays (e.g., Google Home), and vehicles with Android Auto or Android Automotive OS (Compl. ¶98-99). The products are further categorized as "Accused Assistant Products" and "Accused Gemini Products" where specific allegations apply (Compl. ¶100-101).
Functionality and Market Context
The Accused Products are voice-controlled digital assistants that process natural language speech and other inputs to perform a wide variety of tasks for users (Compl. ¶109). According to Google's documentation cited in the complaint, the Assistant functions by receiving a user's request, processing it to "understand the best action to take," and then fulfilling the request (Compl. ¶221). This involves parsing user speech to identify "semantics, i.e. the meaning," understanding context, and selecting an appropriate application or service (a "domain agent") to respond, such as Maps or Search (Compl. ¶117, ¶119). The platforms are "multimodal," meaning they can receive and process combinations of voice, touch, and keyboard inputs on devices with screens and audio outputs (Compl. ¶99). The complaint alleges Google is a major provider of voice recognition products and services (Compl. ¶129).
IV. Analysis of Infringement Allegations
U.S. Patent No. 7,398,209 Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| A method responsive to a user generated natural language speech utterance, comprising: receiving the user generated natural language speech utterance... | Google Assistant responds to user-generated speech utterances, such as "Hey Google, where's the closest dog park?" (Compl. ¶108-111). | ¶108, ¶110 | col. 22:47-51 |
| maintaining a dynamic set of prior probabilities or fuzzy possibilities usable at each stage of processing the received user utterance; | Google Assistant analyzes user questions in combination with "useful information such as recent requests" (Compl. ¶113). | ¶112-113 | col. 22:52-55 |
| recognizing words and phrases contained in the received utterance using information in one or more dictionary and phrase tables; | Google Assistant's NLU uses "trainingPhrases[]" to automatically match words or phrases from user input, which function as phrase tables (Compl. ¶114-115). | ¶114-115 | col. 22:56-59 |
| parsing the recognized words and phrases to determine a meaning of the utterance, wherein determining the meaning includes determining a context for the at least one request...based on one or more keywords... | Google Assistant's NLU models are described as understanding context and identifying "the semantics, i.e. the meaning, of your question" based on keywords to determine what the user is trying to do with a command (Compl. ¶116-117). A Google diagram illustrates routing user utterances based on keywords to a specific "Forecast Intent," which is alleged to be a context (Compl. p. 36). | ¶116-117 | col. 22:60-67 |
| selecting at least one domain agent based on the determined meaning, the selected domain agent being an autonomous executable that receives, processes, and responds to requests associated with the determined context; | Based on the determined context, Google Assistant selects an appropriate service, such as "Maps or Search," to respond to the request. These services are alleged to be domain agents (Compl. ¶118-119). | ¶118-119 | col. 23:1-6 |
| formulating the at least one request contained in the utterance in accordance with a grammar used by the selected domain agent... | Google Assistant is alleged to formulate requests according to a specific structure or grammar for a given domain, such as a weather query, and can "include parameters that partially or entirely fill" the request's parameters (Compl. ¶120-121). | ¶120-121 | col. 23:7-11 |
| invoking the selected domain agent to process the formulated request; and | Google Assistant is described as invoking services like Google Maps to process a query and return results (Compl. ¶122-123). | ¶122-123 | col. 23:12-13 |
| presenting results of the processed request to the user... | Google Assistant provides results to the user "in the appropriate format for your device," which may include a spoken answer or a map with walking directions (Compl. ¶124-125). | ¶124-125 | col. 23:14-18 |
Identified Points of Contention (’209 Patent)
- Scope Questions: A central question may be whether Google's software architecture of "intents" and services like "Google Maps" or "Search" (Compl. ¶119) meets the claim definition of a "domain agent" as an "autonomous executable" (’209 Patent, Claim 1). The analysis could focus on the degree of independence and encapsulated functionality required by the patent's description versus how Google's services are architected and invoked.
- Technical Questions: The claim requires "maintaining a dynamic set of prior probabilities or fuzzy possibilities" ('209 Patent, Claim 1). The complaint cites Google's use of "recent requests" as evidence (Compl. ¶113). A technical question may arise as to whether the specific method by which Google Assistant uses conversational history functionally aligns with the probabilistic or fuzzy logic system described in the patent.
U.S. Patent No. 7,502,738 Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| A system responsive to a user generated natural language speech utterance, comprising: an agent architecture that includes a plurality of domain agents, each...being an autonomous executable... | Google Assistant is described as allowing developers to create applications ("App Actions") for various contexts (e.g., Finance, Games, Transportation), which are alleged to be a plurality of autonomous domain agents (Compl. ¶147-148). The complaint includes a screenshot from Google's developer site listing various application categories (Compl. p. 44). | ¶147-148 | col. 25:7-11 |
| a parser configured to determine a context...and to determine a meaning...wherein the parser selects at least one of the plurality of domain agents based on the determined meaning... | Google Assistant's NLU models are alleged to parse a user's question to identify its meaning and context based on keywords, and then select a domain agent (e.g., Google Maps) to handle the request (Compl. ¶149-150). | ¶149-150 | col. 25:12-20 |
| an event manager configured to coordinate interaction between the parser and the agent architecture; and | Google Assistant is alleged to include software that coordinates interactions between system components, as shown in a Google developer diagram illustrating the flow from user voice request to parser ("Assistant matches request to...BII") to the app ("App processes the fulfillment") (Compl. ¶151-152, p. 46). | ¶151-152 | col. 25:21-23 |
| an update manager that enables the user to purchase one or more domain agents from a third party on a one-time or subscription basis. | Google Assistant allows app makers to "sell subscriptions directly to users" through the Google Play store for applications that include App Actions. This is alleged to be an update manager enabling the purchase of domain agents (Compl. ¶153-154). The complaint provides a screenshot of a news article about this feature and the Google Play store interface (Compl. p. 47). | ¶153-154 | col. 25:24-27 |
Identified Points of Contention (’738 Patent)
- Scope Questions: The infringement theory hinges on the interpretation of the term "update manager that enables the user to purchase one or more domain agents from a third party" (’738 Patent, Claim 1). A key legal question will be whether the Google Play store, which allows users to download or subscribe to third-party applications that integrate with Google Assistant (Compl. ¶154), constitutes the claimed "update manager." The analysis may explore whether "purchase" requires a direct monetary transaction for the "agent" itself, or if it can be construed more broadly to include downloading free apps or apps with in-app subscription models.
- Technical Questions: What evidence does the complaint provide that the "App Actions" available through the Google Play store are "autonomous executables" as required by the "domain agents" limitation? The dispute may focus on the technical implementation of App Actions and whether they are self-contained executable modules or merely API endpoints for existing applications.
V. Key Claim Terms for Construction
The Term: "domain agent" (from ’209 Claim 1 and ’738 Claim 1)
Context and Importance
This term is foundational to the architecture claimed in both lead patents. The infringement case rests on whether Google's "intents," "Actions," and integrated services (like Maps, Search, and third-party apps) qualify as "domain agents." Practitioners may focus on this term because its construction will determine if Google's fundamentally different, service-oriented architecture can be read upon by claims rooted in a model of "autonomous executables."
Intrinsic Evidence for Interpretation
- Evidence for a Broader Interpretation: The specification describes agents as modules that "organize domain specific behavior and information" and can be simple "links to information" (’209 Patent, col. 2:53-54, col. 12:56-57). This language could support construing the term to cover a wide range of software components, including API-based services that handle specific domains.
- Evidence for a Narrower Interpretation: The claims explicitly define the agent as an "autonomous executable" (’209 Patent, Claim 1; ’738 Patent, Claim 1). This language, combined with specification descriptions of agents as "re-distributable packages of executable code" (’209 Patent, col. 11:50-53), may support a narrower construction requiring a self-contained, standalone software package rather than a service endpoint or intent handler.
The Term: "an update manager that enables the user to purchase one or more domain agents from a third party" (from ’738 Claim 1)
Context and Importance
This limitation is the key distinguishing feature of the asserted claim of the ’738 Patent. The viability of the infringement allegation against the Google Play store turns entirely on the scope of "purchase" and whether the store's functionality meets the "update manager" definition.
Intrinsic Evidence for Interpretation
- Evidence for a Broader Interpretation: The specification states the "update manager is used to add new agents to the system" (’738 Patent, col. 2:63-64) and mentions "license management capabilities allowing the sale of agents... on a one time or subscription basis" (’738 Patent, col. 3:1-5). This could be argued to broadly cover any platform, like an app store, that manages the addition of third-party functionality, whether free or paid.
- Evidence for a Narrower Interpretation: The claim's use of the word "purchase" may suggest a requirement for a direct monetary transaction to acquire the "domain agent" itself. An argument could be made that a platform for distributing free applications, or applications where payment is for content rather than the agent functionality itself, does not meet the claim's specific language.
VI. Other Allegations
Indirect Infringement
The complaint alleges both induced and contributory infringement for all asserted patents. Inducement allegations are based on Google providing instructions, documentation, and APIs to third-party developers and end-users, encouraging them to build and use applications (e.g., "App Actions") and devices in a manner that allegedly practices the claimed methods (Compl. ¶132-133, 160-161). Contributory infringement is alleged based on Google selling components, such as processors and memory containing specific software routines, that are especially adapted for infringement and have no substantial non-infringing use (Compl. ¶134, ¶162).
Willful Infringement
Willfulness is a central theme of the complaint. The allegations are primarily based on extensive pre-suit knowledge. The complaint details a multi-year history, from 2007 to 2012, where VoiceBox (the original patent owner) allegedly disclosed its patent portfolio, including issued patents and pending applications that matured into the patents-in-suit, directly to Google's executives, engineers, and patent counsel (Compl. ¶82-90). This alleged knowledge predates the 2016 launch of the accused Google Assistant platform by several years. Post-suit knowledge is also asserted based on the filing of the initial complaint in 2023 (Compl. ¶129).
VII. Analyst’s Conclusion: Key Questions for the Case
A core issue will be one of architectural equivalence: Can Plaintiff demonstrate that Google's modern, service-oriented ecosystem of "intents" and "Actions" built on APIs and integrated with the Google Play store is functionally and structurally equivalent to the patent portfolio's architecture, which describes a system of discrete, "autonomous executable" programs referred to as "domain agents" that are "purchased" via an "update manager"?
A key factual question will center on willfulness and the nature of pre-suit knowledge: Given the extensive allegations that VoiceBox disclosed its specific patents and patent applications to Google executives and lawyers years before the accused products were developed, the case will likely examine what Google knew, when it knew it, and whether its subsequent actions in developing and launching Google Assistant constituted deliberate or willfully blind infringement of patents it was allegedly aware of.
A third pivotal question will be one of claim scope and evolution: How will the court construe foundational terms like "domain agent" and "context," which were patented in the early 2000s, in the context of today's vastly more complex and interconnected conversational AI platforms? The case may turn on whether these terms are interpreted broadly enough to read on modern architectures or are limited to the specific embodiments described in the patents' specifications.