DETAILED ACTION
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 .
Status of Claims
This is in reply to communication filed on 01/26/2026.
Claims 1-34 are currently pending and have been examined.
Response to Arguments
In response to Applicant Arguments /Remarks made in an amendment filled on 01/26/2026:
Regarding 35 USC § 101 rejection:
Applicant argument submitted under the title “Claim Rejections Under 35 USC 101” in pages 15-22, that:
“Claim 1, in its entirety, recites:
A data processing method comprising:
receiving at least one input from a user on an electronic user interface;
identifying one or more data objects from the received input to trigger a master controller LLM (large language model)
agent for executing at least one task wherein the one or more data
objects are associated with one or more application developed by a
codeless platform; and
triggering through a processor, one or more micro LLM agent by the master Controller LLM agent for:
selecting a tool from a tool repository by a tool selector agent wherein the tool repository includes one or more tools configured to execute the at least one task; and invoking the selected tool by a tool execution agent wherein the tool execution agent is configured to update the tool repository and act as a process orchestrator for executing the task.
Applicant respectfully submits the Office Action erred in asserting claim 1 is unpatentable under 35 USC 101 … it is important that if the rejection of claim 1 under 35 USC 101 be sustained as allegedly being drawn to an unpatentable abstract idea (by virtue of being drawn to a "certain method of organizing human activity"), it must do so by also showing the alleged method of organizing human behavior is well-established. The Office Action did not do this. Rather, the Office Action merely stated the Examiner finds the claims to simply recite steps of following rules or instructions to execute a task referring to paragraph [0003] of Applicant's specification. However, this paragraph does not teach the method of claim 1 or show that it is directed to a well-established method of organizing activity. Rather, paragraph [0003] merely teaches:
[0003] … far as claim 1 reciting a mental process, Applicant disagrees. The steps recited in
claim 1 cannot be practically performed by the human mind. The human mind, for example, is incapable of triggering a master controller LLM (large language model) agent for executing at least one task wherein the one or more data objects are associated with one or more application developed by a codeless platform nor can the human mind trigger, through a processor, one or more micro LLM agent by the master Controller LLM agent for: selecting a tool from a tool repository by a tool selector agent wherein the tool repository includes one or more tools configured to execute the at least one task; and invoking the selected tool by a tool execution agent wherein the tool execution agent is configured to update the tool repository and act as a process orchestrator for executing the task. For at least all of the above reasons, Applicant respectfully submits claim 1 is patentable under 35 USC 101 … In the event the Examiner disagrees, Applicant respectfully requests the Examiner set for the reasons in the next Office Action as well as make the Office Action non-Final since Applicant has been presented no argument, observations, or evidence regarding claim 8's limitations to rebut”.
Applicant's arguments have been fully considered but they are not persuasive.
In response, the examiner respectfully disagrees and emphasizes none of the receiving, identifying, triggering, selecting, invoking, updating steps, whether taken individually or collectively, have not been shown to affect any form of technical change or improvement whatsoever, and are abstract idea. Applicant's claims have not been shown to modify, reconfigure, manipulate, or transform the computer, computer software, or any technical elements in any discernible manner, much less yield an improvement thereto. There is simply no showing of implementing any of the claim steps, individually or in combination, amounts to a technological improvement, nor the alleged “solve the technical problems associated with conventional computer systems” suggested by Applicant. Although Applicant asserts that “Large language models are useful, the training of these models for a data flowing in an enterprise application built through a codeless platform is cumbersome. Considering the specific use cases prevalent in supply chain and procurement domain, the training of LLM on such dataset is very difficult. Moreover, the nature of data itself is varied and the limitation of existing LLM to process text data, makes it impossible to processing of data generated in the supply chain domain in other formats” the examiner first notes that managing procurement and supply chain is not reasonably understood as a technology, but instead involves organizing of human activity.
The data collection, recognition, and storage concept described in the claim is similar to the data collection and management concepts that were held to be abstract ideas in Content Extraction, TLI Communications, and Electric Power Group. Although the claim enumerates the type of information that is acquired, stored and analyzed, the Federal Circuit has explained in Electric Power Group and Digitech that the mere selection and manipulation of particular information by itself does not make an abstract concept any less abstract. Further, the claim is not made any less abstract by the invocation of a programmed computer. Unlike Enfish, where the claims were focused on a specific improvement in how the computer functioned, the claim here merely uses the computer as a tool to perform the abstract concepts.
Furthermore, the recited (an electronic user interface; a master controller LLM (large language model) agent; are associated with one or more application developed by a codeless platform; triggering through a processor, one or more micro LLM agent by the master Controller LLM agent; from a tool repository by a tool selector agent; by a tool execution agent), this recitation to the generic computer technology that is being used as a tool to execute the steps that define the abstract idea do not provide for integration at the 2nd prong and do not provide for significantly more at step 2B.
In addition, the second step (Step 2B) in the analysis requires to determine whether the claims do significantly more than simply describe that abstract method. Mayo, 132 S.Ct. at 1297. The examiner should examine the limitations of the claims to determine whether the claims contain an “inventive concept” to “transform” the claimed abstract idea into patent-eligible subject matter. Alice, 134 S.Ct. at 2357 (quoting Mayo, 132 S.Ct. at 1294, 1298). The transformation of an abstract idea into patent-eligible subject matter “requires more than simply stat[ing] the [abstract idea] while adding the words ‘apply it.’ ” Id. (quoting Mayo, 132 S.Ct. at 1294) (alterations in original). “A claim that recites an abstract idea must include ‘additional features' to ensure ‘that the [claim] is more than a drafting effort designed to monopolize the [abstract idea].’ ” Id. (quoting Mayo, 132 S.Ct. at 1297) (alterations in original). Those “additional features” must be more than “well-understood, routine, conventional activity.” Mayo, 132 S.Ct. at 1298.
As for Step 2B requires an evaluation of whether the claim recites additional elements that amount to an inventive concept (aka “significantly more”) than the recited judicial exception (e.g. well-understood, routine, conventional activity). The “well-understood, routine, conventional activity” recited by the applicant is not the only consideration that the examiner should consider. The procedure for evaluating Step 2B remains the same as in prior guidance:
First, the examiner identifies whether there are any additional elements recited in the claim beyond the judicial exception(s), and
Second, the examiner evaluates those additional elements individually and in combination to determine whether they amount to significantly more, using the considerations.
None of these individual steps, viewed “both individually and ‘as an ordered combination,’” transform the nature of the claim into patent-eligible subject matter. See Alice, 134 S.Ct. at 2355 (quoting Mayo, 132 S.Ct. at 1297, 1298). The majority of those steps comprise the abstract concept of managing procurement and supply chain, while the additional limitations utilized by the claimed invention to execute the abstract idea is mere use of the computer connected through the Internet, which does not transform an otherwise abstract idea into patent-eligible subject matter. Instead, the claimed sequence of steps comprises only “conventional steps, specified at a high level of generality,” which is insufficient to supply an “inventive concept.” Id. at 2357 (quoting Mayo, 132 S.Ct. at 1294, 1297, 1300). In other words, Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f), and accordingly do not provide for significantly more at step 2B.
Even assuming, for the sake of argument, that the claims amount to an improvement over prior art techniques for managing procurement and supply chain, such an improvement would be considered, at most, an improvement confined within the abstract idea itself, which is not enough to confer eligibility on the claim. For the reasons above, Applicant’s argument is not persuasive.
Regarding Claim Rejections - 35 USC § 103:
Applicant's arguments have been fully considered but they are not persuasive.
Pei teaches a method that utilizes a framework for transaction categorization personalization. A transaction record is received. a baseline model is selected from a plurality of machine learning models. An account identifier, corresponding to the transaction record using the baseline model, is selected. The account identifier for the transaction record is presented.
Sheikh teaches enabling autonomous agents (AAs) across problem domains. comprises client-agent device (client-AA) to receive service request (SR), generate objective in vector(s) recorded in vector database (VD), send objective to agent-device (AA/micro-AA). Abstract. Sheikh further teaches Large Language Model (LLM) understands the context, identifies key information, and extracts relevant details (Sheikh’s key information and relevant details reads on the claimed data objects) from the service request, [0075], and once the Large Language Model (LLM) processes the service request, it transforms the extracted information into service data represented as the one or more vectors, see [0075], in order to generate protocol specification(PS(s)) for task execution by AA(s) see abstract.
Sheikh further teaches that autonomous agent (further-AA) is configured to implement the at least one protocol specification to execute each task associated with the objective and thereby automatically execute the service request, which reads on the claimed codeless features.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Pei to include the features, as taught by Sheikh, where this would be performed in order to reduce the need for direct human intervention. See Sheikh [0002].
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-34 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception without significantly more.
Step 1:
Claims 1-14 recite a method, which is directed to a process.
Claims 15-30 recite a system, which is directed to a machine.
Claims 31-32 recite a computer program product comprising a non-transitory computer readable storage medium that causes a processor, which is directed to a manufacture.
Claim 33-34 recite a method, which is directed to a process.
Therefore, each claim falls within one of the four statutory categories.
Step 2A, Prong 1 (Is a judicial exception recited?):
The independent claims 1, 15, 31 and 33 recite the abstract idea, which is described by the steps of:
receiving at least one input from a user;
identifying one or more data objects from the received input to trigger for executing at least one task wherein the one or more data objects; and
selecting a tool includes one or more tools configured to execute the at least one task; and
invoking the selected tool update the tools and act as a process orchestrator for executing the task;
These claims recite a certain method of organizing human activity. The claims recite to a certain method of organizing human activity as the above abstract idea limitations are directed to managing personal behavior or relationships or interactions between people. The examiner finds the claims to simply recites a certain method of human organization to follow a certain instruction/ rules to receive data, recognize objects in the received data to choose a specific process/workflow to execute a task, and update information. The Examiner additionally finds the claims to be similar to an example the courts have identified as being a certain method of organizing human activity:
i. filtering content, BASCOM Global Internet v. AT&T Mobility, LLC, 827 F.3d 1341, 1345-46, 119 USPQ2d 1236, 1239 (Fed. Cir. 2016) (finding that filtering content was an abstract idea under step 2A, but reversing an invalidity judgment of ineligibility due to an inadequate step 2B analysis);
ii. considering historical usage information while inputting data, BSG Tech. LLC v. Buyseasons, Inc., 899 F.3d 1281, 1286, 127 USPQ2d 1688, 1691 (Fed. Cir. 2018);
The data collection, recognition, and storage concept described in the claim is similar to the data collection and management concepts that were held to be abstract ideas in Content Extraction, TLI Communications, and Electric Power Group. Although the claim enumerates the type of information that is acquired, stored and analyzed, the Federal Circuit has explained in Electric Power Group and Digitech that the mere selection and manipulation of particular information by itself does not make an abstract concept any less abstract. Further, the claim is not made any less abstract by the invocation of a programmed computer.
The claims recite a mental process. Before computers one could mentally or a human using paper and pen to execute a task. the "mental processes" abstract idea grouping is defined as concepts performed in the human mind, and examples of mental processes include observations, evaluations, judgments, and opinions. The claims are merely directed to a method of receiving data; identifying elements in the data; selecting a tool; update store data; invoke a task . The Examiner find the recited claims to be similar to a claim to "collecting information, analyzing it, and displaying certain results of the collection and analysis," where the data analysis steps are recited at a high level of generality such that they could practically be performed in the human mind, Electric Power Group v. Alstom, S.A., 830 F.3d 1350, 1353-54, 119 USPQ2d 1739, 1741-42 (Fed. Cir. 2016), which the courts have also found to recite a mental process.
Step 2A, Prong 2 (Is the exception integrated into a practical application?):
This judicial exception is not integrated into a practical application because the claims satisfy the following criteria, which indicate that the claims do not integrate the abstract idea into practical application:
The claimed additional limitations are:
Claim 1: on an electronic user interface; a master controller LLM (large language model) agent; are associated with one or more application developed by a codeless platform; triggering through a processor, one or more micro LLM agent by the master Controller LLM agent; from a tool repository by a tool selector agent; by a tool execution agent,
Claim 15: one or more processors; and one or more memory devices including instructions that are executable by the one or more processor for causing the processor; on the electronic user interface; a master controller large language model (LLM) agent; are associated with one or more application developed by a codeless platform; trigger through the one or more processors, one or more micro LLM agent by the master Controller LLM agent; a tool repository by a tool selector agent; by a tool execution agent
Claim 31: on the electronic user interface; a master controller large language model (LLM) agent; are associated with one or more application developed by a codeless platform; trigger through the one or more processors, one or more micro LLM agent by the master Controller LLM agent; a tool repository by a tool selector agent; by a tool execution agent
Claim 33: on the electronic user interface; a micro LLM (large language model) agent; a bot configured to map the intent with the micro LLM agent; from a tool repository by a tool selector agent; by a tool execution agent
The additional limitations are directed to using a generic computer to process information and perform the abstract idea. Therefore, the limitations merely amount to adding the words “apply it” (or an equivalent) to the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea, as discussed in MPEP 2106.05(f).
Step 2B (Does the claim recite additional elements that amount to significantly more that the judicial exception?):
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
As for Step 2B analysis, knowing the consideration is overlapping with Step 2A, Prong 2. The Step 2B considerations have already been substantially addressed under Step 2A Prong 2, see Step 2A Prong 2 analysis above. As discussed above, the additional imitations amount to adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea, as discussed in MPEP 2106.05(f).
In addition, the dependent claims recite:
Step 2A, Prong 1 (Is a judicial exception recited?):
Dependent claims 2-14, 16-30, 32 and 34 recitations further narrowing the abstract idea recited in the independent claims 1, 15, 31 and 33 and therefore directed towards the same abstract idea.
Step 2A, Prong 2 and Step 2B:
The dependent claims 2-14, 16-30, 32 and 34 further narrow the abstract idea recited in the independent claims 1, 15, 31 and 33 and are therefore directed towards the same abstract idea.
The dependent claims recite the following additional limitations, such as; the one or more micro LLM agent, at least one application of the one or more application developed by the codeless platform, micro LLM agent, master controller LLM agent, the one or more application developed by the codeless platform and interact with the one or more micro LLM agents, training dataset, codeless platform, the micro LLM agent are distributed across a plurality of Graphics processing units (GPU) for parallel training of the micro LLM agent, supply chain management (SCM) application, layer(s), a data network, one or more supply chain application dataset, one or more data connectors of the graphical data structure, an AI engine coupled to the processor
However, the examiner finds each of these additional elements to be directed to merely “apply it” or applying a generic technology to perform the recited abstract idea of executing a task, the recitation to the generic computer technology that is being used as a tool to execute the steps that define the abstract idea do not provide for integration at the 2nd prong and do not provide for significantly more at step 2B.
Therefore, the limitations on the invention of claims 1-34, when viewed individually and in ordered combination are directed to in-eligible subject matter.
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.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-7, 9, 11, 13-20, 22, 24-27, 29 and 31-34 are rejected under 35 U.S.C 103 as being unpatentable over Pei et al. (US 20220318925 A1, hereinafter “Pei”) in view of Sheikh et al. (US 20230368284 A1, hereinafter “Sheikh”).
Regarding claims 1 and 15. Pei discloses a data processing method comprising:
receiving at least one input from a user on an electronic user interface; (Pei, [0045]; “Turning to FIG. 2A, at Step 202, a transaction record is received. The transaction record may be received by a server application response to an entity logging into the system and requesting to see a list of transactions. The list of transactions may include the transaction record, which may be displayed by the entity device accessing the system”) for:
selecting a tool from a tool repository (Pei, [0028]; “The repository (105) … may include … the machine learning model data (108), and the training data (109)”) by a tool selector agent (Pei, [0046]; “At Step 204, a baseline model is selected from a group of machine learning models … The baseline model may be selected as a general model when one or more entity usage thresholds have not been met … The selection may be performed by the model selector”) wherein the tool repository includes one or more tools configured to execute the at least one task; (Pei, [0048]; “At Step 206, an account identifier corresponding to the transaction record is selected using the baseline model. The account identifier that is selected identifies the account that is suggested to be assigned to the transaction of the transaction record”, the examiner interprets the selection of an account identifier to satisfy a task being executing) and
invoking the selected tool by a tool execution agent wherein the tool execution agent (Pei, [0053]; “The baseline model (120) is a machine learning model and is either the general model (122) or one of the custom models (123) … The input to the baseline model (120), i.e., the input to either the general model (122) or the custom models (123), is a transaction record (e.g., the transaction record A (125)) and the output is an account identifier (e.g., one of the account identifiers A (136) and B (148))”) is configured to
update the tool repository (Pei, [0088]; “The machine learning model (183) and the update function (186) may be stored in the machine learning model data (108) (of FIG. 1A) of the repository (105) (of FIG. 1A)”) and
act as a process orchestrator for executing the task. (Pei, [0048] At Step 206, an account identifier corresponding to the transaction record is selected using the baseline model … At Step 208, the account identifier for the transaction record is presented … After receiving the account identifier, the entity device may display the account identifier with the transaction record in an entity application”)
Pei substantially discloses the claimed invention; however, Pei fails to explicitly disclose the “identifying one or more data objects from the received input to trigger a master controller LLM (large language model) agent for executing at least one task wherein the one or more data objects are associated with one or more application developed by a codeless platform; and triggering through a processor, one or more micro LLM agent by the master Controller LLM agent”. However, Sheikh teaches
identifying one or more data objects from the received input ([0075]; “The Large Language Model (LLM) understands the context, identifies key information, and extracts relevant details from the service request”) to trigger a master controller LLM (large language model) agent for executing at least one task ([0075]; “once the Large Language Model (LLM) processes the service request, it transforms the extracted information into service data represented as the one or more vectors”) wherein the one or more data objects are associated with one or more application developed by a codeless platform; and (Sheikh, [0131]; “At 506, one or more queries are sent from a context-builder software module of the agent-device (AA or micro-AA) to a machine learning model agent (ML-Model AA) and/or the vector database is accessed to retrieve one or more tasks associated with previous queries in order to obtain a plurality of tasks associated with the objective”)
triggering through a processor, one or more micro LLM agent by the master Controller LLM agent (Sheikh, [0131]; “At 508, there is further implemented, by the context-builder software module, a step of interactively communicating with the machine learning model agent (ML-Model AA) for obtaining an order in which each task is to be executed. At 510, there is further implemented, by the context-builder software module, a step of interactively communicating with the machine learning model agent (ML-Model AA) for: obtaining a list including a plurality of autonomous agents (AAs or micro-AAs)”)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Pei to include identifying one or more data objects from the received input to trigger a master controller LLM (large language model) agent for executing at least one task wherein the one or more data objects are associated with one or more application developed by a codeless platform; and triggering through a processor, one or more micro LLM agent by the master Controller LLM agent, as taught by Sheikh, where this would be performed in order to reduce the need for direct human intervention. See Sheikh [0002].
Regarding claims 2 and 16. The combination of Pei in view of Sheikh discloses the method of claim 1, wherein each of the one or more micro LLM agent is trained on a historical dataset associated with at least one application of the one or more application developed by the codeless platform. (Pei, [0035]; “The training data (109) has pairs of transaction records (e.g., historical transaction records of the entities using the system) and account identifiers that had been assigned to the transaction”)
Regarding claims 3 and 17. The combination of Pei in view of Sheikh discloses the method of claim 2, wherein the at least one application includes a supply chain management application and the at least one task includes a supply chain management application task such as contract management, Purchase order, invoice management, Spend analysis, Sourcing, inventory management, demand planning, quality management, supply planning, should cost modeling, transportation management, warehouse management, forecasting, vendor management, risk assessment management and project management. (Pei, [0024]; “For example, the user may access the system (100) to assign accounts (account identifiers) to transactions (transaction records) and receive suggestions (e.g., the account identifier D (162) of FIG. 1B) from the system (100) for which account to assign to a transaction (e.g., the transaction corresponding to the transaction record A (125) of FIG. 1B)”)
Regarding claims 4 and 18. The combination of Pei in view of Sheikh discloses the method of claim 1, wherein the tool selector agent is a micro LLM agent configured to select and chain tools that can execute the at least one task. (Pei, [0043]; “The model selector (121) identifies which machine learning model is to be used as a baseline model (e.g., between the general model (122) or the custom models (124)). The account selector (160) is configured to select the account from the output of the comparison model (152) and baseline model as selected by the model selector (121)”)
Regarding claims 5 and 19. The combination of Pei in view of Sheikh discloses the method of claim 1, wherein the process orchestrator performs sequence management, execution state control and conversation state management. (Pei, [0059]; “The account embedding model encodes the account information to generate an account vector. The account embedding model is a pre-trained word to vector model that converts an account name to an account vector, which is the output of the account embedding model. For example, the account embedding model may be a word2vec model. Alternative models include GloVe developed by Stanford, fastText developed by Facebook, Inc., amongst other encoding models”)
Regarding claims 6 and 20. The combination of Pei in view of Sheikh discloses the method of claim 1, wherein the master controller LLM agent is configured to adapt to real time changing characteristics of the one or more application developed by the codeless platform and interact with the one or more micro LLM agents to execute the at least one task. (Pei, [0068]; “The adapter model (143) generates the adapter model output (145) from the transaction vector A (128), which may be used instead of the raw data from the transaction record A (125). The raw data that forms transaction records (e.g., the payee name, amount, date description, etc.) may be too sparse and have too few examples to properly train the machine learning models that are in the adapter model (143)”)
Regarding claim 7. The combination of Pei in view of Sheikh discloses the method of claim 2, wherein the one or more micro LLM agent and the master controller LLM agent is trained by:
collecting, storing and pre-processing a plurality of historical data as a training data wherein the historical data is stored in a SCM historical database;
cleansing the training dataset by converting the historical data, removing unwanted text from the historical data and tokenizing the training dataset into sequences of tokens that form the training dataset; and (Pei, [0058]; “The transaction model (126) is a machine learning model that may include multiple neural network layers and inputs that generate the transaction vector A (128) from features extracted from the transaction record A (125). The features extracted from the transaction record A (125) may include the raw features from the different fields of the transaction record A (125) and values derived from the raw features”)
configuring a neural network based on the training dataset wherein the micro LLM agent is trained with supervised and unsupervised learning by presenting a sequence of text to the LLM agent for training the agent to predict next text in the sequence wherein the LLM agent adjusts its weight based on a difference between its prediction and actual text. (Pei, [0035]; “[0035] The training data (109) is the data used to train the machine learning models of the system (100). The training data (109) has pairs of transaction records (e.g., historical transaction records of the entities using the system) and account identifiers that had been assigned to the transaction. The training data (110) may also include the intermediate data generated to train and update the machine learning models of the system. The training data (110) may include the training inputs and expected outputs shown in FIG. 1C. Each model used by the system (100) may have a particular set of data in the training data (109) that includes the specific type of input data for the model and the specific type of expected output for the model. The training data (109) may include the training inputs and expected outputs shown in FIG. 1C”)
Regarding claims 9 and 29. The combination of Pei in view of Sheikh discloses the method of claim 7, wherein the micro LLM agent is configured to be
Pei substantially discloses the claimed invention; however, Pei fails to explicitly disclose the “trained in a distributed structure with artificial intelligence controllers wherein different parts of the micro LLM agent are distributed across a plurality of Graphics processing units (GPU) for parallel training of the micro LLM agent”. However, Sheikh teaches
trained in a distributed structure with artificial intelligence controllers wherein different parts of the micro LLM agent are distributed across a plurality of Graphics processing units (GPU) for parallel training of the micro LLM agent. (Sheikh, [0075]; “the software application utilizes the Large Language Model (LLM) to process the service request. Notably, the Large Language Model (LLM) is an artificial intelligence model trained on a vast amount of textual data, enabling it to understand and generate human-like language responses”)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Pei to include trained in a distributed structure with artificial intelligence controllers wherein different parts of the micro LLM agent are distributed across a plurality of Graphics processing units (GPU) for parallel training of the micro LLM agent, as taught by Sheikh, where this would be performed in order to reduce the need for direct human intervention. See Sheikh [0002].
Regarding claim 11. The combination of Pei in view of Sheikh discloses the method of claim 7, further comprises the step of evaluating performance of the one or more micro LLM agent based on a testing dataset wherein the one or more micro LLM agent is finetuned by adjusting one or more hyperparameters, chaining model architecture or training the micro LLM agent on additional training dataset to improve performance. (Pei, [0021]; “The process (200) of FIG. 2A uses machine learning models to generate a baseline account suggestion … determinations may be performed by performing a test, such as checking a data value to test whether the value is consistent with the tested condition”)
Regarding claims 13 and 22. The combination of Pei in view of Sheikh discloses the method of claim 12, wherein the master controller LLM agent is trained on one or more workflows of the one or more SCM application developed by the codeless platform wherein the master controller LLM agent is configured to generate codes for supplementing operations executed by the one or more SCM application. (Pei, [0035] The training data (109) is the data used to train the machine learning models of the system (100). The training data (109) has pairs of transaction records (e.g., historical transaction records of the entities using the system) and account identifiers that had been assigned to the transaction. The training data (110) may also include the intermediate data generated to train and update the machine learning models of the system. The training data (110) may include the training inputs and expected outputs shown in FIG. 1C. Each model used by the system (100) may have a particular set of data in the training data (109) that includes the specific type of input data for the model and the specific type of expected output for the model. The training data (109) may include the training inputs and expected outputs shown in FIG. 1C.”)
Regarding claims 14 and 27. The combination of Pei in view of Sheikh discloses the method of claim 13, wherein the electronic user interface includes an input component configured to receive the input, wherein (Pei, [0030]; “A transaction record is a text string describing a financial transaction”)
Pei substantially discloses the claimed invention; however, Pei fails to explicitly disclose the “the input component is a chatbot configured to receive a text, image or voice input wherein the image or voice input is converted to text by one or more processors for enabling the master controller LLM agent to identify the intent of the user and the at least one task to be executed”. However, Sheikh teaches
the input component is a chatbot configured to receive a text, image or voice input wherein the image or voice input is converted to text by one or more processors for enabling the master controller LLM agent to identify the intent of the user and the at least one task to be executed (Sheikh, [0060]; “software application within the client-agent device could be a chatbot that interacts with users through natural language processing. Optionally, the software application could be a task management application that allows users to create, assign, and track tasks. Optionally, the software application could function as a virtual assistant application that assists users with various tasks and provides personalized recommendations or services. Optionally, the software application could serve as an intelligent shopping assistant, and so forth”)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Pei to include the input component is a chatbot configured to receive a text, image or voice input wherein the image or voice input is converted to text by one or more processors for enabling the master controller LLM agent to identify the intent of the user and the at least one task to be executed, as taught by Sheikh, where this would be performed in order to reduce the need for direct human intervention. See Sheikh [0002].
Regarding claim 24. The combination of Pei in view of Sheikh discloses the system of claim 23, further comprises one or more large graph models (LGM) configured to interact with the one or more Micro LLM agent and Master controller LLM agent based on alignment of representation basis of graphs and text through paired data enabling interaction through natural language, or by transforming graph structures to text representations including adjacent list, edge list and inserting into LLM agents as prompts, or by aligning behavior of one or more graph models with one or more graph task scripts. (Pei, [0069]; “the adapter model (143) may include multiple logistic regression models that are mapped to the accounts of a chart of accounts for an entity of the system (100) (of FIG. 1A). The logistic regression models may be mapped to the accounts (and corresponding account identifiers) in a one to one correspondence. Each entity may have a set of logistic regression models. The logistic regression models generate logistic regression model outputs”)
Regarding claim 25. The combination of Pei in view of Sheikh discloses the system of claim 15, further comprises: at least one storage layer configured for storing information including memory objects, selected tools information, state of execution, and error messages thereby tracking the information and data objects generated during the process orchestration. (Pei, [0086]; “When a machine learning model is trained in conjunction with another machine learning model, both machine learning models may be updated based on the error between the expected output and the training output generated by the models”)
Regarding claim 26. The combination of Pei in view of Sheikh discloses the system of claim 15, wherein the one or more processors includes a request processor configured for routing the input to tool selector agent and tool executor agents. (Pei, [0116]; “At Step 418 a request is sent from the client application (401) to the server application (402), which initiates the session (416). The request may be for a web page that provides a list of transactions, which allows the entity to assign accounts to transactions and provides suggestions for which account to assign to a transaction”)
Regarding claim 31. Pei discloses a computer program product comprising a non-transitory computer readable storage medium that causes a processor (Pei, [0139]; “Software instructions in the form of computer readable program code to perform embodiments of the invention may be stored, in whole or in part, temporarily or permanently, on a non-transitory computer readable medium”) to:
receive at least one input from a user on the electronic user interface; (Pei, [0045]; “Turning to FIG. 2A, at Step 202, a transaction record is received. The transaction record may be received by a server application response to an entity logging into the system and requesting to see a list of transactions. The list of transactions may include the transaction record, which may be displayed by the entity device accessing the system”) to:
select a tool from a tool repository (Pei, [0028]; “The repository (105) … may include … the machine learning model data (108), and the training data (109)”) by a tool selector agent (Pei, [0046]; “At Step 204, a baseline model is selected from a group of machine learning models … The baseline model may be selected as a general model when one or more entity usage thresholds have not been met … The selection may be performed by the model selector”) wherein the tool repository includes one or more tools configured to execute the at least one task; (Pei, [0048]; “At Step 206, an account identifier corresponding to the transaction record is selected using the baseline model. The account identifier that is selected identifies the account that is suggested to be assigned to the transaction of the transaction record”, the examiner interprets the selection of an account identifier to satisfy a task being executing) and
invoke the selected tool by a tool execution agent wherein the tool execution agent (Pei, [0053]; “[0053] The baseline model (120) is a machine learning model and is either the general model (122) or one of the custom models (123) … The input to the baseline model (120), i.e., the input to either the general model (122) or the custom models (123), is a transaction record (e.g., the transaction record A (125)) and the output is an account identifier (e.g., one of the account identifiers A (136) and B (148))”) is configured to update the tool repository (Pei, [0088]; “The machine learning model (183) and the update function (186) may be stored in the machine learning model data (108) (of FIG. 1A) of the repository (105) (of FIG. 1A)”) and act as a conversation orchestrator with the user for executing the task. (Pei, [0048] At Step 206, an account identifier corresponding to the transaction record is selected using the baseline model … At Step 208, the account identifier for the transaction record is presented … After receiving the account identifier, the entity device may display the account identifier with the transaction record in an entity application”)
Pei substantially discloses the claimed invention; however, Pei fails to explicitly disclose the “identify one or more data objects from the received input to trigger a master controller large language model (LLM) agent for executing at least one task wherein the one or more data objects are associated with one or more application developed by a codeless platform; and trigger through the one or more processors, one or more micro LLM agent by the master Controller LLM agent”. However, Sheikh teaches
identify one or more data objects from the received input ([0075]; “The Large Language Model (LLM) understands the context, identifies key information, and extracts relevant details from the service request”) to trigger a master controller large language model (LLM) agent for executing at least one task ([0075]; “once the Large Language Model (LLM) processes the service request, it transforms the extracted information into service data represented as the one or more vectors”) wherein the one or more data objects are associated with one or more application developed by a codeless platform; and (Sheikh, [0131]; “At 506, one or more queries are sent from a context-builder software module of the agent-device (AA or micro-AA) to a machine learning model agent (ML-Model AA) and/or the vector database is accessed to retrieve one or more tasks associated with previous queries in order to obtain a plurality of tasks associated with the objective”)
trigger through the one or more processors, one or more micro LLM agent by the master Controller LLM agent (Sheikh, [0131]; “At 508, there is further implemented, by the context-builder software module, a step of interactively communicating with the machine learning model agent (ML-Model AA) for obtaining an order in which each task is to be executed. At 510, there is further implemented, by the context-builder software module, a step of interactively communicating with the machine learning model agent (ML-Model AA) for: obtaining a list including a plurality of autonomous agents (AAs or micro-AAs)”)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Pei to include identify one or more data objects from the received input to trigger a master controller large language model (LLM) agent for executing at least one task wherein the one or more data objects are associated with one or more application developed by a codeless platform; and trigger through the one or more processors, one or more micro LLM agent by the master Controller LLM agent, as taught by Sheikh, where this would be performed in order to reduce the need for direct human intervention. See Sheikh [0002].
Regarding claim 32. The combination of Pei in view of Sheikh discloses the non-transitory computer program product of claim 31, wherein the method is performed in a cloud or cloud-based computing environment. (Pei, [0038]; “the server (101) may be one of a set of virtual machines hosted by a cloud services provider to deploy the training application (103) and the server application (102)”)
Regarding claim 33. Pei discloses a method comprising:
receiving at least one input from a user on the electronic user interface; (Pei, [0045]; “Turning to FIG. 2A, at Step 202, a transaction record is received. The transaction record may be received by a server application response to an entity logging into the system and requesting to see a list of transactions. The list of transactions may include the transaction record, which may be displayed by the entity device accessing the system”)
selecting a tool from a tool repository (Pei, [0028]; “The repository (105) … may include … the machine learning model data (108), and the training data (109)”) by a tool selector agent (Pei, [0046]; “At Step 204, a baseline model is selected from a group of machine learning models … The baseline model may be selected as a general model when one or more entity usage thresholds have not been met … The selection may be performed by the model selector”) wherein the tool repository includes one or more tools configured to execute the at least one task; (Pei, [0048]; “At Step 206, an account identifier corresponding to the transaction record is selected using the baseline model. The account identifier that is selected identifies the account that is suggested to be assigned to the transaction of the transaction record”, the examiner interprets the selection of an account identifier to satisfy a task being executing”) and
invoking the selected tool by a tool execution agent wherein the tool execution agent (Pei, [0053]; “[0053] The baseline model (120) is a machine learning model and is either the general model (122) or one of the custom models (123) … The input to the baseline model (120), i.e., the input to either the general model (122) or the custom models (123), is a transaction record (e.g., the transaction record A (125)) and the output is an account identifier (e.g., one of the account identifiers A (136) and B (148))”) is configured to update the tool repository (Pei, [0088]; “The machine learning model (183) and the update function (186) may be stored in the machine learning model data (108) (of FIG. 1A) of the repository (105) (of FIG. 1A)”) and act as a process orchestrator for executing the task. (Pei, [0048] At Step 206, an account identifier corresponding to the transaction record is selected using the baseline model … At Step 208, the account identifier for the transaction record is presented … After receiving the account identifier, the entity device may display the account identifier with the transaction record in an entity application”)
Pei substantially discloses the claimed invention; however, Pei fails to explicitly disclose the “identifying one or more data objects from the received input to trigger a micro LLM (large language model) agent for executing at least one task wherein an intent of the user is determined based on the identified data objects by a bot configured to map the intent with the micro LLM agent”. However, Sheikh teaches
identifying one or more data objects from the received input ([0075]; “The Large Language Model (LLM) understands the context, identifies key information, and extracts relevant details from the service request”) to trigger a micro LLM (large language model) agent ([0075]; “once the Large Language Model (LLM) processes the service request, it transforms the extracted information into service data represented as the one or more vectors”) for executing at least one task (Sheikh, [0131]; “At 506, one or more queries are sent from a context-builder software module of the agent-device (AA or micro-AA) to a machine learning model agent (ML-Model AA) and/or the vector database is accessed to retrieve one or more tasks associated with previous queries in order to obtain a plurality of tasks associated with the objective”) wherein an intent of the user is determined based on the identified data objects (Sheikh, [0067] The term “objective” as used herein refers to a desired outcome or goal that the client-agent device (client-AA) aims to achieve based on the service request received therethrough. Optionally, the objective defines the purpose or intent behind the service request”) by a bot configured to map the intent with the micro LLM agent; (Sheikh, [0094]; “the context-builder software module ensures that there is a proper mapping between tasks and the autonomous agents, ensuring that each task is assigned to at least one autonomous agent associated with the objective”)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Pei to include identifying one or more data objects from the received input to trigger a micro LLM (large language model) agent for executing at least one task wherein an intent of the user is determined based on the identified data objects by a bot configured to map the intent with the micro LLM agent, as taught by Sheikh, where this would be performed in order to reduce the need for direct human intervention. See Sheikh [0002].
Regarding claim 34. The combination of Pei in view of Sheikh discloses the method of claim 33, wherein the one or more application includes a supply chain management application and the at least one task includes a supply chain management application task such as contract management, Purchase order, invoice management, Spend analysis, Sourcing, inventory management, demand planning, quality management, supply planning, should cost modeling, transportation management, warehouse management, forecasting, vendor management, risk assessment management and project management. (Pei, [0024]; “For example, the user may access the system (100) to assign accounts (account identifiers) to transactions (transaction records) and receive suggestions (e.g., the account identifier D (162) of FIG. 1B) from the system (100) for which account to assign to a transaction (e.g., the transaction corresponding to the transaction record A (125) of FIG. 1B)”)
Claims 10 and 30 are rejected under 35 U.S.C 103 as being unpatentable over Pei in view of Sheikh in view of Raumann et al. (US 20220197714 A1, hereinafter “Raumann”).
Regarding claims 10 and 30. The combination of Pei in view of Sheikh discloses the method of claim 9, wherein
The combination of Pei in view of Sheikh substantially discloses the claimed invention; however, the combination of Pei in view of Sheikh fails to explicitly disclose the “the parallel training of the micro LLM agent includes data parallelism, sequence parallelism, pipeline parallelism and tensor parallelism”. However, Raumann teaches
the parallel training of the micro LLM agent includes data parallelism (Raumann, [0298]; “Parallelism can occur in multiple forms such as fine-grained and coarse-grained pipeline parallelism, data parallelism, and task parallelism”), sequence parallelism, pipeline parallelism and tensor parallelism. (Raumann, [0162]; “FIG. 17A is a message sequence chart 1700A illustrating one implementation of asynchronous tensor streaming in which a next tensor is buffered while a reconfigurable processor is processing a current tensor … the series of data units 1712 includes a sequence of tensors 1 to N”)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Pei to include the parallel training of the micro LLM agent includes data parallelism, sequence parallelism, pipeline parallelism and tensor parallelism, as taught by Raumann, where this would be performed in order to train larger and more powerful neural networks effectively. See Raumann [0044].
Distinguished Over Prior Art
Claims 8, 12, 21, 23 and 28 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
1. THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
2. Any inquiry concerning this communication or earlier communications from the examiner should be directed to AVIA SALMAN whose telephone number is (313)446-4901. The examiner can normally be reached Monday thru Friday; 9:00 AM to 5:00 PM EST.
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, FAHD OBEID can be reached at (571) 270-3324. 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.
/AVIA SALMAN/Primary Patent Examiner, Art Unit 3627
Prosecution Insights