Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
Claims 1-7 and 11-20 are pending. Claims 1, 13, and 19 are independent. Independent Claims are amended. Claim 11 was amended to overcome an Objection.
This Application was published as U.S. 20250272509.
Apparent priority: 27 February 2024 (claiming priority to four provisional applications).
Applicant’s amendments and arguments are considered but are either unpersuasive or moot in view of the new grounds of rejections.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 8/25/2026 has been entered.
Response to Amendments and Arguments
Applicant’s arguments are directed to the amended language and are moot in view of the modified grounds of rejection. Seibel is a very strong reference and maintained and response to arguments regarding Seibel are provided below.
Amendments to Claim 11 overcome the Objection to this Claim.
Claim 1 is amended as follows and the other independent Claims are amended similarly:
1. A computing services environment comprising:
a database system storing a plurality of database records for a plurality of client organizations accessing computing services via the computing services environment, the computing services including a conversational chat assistant;
an application server receiving user input for the conversational chat assistant via the Internet;
a generative language model interface providing access to one or more generative language models;
processor providing an orchestration and planning service configured to identify a plurality of actions based on the user input and to execute the plurality of actions to determine a natural language response message,
wherein identifying the plurality of actions includes:
(1) transmitting a topic identification input prompt to a generative language model and receiving a topic identification prompt completion identifying a topic of a plurality of topics, the topic corresponding to a respective topic-based subset of a plurality of predetermined actions, and
(2) transmitting, based on the identified topic, an intent identification input prompt identifying the topic-based subset by respective action descriptions and unique identifiers to the generative language model and receiving an intent identification input prompt completion identifying the plurality of actions by a corresponding plurality of unique identifiers, the plurality of actions being a subset of the topic-based subset of the plurality of predetermined actions;
a storage system storing a metadata framework specifying information related to the conversational chat assistant, the metadata framework including a definition associated with an action of the plurality of actions, the definition including one or more inputs, one or more outputs, and one or more operations performed via the computing services environment; and
a communication interface configured to transmit the natural language response message to a client machine via the application server.
Applicant has pointed to [0066], [0070], [0078], [0086], [0109], and [0113] of Specification as filed which are the same as those published:
PNG
media_image1.png
198
726
media_image1.png
Greyscale
PNG
media_image2.png
164
728
media_image2.png
Greyscale
PNG
media_image3.png
128
702
media_image3.png
Greyscale
PNG
media_image4.png
276
716
media_image4.png
Greyscale
PNG
media_image5.png
190
708
media_image5.png
Greyscale
PNG
media_image6.png
366
698
media_image6.png
Greyscale
PNG
media_image7.png
196
700
media_image7.png
Greyscale
Applicant’s arguments try to address the question posed in the previous Office action regarding the significance of the action identifiers?
PNG
media_image8.png
86
650
media_image8.png
Greyscale
PNG
media_image9.png
172
660
media_image9.png
Greyscale
PNG
media_image10.png
390
638
media_image10.png
Greyscale
PNG
media_image11.png
330
660
media_image11.png
Greyscale
Response at 8.
It appears that:
1) The user inputs what he wants in Natural language to an LLM.
2) The LLM takes the description of user’s intent and generates a plan of action.
3) The plan of action includes a series of actions which are identified by action identifiers.
However, between the first and second steps above, the LLM identifies topics in the user’s natural language statement of intent. Each topic is associated with a subset of the set of all of the actions. So, in order to generate a plan of actions the LLM provides the subset of action actions as identified by their action identifiers as output.
Intent [Wingdings font/0xE8] Topic [Wingdings font/0xE8] Subset of all Available Actions corresponding to Topic [Wingdings font/0xE8] Plan
The actions in a subset have to be identified somehow in the plan of action and they are identified by their identifiers.
Examiner had requested reference to a Figure.
The steps mentioned fall in different flowcharts:
PNG
media_image12.png
210
692
media_image12.png
Greyscale
PNG
media_image13.png
638
1070
media_image13.png
Greyscale
Figure 6 is used as most representative of the Claim:
PNG
media_image14.png
562
354
media_image14.png
Greyscale
Siebel, Figure 1, “pre-processing 104” which identifies the actions in the “input 102” and sends them to the proper “Agent 106-1 …N” is an LLM. See [0028].
Each Agent 106-1 … N is associated with a subset of all of Tools 108-1 … N.
This is like each topic of the Claim being associated with a subset of actions.
Agents are topic/domain/subject specific.
Tools are more elemental and groups/subsets of all of the tools are put together for each domain/subject/topic/Agent.
PNG
media_image15.png
380
794
media_image15.png
Greyscale
PNG
media_image16.png
564
721
media_image16.png
Greyscale
So far, there is a one-to-one correspondence between Seibel and Claim or Application.
Remains the issue of PROMPTS to the LLM.
The Claim lets the LLM identify the Topics and then sends another prompt to the LLM to identify the subset of actions.
1) Topic Identification Input Prompt [Wingdings font/0xE8] topics
2) Intent Identification Input Prompt [Wingdings font/0xE8] actions (or action identifiers)
1. A computing services environment comprising:
a database system storing a plurality of database records for a plurality of client organizations accessing computing services via the computing services environment, the computing services including a conversational chat assistant;
an application server receiving user input for the conversational chat assistant via the Internet;
a generative language model interface providing access to one or more generative language models;
processor providing an orchestration and planning service configured to identify a plurality of actions based on the user input and to execute the plurality of actions to determine a natural language response message,
wherein identifying the plurality of actions includes:
(1) transmitting a topic identification input prompt to a generative language model and receiving a topic identification prompt completion identifying a topic of a plurality of topics, the topic corresponding to a respective topic-based subset of a plurality of predetermined actions, and
(2) transmitting, based on the identified topic, an intent identification input prompt identifying the topic-based subset by respective action descriptions and unique identifiers to the generative language model and receiving an intent identification input prompt completion identifying the plurality of actions by a corresponding plurality of unique identifiers, the plurality of actions being a subset of the topic-based subset of the plurality of predetermined actions;
a storage system storing a metadata framework specifying information related to the conversational chat assistant, the metadata framework including a definition associated with an action of the plurality of actions, the definition including one or more inputs, one or more outputs, and one or more operations performed via the computing services environment; and
a communication interface configured to transmit the natural language response message to a client machine via the application server.
However, the existence of two prompts in this order is not supported by the Disclosure as filed and is considered new matter.
See Figures 5 and 6 and description.
[0083] Input is received via a communication channel at 502. In some embodiments, the input may include natural language text. Alternatively, or additionally, the input may include other types of information, such as a selection of an action to perform based on a button provided in a chat interface, a request sent by an application or workflow, or another such input indicator.
[0084] In some embodiments, the communication channel may be a conversational chat interface. For instance, a conversational chat interface may be provided via a web application, mobile application, or other such service. Alternatively, the communication channel may be a messaging service such as email, SMS, Slack, WhatsApp, or any other suitable service for sending and receiving messages.
[0085] A plan to execute the user's intent as reflected in the input is determined at 504. In some embodiments, the user input may include an explicit selection of a workflow, action, or other predefined operations. For instance, the input may include a selection of a button corresponding to an action and presented in a conversational chat interface. In such a situation, the action or actions to be performed may be selected from the predefined operations.
[0086] In some embodiments, the user input may be provided via natural language. In such a situation, the user's intent may be less clear and may be determined based on one or more interactions with a generative language model. For instance, natural language text included in the input may be used to determine an intent identification input prompt. The intent identification input prompt may include the input text, a natural language request executable by a generative language model, and/or other types of information. For instance, the intent identification input prompt may include a description of actions capable of being performed via the conversational chat assistant. The generative language model may then generate novel text that includes one or more identifiers corresponding with the actions to be performed based an analysis of the intent in the input text by the generative language model. Additional details regarding a method for determining the user's intent are discussed with respect to the method 600 shown in FIG. 6.
PNG
media_image17.png
480
747
media_image17.png
Greyscale
Figure 6 was provided above. The Description provides:
[0099] A topic selection input prompt is determined at 606. The topic selection input prompt includes some or all of the natural language user input and a description of a set of topics. The topic selection input prompt may instruct the generative language model to select from the set of topics for the purpose of identifying prospective actions to perform to fulfill the intent reflected in the user's input.
[0100] According to various embodiments, the particular topics that may be selectable may depend upon the context. For example, the computing services environment 150 may provide a set of default topics, such as database system interaction, service-related operations, sales-related operations, and the like. As another example, one or more topics may be tailored to specific industries, organizations, individuals, or other contexts.
[0101] The topic is identified at 608 based on a topic selection prompt completion provided by a generative language model. For instance, the generative language model may generate novel text that includes an identifier corresponding to the topic that the generative language model identifies as being most closely related to the user's intent. The identifier may be extracted from the topic selection prompt completion by the computing services environment 150.
[0102] In some embodiments, the generative language model may identify more than one topic. For instance, the generative language model may identify the user's intent as being related to sales operations and payment processing topics.
[0103] A subset of available actions is determined at 610 based on the identified topic. In some embodiments, an action may be any operation or combination of operations capable of being performed via the computing services environment 150. For instance, an action may include a prompt completed by a generative language model, one or more database operations, an API request, or another type of operation.
Based on the Disclosure:
User inputs what it wants (intent) and it could be by direct selection on a screen which leaves no ambiguity.
BUT, if his input is in NL and is ambiguous then a LLM is recruited and an Intent Identification Input Prompt is input to the LLM which generates the actions (action identifiers) that satisfy the intent if executed.
IF we want a more sophisticated operation instead of the Intent Identification Input Prompt, a Topic Identification Input Prompt is input to the LLM which generates a set of Topics and each Topic has its own group (subset) of actions.
It is one or the other; not both; and not in the succession claimed.
Channapattan and Fozdar
Siebel teaches all the main concepts.
Channapattan remains as a reference because neither Siebel nor Fozdar use the word “client” and don’t discuss transmission in a distributed system.
Fozdar is also very on point and is cited because it teaches the “identifier” expressly. The Tools/actions of Seibel must have identifiers specially that there a pool of Tools and then the Agents are associated with overlapping subsets. But the word identifier is not in Seibel. Further Fozdar teaches going back to the LLM like the current Claim (although not supported in the Claim).
“[0023] … As another example, the present techniques utilize an LLM to identify the specific data sources that are relevant to the natural language query, which is a complex task due to the large, evolving nature of data source and destination schema types. In particular, data integration applications often support over one hundred different connector types, and enterprises may utilize hundreds of different data sources. The complexity of this task is also increased due to the potential variability and vague nature of the provided user input. Therefore, by utilizing an LLM to perform this process, the present techniques provide a significant advantage over conventional data integration techniques. As another example, the present techniques utilize an LLM to conduct the appropriate reformulations for translating the natural language query into a set of ordered, dependent activities.”
“[0026] … The task identification is a unique identifier for task planning, where the order of the tasks is used as a reference to ensure that activities are generated in the correct order. …”
PNG
media_image18.png
472
707
media_image18.png
Greyscale
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-7 and 11-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Independent Claims include new matter and the dependent inherit.
1. A computing services environment comprising:
a database system storing a plurality of database records for a plurality of client organizations accessing computing services via the computing services environment, the computing services including a conversational chat assistant;
an application server receiving user input for the conversational chat assistant via the Internet;
a generative language model interface providing access to one or more generative language models;
processor providing an orchestration and planning service configured to identify a plurality of actions based on the user input and to execute the plurality of actions to determine a natural language response message,
wherein identifying the plurality of actions includes:
(1) transmitting a topic identification input prompt to a generative language model and receiving a topic identification prompt completion identifying a topic of a plurality of topics, the topic corresponding to a respective topic-based subset of a plurality of predetermined actions, and
(2) transmitting, based on the identified topic, an intent identification input prompt identifying the topic-based subset by respective action descriptions and unique identifiers to the generative language model and receiving an intent identification input prompt completion identifying the plurality of actions by a corresponding plurality of unique identifiers, the plurality of actions being a subset of the topic-based subset of the plurality of predetermined actions;
a storage system storing a metadata framework specifying information related to the conversational chat assistant, the metadata framework including a definition associated with an action of the plurality of actions, the definition including one or more inputs, one or more outputs, and one or more operations performed via the computing services environment; and
a communication interface configured to transmit the natural language response message to a client machine via the application server.
The Specification and drawings do not provide for two input prompts in succession in the manner claimed.
See the Response to Arguments section for a detailed analysis under the heading of “Remains the issue of PROMPTS to the LLM.”
There is one prompt to the LLM. In Figure 5 it is called intent identification and Figure 6 topic identification. Both lead to the subset of actions. Figure 6 has an extra step of identifying topics first and actions based on topics whereas Figure 5 goes to actions straight.
Figure 6, which begins with topic identification prompt, does not follow with an intent identification prompt for identifying the actions. Once topic is determined, the actions associated with topic are known or at any rate are not generated by the LLM.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-7 and 11-20 are rejected under 35 U.S.C. 103 as being unpatentable over Siebel (U.S. 20240202225) in view of Channapattan (U.S. 20250110976) and further in view of Fozdar (U.S. 20250077538).
Regarding Claim 1, Siebel teaches:
1. A computing services environment comprising:
a database system storing a plurality of database records for a plurality of client organizations accessing computing services via the computing services environment, [Siebel, Figure 3, “datastore 318” and “DB engine 316.” Figure 4 shows “enterprise systems 404-1 …N” which teach the “client organizations” of the Claim. “[0052] … Enterprise systems 404 can include data flow and management of different processes (e.g., of one or more organizations) and can provide access to systems and users of the enterprise while preventing access from other systems and/or users. … references to enterprise information environments can also include enterprise systems, …. In various embodiments, functionality of the enterprise systems 404 may be performed by one or more servers (e.g., a cloud-based server) and/or other computing devices.” “[0048] In some embodiments, generative artificial intelligence models (e.g., large language models of an orchestrator) of the enterprise generative artificial intelligence system 402 can interact with agents (e.g., retrieval agents, retriever agents) to retrieve and process information from various data sources. For example, data sources can store data records and/or segments of data records which may be identified by the enterprise generative artificial intelligence system 402 based on embedding values (e.g., vector values associated with data records and/or segments). Data records can include tables, text, images, audio, video, code, application outputs (e.g., predictive analysis and/or other insights generated by artificial intelligence applications), and/or the like.”] the computing services including a conversational chat assistant; [Siebel, Figures 1 and 2 show the “input 102” and the “input layer 202” “[0032] The input layer 202 represents a layer of the enterprise generative artificial intelligence system architecture that receives an input (e.g., a query, complex input, instruction set, and/or the like) from a user or system. For example, an interface module of the enterprise generative artificial intelligence system may receive the input.” Figure 3, “chat 350” and “[0159] … For example, a previous conversation 864 (e.g., as part of a chat with a chat bot) may have included a conversation about France. …”]
an application server receiving user input for the conversational chat assistant via the Internet; [Siebel, Figure 3, “application hosting and application engine 360.” “[0042] In some embodiments, a user query 362 and/or other inputs may be received by an application hosting an application engine 360 which can communicate with a low latency engine 358 to provide the input, or a transformed input, to the orchestrator 342. The orchestrator 342 may utilize the various agents, large language models, and other features to generate an accurate and reliable (e.g., without hallucination) answer to the user query 362.”]
a generative language model interface providing access to one or more generative language models; [Siebel, Figure 3, “[0041] In the example of FIG. 3, the enterprise generative artificial intelligence system includes an orchestrator 342 with a fine-tuned large language model. The orchestrator 342 and/or agents 326-339 may include and/or access task-specific large language models 348-356, as well as external or third-party large language models 340 in some embodiments….”]
processor providing [Siebel, Figure 15, “processor 1504.”] an orchestration and planning service configured to identify a plurality of actions based on the user input and to execute the plurality of actions to determine a natural language response message; [Siebel. The “actions” of the Claim are mapped to the “tools” or “tasks” of Siebel that are performed by the various “agents.” “[0023] … Agents can include one or more multimodal models (e.g., large language models) to accomplish the prescribed tasks using a variety of different tools. Different agents can use various tools to execute and process unstructured data retrieval requests, structured data retrieval requests, API calls (e.g., for accessing artificial intelligence application insights), and the like. Tools can include one or more specific functions and/or machine learning models to accomplish a given task (or set of tasks).” Figure 6, “orchestrator 604” receiving the “user query 602” and “choosing tool 608” / “plurality of actions” to orchestrate the query and then finally generate the “summary 620” / “natural language response message” to be output as the response at 632. Figure 3, 342. “[0041] In the example of FIG. 3, the enterprise generative artificial intelligence system includes an orchestrator 342 with a fine-tuned large language model.…” “[0025] The orchestrator manages the agents to efficiently process disparate inputs or different portions of an input. For example, an input may require the system to access and retrieve data records from disparate data sources (e.g., unstructured datastores, structured datastores, timeseries datastores, and the like), database tables from different types of databases, and machine learning insights from different machine learning applications. The different agents can each separately, and in parallel, handle each of these requests, greatly increasing computational efficiency.”]
wherein identifying the plurality of actions includes: [Siebel, Figure 1, “pre-processing 104” which identifies the tasks/actions in the “input 102” and sends them to the proper “Agent 106-1 …N” is an LLM. (Agents 106 are mapped to topics of the Claim each of which in turn correspond to a plurality of Tools 108 which are mapped to the actions of the Claim. Both Seibel and instant Application use “action”.) Figure 2, the “supervisory layer 210” is shown as a “language model 206.” “[0028] An orchestrator agent (or, simply, orchestrator) can pre-process the input in step 104. Pre-processing can include … input identification (e.g., identifying different portions of the input 102 for processing by different agents). The orchestrator can use a multimodal model (e.g., large language model) to further process the input 102 to create a plan for determining a result (step 112) for the input. The plan may include a prescribed set of tasks, such as structured data retrieval tasks, unstructured data retrieval tasks, timeseries processing tasks, visualization tasks, and the like. In some embodiments, the plan can designate which tools 108 should be used to execute the tasks, and the orchestrator can select the agents based on the designated tools. In some embodiments, the plan can designate which agents should be used to execute the tasks, and the agents can independently designate which tools 108 should be used to execute the tasks.” “[0033] The supervisory layer 210 represents a layer of the enterprise generative artificial intelligence system architecture that includes one or more large language models (e.g., of an orchestrator module) that can develop a plan for responding to the input received in the input layer 202….”]
(1) transmitting a topic identification input prompt to a generative language model and receiving a topic identification prompt completion identifying a topic of a plurality of topics, the topic corresponding to a respective topic-based subset of a plurality of predetermined actions, and [Siebel, the actions of Seibel are sent to particular Agents 106-1 … N and then the Agent selects the Tools 108-1 …N and one Agent/Topic of the Claim may select several tools as shown in Figure 1. The input that identifies the Agent 106 (topic of this Claim) teaches the “topic identification prompt” whose completion identifies the topics/Agents from a plurality of topics /Agents 106-1 …N. Basically, the user NL input to the system teaches the topic identification input prompt of this Claim. See response to arguments.]
(2) transmitting, based on the identified topic, an intent identification input prompt identifying the topic-based subset by respective action descriptions and unique identifiers to the generative language model and receiving an intent identification input prompt completion identifying the plurality of actions by a corresponding plurality of unique identifiers, the plurality of actions being a subset of the topic-based subset of the plurality of predetermined actions; [Siebel, the Tools 108-1…N of Figure 1 or 252 of Figure 2 teaches the “plurality of actions” of this Claim. The plurality of tools 108/actions fall under particular Agents 106 /topics and are subsets of the Agents 106. In Figure 1, e.g., Agent 106-1 (topic of this Claim) invokes Tools 108-1 and 108-2 (actions of this Claim that fall under a particular topic) whereas Agent 106-2 (different topic) invokes only Tool 108-2. So, each Agent/Topic has a subset of Tools/Actions associated with it. Figure 7 also shows a number of “Tool Choose …” (choice of actions that are a subset of a topic) options. “[0030] The agents 106 can select the appropriate tools 108 to accomplish a set of prescribed tasks (e.g., tasks prescribed by the orchestrator). The tools 108 can make the appropriate function calls to retrieve disparate data records among other functions….” Thus the “intent identification input prompt completion” that identifies a plurality of actions/tools 108, is the output of the agents 106 which are LLMs in their own right. Seibel as provided in the response to Arguments does not input a second prompt (but neither does the instant Application as noted in the 112(a) new matter rejection.)]
a storage system storing [Siebel, Figure 15, “memory 1506” and “storage 1508.”] a metadata framework specifying information related to the conversational chat assistant, the metadata framework including a definition associated with an action of the plurality of actions, the definition including one or more inputs, one or more outputs, and one or more operations performed via the computing services environment; and [Siebel, Figure 3, 318: Metadata Store. “[0039] … The datastores 318 can include vector datastores (e.g., FAISS implementation), metadata datastores ….” “[0134] The system may leverage characteristics of a model driven architecture, which represent system objects (e.g., components, functionality, data, etc.) using rich and descriptive metadata, to dynamically generate queries for conducting searches across a wide range of data domains (e.g., documents, tabular data, insights derived from AI applications, web content, or other data sources)….”]
a communication interface configured to transmit the natural language response message to a client machine via the application server. [Siebel, Figure 4, “communication network 408” / “communication interface” operating between servers and clients and the generative AI system 402: “[0052] The enterprise systems 404 can include enterprise applications (e.g., artificial intelligence applications), enterprise datastores, client systems, and/or other systems of an enterprise information environment. As used herein, an enterprise information environment can include one or more networks (e.g., cloud, on premise, air-gapped or otherwise) of enterprise systems (e.g., enterprise applications, enterprise datastores), client systems (e.g., computing systems for access enterprise systems)….”]
Siebel is arguably a 102 reference.
Siebel does not include the term “client organization” expressly.
Channapattan teaches:
a database system storing a plurality of database records for a plurality of client organizations accessing computing services via the computing services environment, the computing services including a conversational chat assistant; [Channapattan, “Methods, systems, and devices for processing a natural language request are described. An identity management system may receive a user request for information maintained in the identity management system and related to a client organization. The request may be received in a natural language form. In response to the user request, a machine learning model may be employed to generate a query in a machine-readable language that is understandable by the identity management system…. Based on receiving a selection of a portion of the information output for display, the machine learning model may be employed to generate a natural language explanation of the selected portion. In some cases, the natural language explanation may be a summarization of information associated with the selected portion and retrieved from multiple data sources.” Abstract. “[0065] The described techniques provide a simplified and streamlined way for administrators of client organizations to conveniently access information maintained in the identity management system about their client organization, thereby improving the user experience and ensuring that the client organizations are able to access the most accurate and up-to-date information. Further, by employing a machine learning model to generate machine-readable queries to retrieve information responsive to a user's query, the identity management system may avoid receiving queries written and executed by administrators who may issue queries that are poorly written or not optimized for the identity management system's databases. Executing such queries may result in long running queries that degrade performance at the identity management system.”]
(1) transmitting a topic identification input prompt to a generative language model and receiving ….[Channapattan includes the back and forth transmissions: “[0082]The user query may be transmitted to one or more modules of the natural language interface system 230. In some cases, the user query may initially be transmitted to the prompt injection prevention module 240….”]
(2) transmitting, based on the identified topic, an intent identification input prompt … [Channapattan includes the back and forth transmissions because it hasa distributed system: [0127] For example, the software module 720 may include a user request component 725, a generative AI component 730, a data retrieval component 735, a response output component 740, or any combination thereof. In some examples, the software module 720, or various components thereof, may be configured to perform various operations (e.g., receiving, monitoring, transmitting) using or otherwise in cooperation with the input module 710, the output module 715, or both. For example, the software module 720 may receive information from the input module 710, send information to the output module 715, or be integrated in combination with the input module 710, the output module 715, or both to receive information, transmit information, or perform various other operations as described herein.”]
PNG
media_image19.png
594
818
media_image19.png
Greyscale
Siebel and Channapattan pertain to the use of natural language interface to query databases pertaining to data of various enterprises and it would have been obvious to combine the express mention of Channapattan to client organizations having their own databases with the system of Siebel which does include implied and indirect teachings of client organizations for a more solid teaching. This combination falls under combining prior art elements according to known methods to yield predictable results or use of known technique to improve similar devices (methods, or products) in the same way. See MPEP 2141, KSR, 550 U.S. at 418, 82 USPQ2d at 1396.
Neither teach the “identifying the plurality of actions by a corresponding plurality of unique identifiers” or a second prompt.
Fozdar teaches:
PNG
media_image20.png
680
456
media_image20.png
Greyscale
(2) transmitting, based on the identified topic, an intent identification input prompt identifying the topic-based subset by respective action descriptions and unique identifiers to the generative language model [Fozdar, Figure 2, teaches at 208 generate via LLM set of ordered activities for a task which requires a prompt to the LLM and at 210 again select via LLM the APIS for each activity which requires a second prompt to the LLM and thus teaches the two prompts of this Claim. “[0026] … The task identification is a unique identifier for task planning, where the order of the tasks is used as a reference to ensure that activities are generated in the correct order. …” “[0036] At block 208, a set of ordered activities corresponding to the data integration task represented by the natural language input is generated via an LLM….Moreover, in some such embodiments, the natural language input is parsed into multiple activities via the LLM based on … a task identification… for example.” “[0037] At block 210, one or more APIs for performing each activity within the set of ordered activities are selected via the LLM….”] and receiving an intent identification input prompt completion identifying the plurality of actions by a corresponding plurality of unique identifiers, the plurality of actions being a subset of the topic-based subset of the plurality of predetermined actions; [Fozdar, Figure 1, the “task identification” is a “unique identifier” for the particular task/action. “[0026] More specifically, the natural language input 106 is provided to an activity planning module 110 of the natural language-based data integration system 100. In various embodiments, the activity planning module 110 then determines one or more datasets corresponding to the natural language input 106 and performs a dataset lookup operation, as indicated by box 112. The activity planning module 110 also calls the LLM described herein, as indicated by box 114, and utilizes the LLM to parse the natural language input 106 into multiple tasks, where each task corresponds to a distinct activity within a corresponding pipeline. The activity planning module 110 further utilizes the LLM to determine the activity execution order and activity dependencies. In various embodiments, this is accomplished by utilizing specification-based instructions and demonstration-based examples in the corresponding prompt, where the specification-based instructions may consist of a uniform template that allows the LLM to perform task parsing via slot filling. As a non-limiting example, there may be at least four slots for task parsing: task type, task identification, task dependencies, and task arguments. In this example, the task type covers all the activities that can be generated (e.g., get-metadata, for-each, copy, lookup, email notifications, and the like). The task identification is a unique identifier for task planning, where the order of the tasks is used as a reference to ensure that activities are generated in the correct order. …” See also: “[0020] Moreover, to build complex and realistic data pipelines, the LLM is utilized to control, manage, and call various APIs that bring in the correct contextual information for generating each step of the data pipeline. In particular, in various embodiments, during the API selection phase of the data integration process, the LLM is utilized to call one or more activity-specific APIs for each activity identified during the activity planning phase, where such APIs are intended to provide the context for generating each corresponding activity JSON correctly. Furthermore, in various embodiments, during the API execution phase of the data integration process, each invoked API is then executed, and the results are returned.” “[0021] …More specifically, in various embodiments, after the LLM generates the ordered set of activities for the data pipeline and identifies the appropriate API(s) for each activity, the JSON generation phase of the data integration process is executed….” “[0023] …As another example, the present techniques utilize an LLM to identify the specific data sources that are relevant to the natural language query, which is a complex task due to the large, evolving nature of data source and destination schema types…”]
Siebel/Channapattan and Fozdar pertain to the use of LLMs for planning a task as a series of sub-tasks (actions, intents, skills) and it would have been obvious to use the “unique identifier” of Fozdar that is used for identifying tasks in the system of combination in order to assign the intents/actions/tasks each a unique identifier which streamlines the process and avoids ambguity. This combination falls under combining prior art elements according to known methods to yield predictable results or use of known technique to improve similar devices (methods, or products) in the same way. See MPEP 2141, KSR, 550 U.S. at 418, 82 USPQ2d at 1396.
Regarding Claim 2, Siebel teaches:
2. The computing services environment recited in claim 1, wherein an action of the plurality of actions comprises retrieving one or more database records from the database system, the one or more database records being associated with a client organization of the plurality of client organizations. [Siebel, “tools 108” and “tasks” of Siebel teach the “action” of the Claim and the performance of actions/tasks involve retrieving data records in Siebel: “[0025] … For example, an input may require the system to access and retrieve data records from disparate data sources ….” “[0030] The agents 106 can select the appropriate tools 108 to accomplish a set of prescribed tasks (e.g., tasks prescribed by the orchestrator). The tools 108 can make the appropriate function calls to retrieve disparate data records among other functions….” “[0030] The agents 106 can select the appropriate tools 108 to accomplish a set of prescribed tasks (e.g., tasks prescribed by the orchestrator). The tools 108 can make the appropriate function calls to retrieve disparate data records among other functions….”]
Regarding Claim 3, Siebel teaches:
3. The computing services environment recited in claim 2,
wherein an action of the plurality of actions comprises generating a summary of the one or more database records via a generative language model, and [Siebel, Figure 3, “Task Specific Fine Tuned LLMs 346” include the “Summarization 352.” Figure 6, “Query DB for relevant Docs 614” and “Query DB for relevant data 622.” Both paths end in “summary 620” and “visualization summary 630.”]
wherein the natural language response message includes the summary. [Siebel, Figure 6, “User Query 602” as input and “Summary of Tool Outputs as Response 632” as output. “[0026] Agents can process the disparate data returned by the different agents and/or tools. For example, large language models typically receive inputs in natural language format. The agents may receive information in a non-natural language format (e.g., database table, image, audio) from a tool and transform it into natural language describing the tool output in a format understood by large language models. A large language model can then process that input to “answer,” or otherwise satisfy the initial input.”]
Regarding Claim 4, The “actions” of the Claim were mapped to the “tools” or “tasks” of Siebel that are performed by the various “agents.” [0023] … Agents can include one or more multimodal models (e.g., large language models) to accomplish the prescribed tasks using a variety of different tools. Different agents can use various tools to execute and process unstructured data retrieval requests, structured data retrieval requests, API calls (e.g., for accessing artificial intelligence application insights), and the like. Tools can include one or more specific functions and/or machine learning models to accomplish a given task (or set of tasks).”
Siebel does not expressly teach storing information in databases.
Channapattan teaches:
4. The computing services environment recited in claim 1,
wherein an action of the plurality of actions comprises storing information to the database system. [Channapattan teaches the storage of several types of information including the generated output: “[0084] … The anonymization module 250 may further cache, or otherwise persist or store, the identified personally-identifiable information. Such techniques may prevent or reduce a likelihood of having personal or sensitive information injected into the machine learning model….” “[0086 ]… The generative AI module 260 may select the one or more prompts from a prompt store. The prompt store may be maintained in the database 290. The generative AI module 260 may select the one or more prompts based on the user query, such as based on a determined intent of the user query….” “[0126] The output module 715 may manage output signals for the device 705. For example, the output module 715 may receive signals from other components of the device 705, such as the software module 720, and may transmit these signals to other components or devices. In some examples, the output module 715 may transmit output signals for display in a user interface, for storage in a database or data store, for further processing at a server or server cluster, or for any other processes at any number of devices or systems. In some cases, the output module 715 may be a component of an I/O controller 910 as described with reference to FIG. 9.” “[0172] The database controller 915 may manage data storage and processing in a database 935. In some cases, a user may interact with the database controller 915. In other cases, the database controller 915 may operate automatically without user interaction. The database 935 may be an example of a single database, a distributed database, multiple distributed databases, a data store, a data lake, or an emergency backup database.”]
Siebel and Channapattan pertain to the use of natural language interface to query databases and it would have been obvious to add the storing of the output to databases of the system, as taught by Channapattan, to the system of Siebel as one additional use. This combination falls under combining prior art elements according to known methods to yield predictable results or use of known technique to improve similar devices (methods, or products) in the same way. See MPEP 2141, KSR, 550 U.S. at 418, 82 USPQ2d at 1396.
Regarding Claim 5, Siebel teaches and the teaching suggests:
5. The computing services environment recited in claim 1,
wherein an action of the plurality of actions comprises authenticating a user account associated with the user input. [Seibel, Figure 5, “enterprise access control module 514” evaluated authorization for access which suggests authenticating the user. “[0120] In some implementations, the enterprise access control module 514 may evaluate (e.g., using access control lists) whether a user is authorized to access all or only a portion of a result (e.g., answer). For example, a user can provide a query associated with a first department or sub-unit of an organization. Members of that department or sub-unit may be restricted from accessing certain pieces of data, types of data, data models, or other aspects of a data domain in which a search is to be performed….”] (Authentication means verifying that a user is actually who he says he is. Evaluating authorization for access is not the same as authentication because it can be done by other means but does include and therefore does suggest authentication.)
Regarding Claim 6, Siebel teaches:
6. The computing services environment recited in claim 1, further comprising a conversational chat studio configured to customize the conversational chat assistant based on graphical user input provided via a graphical user interface. [Siebel, Figure 2, “dashboard agent 226” generates a GUI “[0036] The dashboard agent 226 may be configured to generate one or more visualizations and/or graphical user interfaces, such as dashboards….” “[0139] In some embodiments, the interface module 528 can function to generate graphical user interface components (e.g., server-side graphical user interface components) that can be rendered as complete graphical user interfaces on the enterprise generative artificial intelligence system 402 and/or other systems. For example, the interface module 528 can function to present an interactive graphical user interface for displaying and receiving information.” Siebel does use a conversational chat 350 as one of its “task-specific fine-tuned LLMs 346” and also includes a “visualization agent 334” and a “visualization tool 628” which generates a “visualization summary 630” which is a graphical summary of the tables and results.]
Regarding Claim 7, Siebel teaches:
7. The computing services environment recited in claim 1,
wherein the conversational chat assistant is one of a plurality of conversational chat assistants accessible via the computing services environment, and [Siebel, Figure 3, the “task-specific fined tuned LLMs 346” are all conversational because they take in natural language as input and provide natural language as output and as shown in Figure 3, each has a particular skill.]
wherein the conversational chat assistant is specific to a client organization of the plurality of client organizations. [Siebel, Figure 4, “[0052] … Enterprise systems 404 can include data flow and management of different processes (e.g., of one or more organizations) and can provide access to systems and users of the enterprise while preventing access from other systems and/or users….” “[0053] The external systems 406 can include applications, datastores, and systems that are external to the enterprise information environment. In one example, the enterprise systems 404 may be a part of an enterprise information environment of an organization that cannot be accessed by users or systems outside that enterprise information environment and/or organization….”]
Regarding Claim 11, Seibel teaches:
11. The computing services environment recited in claim 1, wherein identifying the plurality of actions comprises:
determining a topic identification input prompt that includes the natural language user input and a second one or more natural language instructions executable by the generative language model to identify a topic based on the natural language user input; [Seibel, Figure 1, “input 102” is a topic identification input prompt in natural language: “[0027] … As shown, an initial input 102 is received by the system from either a user (e.g., a natural language input) or another system (e.g., a machine-readable input).” The input 102 teaches both input prompt and the second … instructions because it can be a compound command which invokes several agents and tools: “[0029] … More specifically, the orchestrator may use one or more multimodal models (e.g., language, video, audio, statistical models, etc.), and/or other machine learning models, to interpret the input 102 to select appropriate agents 106 and appropriate tools 108…”]
transmitting the topic identification input prompt to the generative language model for completion; [Seibel, Figure 2, the “input 204” is sent from the “input layer 202” to the “supervisory layer 210” which is a “language model 206.”]
receiving an topic identification input prompt completion from the generative language model; and [Seibel, Figure 2, the generative language model / “language model 206” provides its output (completion) by selecting the appropriate Agents at the Agent Layer 220.]
identifying a topic of a plurality of topics by parsing the intent identification prompt completion, wherein each of the plurality of topics corresponds with a respective topic-based subset of the plurality of actions. [Seibel, Figure 2, the Topics are mapped to the Agents of Seibels and the selected Agents/topics are a subset of the all of the Agents/topics shown in Figure 1.]
There is no express transmitting in Seibel.
Channapattan teaches:
transmitting the topic identification input prompt to the generative language model for completion; [Channapattan, Figures 1, 2, and 3 show a user 185 at a client device 105 which is remote from the on-premise system 115 and therefore the input by the user is transmitted to the main system. See Figures 7 and 8. “[0082] … As an illustrative example, the administrator 285 may input to the natural language UI 210, an English language statement such as “give me all the users that logged in today.” The user query may be transmitted to one or more modules of the natural language interface system 230….” “[0140] The communication module 530 can function to send requests, transmit and receive communications, and/or otherwise provide communication with one or more of the systems, modules, engines, layers, devices, datastores, and/or other components described herein….”]
Siebel and Channapattan pertain to natural language requests and responses and while Seibel does not expressly show the distributed system of the Claim which requires transmission and receiving of the inputs and outputs, this feature is well-known in modern systems that perform the more rigorous processing offsite and it would have been obvious to combine the separate client device and central device of Channapattan with the system of Siebel to achieve that as combining prior art elements according to known methods to yield predictable results or simple substitution of one known element for another to obtain predictable results. See MPEP 2141, KSR, 550 U.S. at 418, 82 USPQ2d at 1396.
Regarding Claim 12, Siebel teaches:
12. The computing services environment recited in claim 11,
wherein the intent identification input prompt identifies a plurality of predetermined actions executable by the computing services environment, [Siebel, Figure 10, 1004 and 1008 where the LLMs and agents and tools are selected. “[0180] In some embodiments, the orchestrator parses the input into different portions (e.g., segments) and routes each portion to a respect agent….” Figure 11, 1106: selecting a first agent from a plurality of different agents.]
wherein the plurality of actions are a subset of the plurality of predetermined actions, [Siebel, the plurality of different Agents and LLMs is predetermined and not indefinite. The list is shown in Figure 1 or 3.]
wherein the plurality of actions are identified in the intent identification prompt completion, and [Siebel, Figure 11, 1106, 1110 each identifies different LLMs for performance of the tasks.]
wherein the plurality of predetermined actions are those corresponding with the identified topic. [Seibel, mapping was changed in the amended Claim to map the topic to Agents of Seibel. The input that identifies the Agent 106 (topic of this Claim) teaches the “topic identification prompt” whose completion identifies the topics/Agents from a plurality of topics /Agents 106-1 …N.]
Claim 13 is a method claim with limitations corresponding to the limitations of Claim 1 and is rejected under similar rationale.
Claim 14 is a method claim with limitations corresponding to the limitations of Claim 2 and is rejected under similar rationale.
Claim 15 is a method claim with limitations corresponding to the limitations of Claim 3 and is rejected under similar rationale.
Claim 16 is a method claim with limitations corresponding to the limitations of Claim 4 and is rejected under similar rationale.
Claim 17 is a method claim with limitations corresponding to the limitations of Claim 5 and is rejected under similar rationale.
Claim 18 is a method claim with limitations corresponding to the limitations of Claim 6 and is rejected under similar rationale.
Claim 19 is a computer program product system claim with limitations corresponding to the limitations of Claim 1 and is rejected under similar rationale.
Claim 20 is a computer program product system claim with limitations corresponding to the limitations of Claims 2 and 3 and is rejected under similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to FARIBA SIRJANI whose telephone number is (571)270-1499. The examiner can normally be reached 9 to 5, M-F.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Pierre Desir can be reached at 571-272-7799. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Fariba Sirjani/
Primary Examiner, Art Unit 2659
Prosecution Insights