DETAILED ACTION
The following is a Final Office action. In response to Examiner’s communication of 4/16/2026, Applicant, on 5/27/2026, amended claims 1, 8, and 15. Claims 1-3, 5-10, 12-17, 19, 20, and 24-26 are now pending and have been rejected as indicated below.
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 .
Information Disclosure Statement
No Information Disclosure Statement has yet been filed. As such, No Information Disclosure Statement has been considered.
Response to Amendment
Applicant’s amendments are acknowledged.
The 35 USC 101 rejection of claims 1-3, 5-10, 12-17, 19, 20, and 24-26 in regard to abstract ideas has been withdrawn in light of Applicant’s amendments and explanations.
New 35 USC 103 rejections of claims 1-3, 5-10, 12-17, 19, 20, and 24-26 have been applied in light of Applicant’s amendments and explanations.
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 of this title, 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 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 1-3, 5-10, 12-17, 19, 20 and 24-26 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Number 12223456 to Manohar et al. (hereafter referred to as Manohar) in view of U.S. Patent Application Publication Number 2025/0272652 to Makhija et al. (hereafter referred to as Makhija) in further view of U.S. Patent Application Publication Number 2025/0135350 to Gromenko et al. (hereafter referred to as Gromenko) and in even further view of U.S. Patent Application Publication Number 2024/0402692 to Sheppard-Guzelaydin et al. (hereafter referred to as Sheppard-Guzelaydin).
As per claim 1, Manohar teaches:
A system comprising: a memory that stores instructions; and one or more processors coupled to the memory and configured to execute the instructions to perform operations comprising: (Col. 16 lines 30-61 teach the one or more hardware processors 110, as used herein, means any type of computational circuit, such as, but not limited to, a microprocessor unit, microcontroller, complex instruction set computing exceptionally long processor unit, reduced instruction set computing microprocessor unit, very long instruction word microprocessor unit, explicitly parallel instruction computing microprocessor unit, graphics processing unit, digital signal processing unit, or any other type of processing circuit. The one or more hardware processors 110 may also include embedded controllers, such as generic or programmable logic devices or arrays, application-specific integrated circuits, single-chip computers, and the like. The memory 112 may be a non-transitory volatile memory and a non-volatile memory. The memory 112 may be coupled to communicate with the one or more hardware processors 110, such as being a computer-readable storage medium. The one or more hardware processors 110 may execute machine-readable instructions and/or source code stored in the memory 112. A variety of machine-readable instructions may be stored in and accessed from the memory 112).
determining a set of formulas to calculate a set of key performance indicators (KPIs) (Col. 15 lines 43-56 teach the first artificial intelligence (AI) model is trained with one or more pre-defined rules and criteria to assess the one or more pre-defined key performance indicators (KPIs) and one or more pre-defined goals defined in at least one of: the service level agreements (SLAs) and the service level objectives (SLOs). The computer-implemented system 102 may be further configured to automatically update the one or more service level objectives (SLOs) and associated metrics corresponding to the one or more services specified in the one or more service level agreements (SLAs), based on deviations of the one or more actual performance levels of the one or more services from the one or more expected performance levels, using the first artificial intelligence (AI) model).
each formula of the set of formulas comprising a set of variables used in the formula and a mathematical relationship between the set of variables and a corresponding KPI (Col. 15 line 57 - Col. 16 line 3 teaches the computer-implemented system 102 may be further configured to generate one or more insights associated with one or more actions, to be applied to one or more corresponding services, to be performed to automatically manage the one or more applications based on the automatic updates on the one or more service level objectives (SLOs) and associated metrics corresponding to the one or more services specified in the one or more service level agreements (SLAs), using the first artificial intelligence (AI) model. The computer-implemented system 102 may be further configured to provide information associated with the one or more actions, to be applied to the one or more corresponding services to automatically manage the one or more applications, to the one or more environments 116. Col. 16 line 62 - Col. 17 line 16 teaches the storage unit 204 may be a cloud storage or a repository such as those shown in FIG. 1. The storage unit 204 may store, but is not limited to, the service level agreements (SLAs), the applications (e.g., the enterprise applications), the service level objectives (SLOs), policy records, measurable performance metrics, the services, performance requirements, policy action(s) for each service, error files, standard SLI record, data compliance, security requirements, encryption requirements, governance requirements, requirements, objectives, constraints, a location, availability requirements, pricing requirements, security requirements associated with the one or more ERP applications, response time, availability, throughput, performance indicators, service name, a current performance level of a performance metric under consideration, an empty actions service name, performance required metric, action to improve or maintain the record, any other data, and combinations thereof. (Examiner asserts that the repository is storing information that is equivalent to mathematical relationships such as constraints and governance requirements. Additionally, these relationships are between sets of variables and the various KPIs as described in Col. 15 lines 43-56)).
generating a first prompt for a large language model (LLM), the first prompt comprising names from the variables used in the set of formulas and, for each of the plurality of data sources, a set of names of key figures in the data source (Col. 29 line 60 - Col. 30 line 5 teaches the multi-prompt method 1004 may be configured to apply few-shot prompting to generate ReAct trajectories with a strong large language model, including GPT-4. This process includes creating thought-action-observation rounds to guide the large language model through task-solving processes. The multi-prompt method 1004 may be further configured to convert Chain of Thought (CoT) prompts into one-round ReAct trajectories where the “thought” is the reasoning step, and the “action” is the answer to the question. The multi-prompt method 1004 may be further configured to provide Reflexion trajectories by prompting for reflections at specific rounds, allowing the large language model to adjust its strategy in solving the one or more tasks. Col. 31 lines 10-22 teach the fine-tuned language model good at providing the information associated with the SLA, SLO and associated metrics, is static due to the nature of data is inputted. In an embodiment, the fine-tuned language model may reason to a certain extent with such static data. The static model is fine-tuned with the relevant information is an initial step of the process. For generating the one or more actions to ensure the SLO and associated metrics measurements meet the SLA, requires comprehensive understanding of the real-time monitoring data of the application for the language model. In an embodiment, the static model may be fine-tuned with relevant real-time data to generate the dynamic model. (See Col. 16 line 62 - Col. 17 line 16 in regard to categories and names of data and how they relate to the KPIs. See also Col. 28 lines 6-21)).
using the structured data object to access, as a value for each of the variables used in the set of formulas, the key figure identified by the LLM from the data source identified by the LLM (Col. 42 line 56 - Col 43 line 4 teaches the first artificial intelligence (AI) model (e.g., the large language model (LLM)) determines whether the one or more actual performance levels of the one or more services associated with the one or more applications, are compliant with the one or more expected performance levels by comparing the one or more actual performance levels with one or more pre-defined key performance indicators (KPIs) and the one or more pre-defined goals of the one or more services defined in at least one of: the service level agreements (SLAs) and the service level objectives (SLOs). The first artificial intelligence (AI) model is trained with the one or more pre-defined rules and criteria to assess the one or more pre-defined key performance indicators (KPIs) and the one or more pre-defined goals defined in at least one of: the service level agreements (SLAs) and the service level objectives (SLOs)).
generating a second prompt for the LLM, the second prompt comprising the set of formulas and the values for the variables used in the set of formulas receiving, from the LLM and in response to the second prompt, the set of KPIs (Col. 18 line 58 - Col. 19 line 3 teaches the plurality of subsystems 114 further includes the data extraction subsystem 220 that is communicatively connected to the one or more hardware processors 110. The data extraction subsystem 220 is configured to extract one or more second real-time data from the one or more first real-time data for fine-tuning the large language model (LLM) using a second artificial intelligence (AI) model. In an embodiment, the second artificial intelligence (AI) model may be a small language model (SLM). For extracting the one or more second real-time data from the one or more first real-time data, the data extraction subsystem 220 is configured to obtain the one or more first real-time data from the one or more monitoring platforms using a data ingestion layer. The data ingestion layer is configured to determine whether the one or more first real-time data are obtained efficiently and to preprocess the one or more first real-time data to determine whether the one or more first real-time data include consistency and compatibility across the one or more monitoring platforms. Col. 29 line 60 - Col. 30. line 5 teaches the multi-prompt method 1004 may be configured to apply few-shot prompting to generate ReAct trajectories with a strong large language model, including GPT-4. This process includes creating thought-action-observation rounds to guide the large language model through task-solving processes. The multi-prompt method 1004 may be further configured to convert Chain of Thought (CoT) prompts into one-round ReAct trajectories where the “thought” is the reasoning step, and the “action” is the answer to the question. The multi-prompt method 1004 may be further configured to provide Reflexion trajectories by prompting for reflections at specific rounds, allowing the large language model to adjust its strategy in solving the one or more tasks. Col. 42 line 56 - Col 43 line 4 teaches the first artificial intelligence (AI) model (e.g., the large language model (LLM)) determines whether the one or more actual performance levels of the one or more services associated with the one or more applications, are compliant with the one or more expected performance levels by comparing the one or more actual performance levels with one or more pre-defined key performance indicators (KPIs) and the one or more pre-defined goals of the one or more services defined in at least one of: the service level agreements (SLAs) and the service level objectives (SLOs). The first artificial intelligence (AI) model is trained with the one or more pre-defined rules and criteria to assess the one or more pre-defined key performance indicators (KPIs) and the one or more pre-defined goals defined in at least one of: the service level agreements (SLAs) and the service level objectives (SLOs). (See Col. 16 line 62 - Col. 17 line 16 in regard to categories and names of data and how they relate to the KPIs. See also Col. 28 lines 6-21)).
causing presentation of the set of KPIs in the user interface. (Col. 13 lines 12-45 teaches the one or more electronic devices 106 may present to the one or more user interfaces for the one or more user to interact with the computer-implemented system 102 and/or to the one or more databases 104 for artificial intelligence (AI) driven autonomic application management framework need. The one or more electronic devices 106 may be at least one of, an electrical, an electronic, an electromechanical, and a computing device. The one or more electronic devices 106 may include, but is not limited to, a mobile device, a smartphone, a personal digital assistant (PDA), a tablet computer, a phablet computer, a wearable computing device, a virtual reality/augmented reality (VR/AR) device, a laptop, a desktop, a server, and the like. Col. 28 lines 40-56 teach the large language model is configured to perform complex reasoning tasks requiring expert knowledge across a wide range of fields, including in specialized domains such as programming and creative writing. The large language model may interact with humans through intuitive chat interfaces, which has led to rapid and widespread adoption among public).
Manohar teaches calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources, but does not explicitly teach input-output pairs and output that explicitly links two pieces of data together which is taught by the following citations from Makhija:
receiving, via a user interface, a selection of a plurality of data sources (Paragraph Number [0149] teaches the data processing system enables the user to consume out-of-the-box template workflows to solve their integration needs. The AI Engine of the data processing system is trained to select and generate workflows based on the various prompts and templates. As an example, the user would interact with the conversational assistant and prompt the system that they would like to find available integrations for their operational scenarios. The system would identify the intent and ask the user to select their source and target systems. The User can select from the options provided by the system or enter manually the details of the source and target systems. Paragraph Number [0150] teaches the user selects the integration in which they want to update the mapping. The AI engine checks if the integration has more than one mapping. If there is more than one mapping, the system identifies the mapping to be updated based on the AI engine. Once the mapping is identified, the Source and Target schemas are uploaded. If there is only one mapping, the user is asked to upload the Source and Target schemas. Once the user uploads the schemas, the AI engine determines the possible field to field mapping and provides the recommended mapping to the user and the mapping is updated into the integration. Further, if the user provides a Pseudocode mapping between Source and target, the data processing system utilizes the AI engine to generate the mapping).
receiving, from the LLM and in response to the prompt, a data source and key figure pair for each variable name in the set of variable names… based on the data source and key figure pairs (Paragraph Number [0105] teaches a method of fine tuning the one or more LLM agent includes the step of loading a plurality of historical dataset related to one or more application workflows of the codeless platform into a vector index to enable semantic search. The finetuning method includes the step of triggering one or more unit of task action descriptions index and a knowledge graph on units of task as additional tools for the one or more LLM agent. The finetuning method includes generating variations of the input requiring cross-referencing the unit of task action descriptions and knowledge graph including substituting steps within the workflow or augmenting the input with additional flows. The finetuning further includes running the input including the variations through a reference LLM, identifying one or more high reward input-output pair for fine tuning the LLM agents, and evaluating on a testing dataset, contextualization and substitution ability of the one or more LLM agents through the description index wherein a matrix is utilized on the testing dataset to assess the one or more LLM agents).
into a structured data object that enforces a schema defining the data source and key figure pair for each variable name (Paragraph Number [0060] teaches the plurality of configurable components includes one or more data layer configurable components including but not limited to Query builder, graph database parser, data service connector, transaction handler, document structure parser, event store parser and tenant access manager. The data layer provides abstracted layers to the SCM service to perform data operations like Query, insert, update, delete and Join on various types of data stores document database (DB) structure, relational structure, key value structure and hierarchical structure. Paragraph Number [0150] teaches the data processing system for integration of one application with another application utilizes automapping triggered by the AI engine. Mapping is the method of creating data transformation logic between source and target for the data being transmitted through interface. For example, in a scenario when data is received from a source system in a certain format, it is automatically transformed into a format that is understandable by the target system through Artificial intelligence AI engine. As an example, the user opens the conversational assistant/chat window and enters the prompt that they would like to update the mapping in their integration. Based on the integration details provided by the user, the system responds with one or more integrations that match the details. The system will request the user to select the integration that they intend to update the mapping. The user selects the integration in which they want to update the mapping. The AI engine checks if the integration has more than one mapping. If there is more than one mapping, the system identifies the mapping to be updated based on the AI engine. Once the mapping is identified, the Source and Target schemas are uploaded. If there is only one mapping, the user is asked to upload the Source and Target schemas. Once the user uploads the schemas, the AI engine determines the possible field to field mapping and provides the recommended mapping to the user and the mapping is updated into the integration. Further, if the user provides a Pseudocode mapping between Source and target, the data processing system utilizes the AI engine to generate the mapping).
Both Manohar and Makhija are directed to analysis of data through use of LLMs. Manohar discloses calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources. Makhija improves upon Manohar by disclosing input-output pairs and output that explicitly links two pieces of data together. One of ordinary skill in the art would be motivated to further include input-output pairs and output that explicitly links two pieces of data together, to efficiently compare data sets and results and to increase accuracy of the data training for the LLMs. Accordingly, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system and method of calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources in Manohar to further utilize input-output pairs and output that explicitly links two pieces of data together as disclosed in Makhija, since the claimed invention is merely a combination of old elements, and in combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Manohar teaches calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources, but does not explicitly teach a key figure in one of the data sources not being an exact match for any of the names of the variables which is taught by the following citations from Gromenko:
a key figure in one of the data sources not being an exact match for any of the names of the variables (Paragraph Number [0099] teaches task Matching: Create a function to match tasks across different sources. This function should take an identifier and compare it to the task descriptions in each source. The comparison could be an exact match or use some form of fuzzy matching (like Levenshtein distance or cosine similarity). Paragraph Number [0103] teaches Fuzzy Matching: This approach allows for approximate matches, which can be useful if the task descriptions aren't exactly the same but are still referring to the same task. Paragraph Number [0107] teaches choosing between exact and fuzzy matching (and choosing which fuzzy matching technique to use) would depend on your specific use case. You may want to experiment with different approaches and see which works best for your data).
the key figure that is not an exact match for any of the names of the variables being included in one of the data source and key figure pairs (Paragraph Number [0099] teaches task Matching: Create a function to match tasks across different sources. This function should take an identifier and compare it to the task descriptions in each source. The comparison could be an exact match or use some form of fuzzy matching (like Levenshtein distance or cosine similarity). Paragraph Number [0103] teaches Fuzzy Matching: This approach allows for approximate matches, which can be useful if the task descriptions aren't exactly the same but are still referring to the same task. Paragraph Number [0107] teaches choosing between exact and fuzzy matching (and choosing which fuzzy matching technique to use) would depend on your specific use case. You may want to experiment with different approaches and see which works best for your data).
Both the combination of Manohar and Makhija and Gromenko are directed to analysis of data through use of LLMs. The combination of Manohar and Makhija discloses calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources. Gromenko improves upon the combination of Manohar and Makhija by teaching a key figure in one of the data sources not being an exact match for any of the names of the variables. One of ordinary skill in the art would be motivated to further include a key figure in one of the data sources not being an exact match for any of the names of the variables, to efficiently allow for approximate matches, which can be useful if the task descriptions aren't exactly the same but are still referring to the same task. Accordingly, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system and method of calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources in the combination of Manohar and Makhija to further utilize a key figure in one of the data sources not being an exact match for any of the names of the variables as disclosed in Gromenko, since the claimed invention is merely a combination of old elements, and in combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Manohar teaches calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources, but does not explicitly teach passing tokenized information through a second prompt by parsing the data output of the first prompt which is taught by the following citations from Sheppard-Guzelaydin:
wherein the LLM is not provided with contents of the plurality of data sources in the first prompt, such that token consumption of the LLM is reduced relative to providing the contents of the plurality of data sources to the LLM (Paragraph Number [0034] teaches the systems can include a plurality of machine learning models that may be configured using integrated or disparate data sources. This can facilitate more integrated user experiences or more specialized (and/or lower computational usage for) data processing and output generation. Outputs from one or more first systems, such as one or more first algorithms or machine learning models, can be provided at least as part of inputs to one or more second systems, such as one or more second algorithms or machine learning models. For example, a first language model can be configured to process unstructured inputs (e.g., text, speech, images, etc.) into a structure output format compatible for use by a second system, such as a root cause prediction algorithm or equipment configuration model. Paragraph Number [0043] teaches the first model 104 can include at least one GPT model. The GPT model can receive an input sequence, and can parse the input sequence to determine a sequence of tokens (e.g., words or other semantic units of the input sequence, such as by using Byte Pair Encoding tokenization). The GPT model can include or be coupled with a vocabulary of tokens, which can be represented as a one-hot encoding vector, where each token of the vocabulary has a corresponding index in the encoding vector; as such, the GPT model can convert the input sequence into a modified input sequence, such as by applying an embedding matrix to the token tokens of the input sequence (e.g., using a neural network embedding function), and/or applying positional encoding (e.g., sin-cosine positional encoding) to the tokens of the input sequence. The GPT model can process the modified input sequence to determine a next token in the sequence (e.g., to append to the end of the sequence), such as by determining probability scores indicating the likelihood of one or more candidate tokens being the next token, and selecting the next token according to the probability scores (e.g., selecting the candidate token having the highest probability scores as the next token). For example, the GPT model can apply various attention and/or transformer based operations or networks to the modified input sequence to identify relationships between tokens for detecting the next token to form the output sequence).
parsing the response from the LLM to the first prompt, using an output parser (Paragraph Number [0043] teaches the first model 104 can include at least one GPT model. The GPT model can receive an input sequence, and can parse the input sequence to determine a sequence of tokens (e.g., words or other semantic units of the input sequence, such as by using Byte Pair Encoding tokenization). The GPT model can include or be coupled with a vocabulary of tokens, which can be represented as a one-hot encoding vector, where each token of the vocabulary has a corresponding index in the encoding vector; as such, the GPT model can convert the input sequence into a modified input sequence, such as by applying an embedding matrix to the token tokens of the input sequence (e.g., using a neural network embedding function), and/or applying positional encoding (e.g., sin-cosine positional encoding) to the tokens of the input sequence. The GPT model can process the modified input sequence to determine a next token in the sequence (e.g., to append to the end of the sequence), such as by determining probability scores indicating the likelihood of one or more candidate tokens being the next token, and selecting the next token according to the probability scores (e.g., selecting the candidate token having the highest probability scores as the next token). For example, the GPT model can apply various attention and/or transformer based operations or networks to the modified input sequence to identify relationships between tokens for detecting the next token to form the output sequence).
Both the combination of Manohar, Makhija, and Gromenko and Sheppard-Guzelaydin are directed to analysis of data through use of LLMs. The combination of Manohar, Makhija, and Gromenko discloses calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources. Sheppard-Guzelaydin improves upon the combination of Manohar, Makhija, and Gromenko by teaching passing tokenized information through a second prompt by parsing the data output of the first prompt. One of ordinary skill in the art would be motivated to further include passing tokenized information through a second prompt by parsing the data output of the first prompt, to efficiently utilize only certain data when implementing more than one LLM model to reduce extraneous computations and improve categorization of data. Accordingly, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system and method of calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources in the combination of Manohar, Makhija, and Gromenko to further utilize passing tokenized information through a second prompt by parsing the data output of the first prompt as disclosed in Sheppard-Guzelaydin, since the claimed invention is merely a combination of old elements, and in combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 17, Manohar teaches:
A non-transitory computer-readable medium that stores instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising: (Col. 16 lines 30-61 teach the one or more hardware processors 110, as used herein, means any type of computational circuit, such as, but not limited to, a microprocessor unit, microcontroller, complex instruction set computing exceptionally long processor unit, reduced instruction set computing microprocessor unit, very long instruction word microprocessor unit, explicitly parallel instruction computing microprocessor unit, graphics processing unit, digital signal processing unit, or any other type of processing circuit. The one or more hardware processors 110 may also include embedded controllers, such as generic or programmable logic devices or arrays, application-specific integrated circuits, single-chip computers, and the like. The memory 112 may be a non-transitory volatile memory and a non-volatile memory. The memory 112 may be coupled to communicate with the one or more hardware processors 110, such as being a computer-readable storage medium. The one or more hardware processors 110 may execute machine-readable instructions and/or source code stored in the memory 112. A variety of machine-readable instructions may be stored in and accessed from the memory 112).
The remainder of the claim limitations are substantially similar to those found in claim 1 and are rejected for the same reasons put forth in regard to claim 1.
As per claim 15, claim 15 recites a method that is substantially similar to that performed by the system of claim 1 and is rejected for the same reasons put forth in regard to claim 1.
As per claims 2, 9, and 16, the combination of Manohar, Makhija, Gromenko and Sheppard-Guzelaydin teaches each of the limitations of claims 1, 8, and 15 respectively.
In addition, Manohar teaches:
wherein the operations further comprise receiving a user request for the set of KPIs via the user interface. (Col. 13 lines 12-45 teaches the one or more electronic devices 106 may present to the one or more user interfaces for the one or more user to interact with the computer-implemented system 102 and/or to the one or more databases 104 for artificial intelligence (AI) driven autonomic application management framework need. Col 27 lines 26-46 teaches the service level statement “SLO for performance refers to minimum response rate of less than 30 ms for web applications” is broken to determine how the large language model learns from the documents. In an embodiment, the web application may be selected to explain the semantic analysis, structure analysis, and implicit rules. In the semantic analysis, a meaning statement describes a service level objective (SLO) related to the performance of the web applications. Specifically, the meaning statement specifies a criterion for performance, stating that the minimum response rate should be less than 30 milliseconds (ms). This means that the web application should ideally respond to user requests within this timeframe to meet the defined performance standard).
As per claims 3, 10, and 17, the combination of Manohar, Makhija, Gromenko and Sheppard-Guzelaydin teaches each of the limitations of claims 1, 8, and 15 respectively.
In addition, Manohar teaches:
wherein the determining of the set of formulas comprises accessing a KPI formula database. (Col. 29 lines 11-23 teach the one or more instruction sets are used for fine tuning the large language model are based on ReACT trajectories. The one or more data which are stored in the vector database 802 may have private data (i.e., obtained from the documents) and public data (i.e., obtained from gpt3, and google search). The one or more data 702 need to be converted into the one or more instruction sets, as a first set and the one or more data obtained from the vector database 802 need to be converted into one or more question answer formats 1002 to cover the reasoning and facts. The one or more question answer formats 1002 may include HotpotQA, StrategyQA, and Massive Multitask Language Understanding (MMLU). Col. 33 lines 24-35 teach the one or more instance data stored in the interim database 1506 need to be translated, as shown in 1508, into one or more formats for fine-tuning the instance model. The one or more instance data need to be mapped to specific SLA, SLO and metric available in the static model fine-tuned).
As per claims 5, 12, and 19, the combination of Manohar, Makhija, Gromenko and Sheppard-Guzelaydin teaches each of the limitations of claims 1, 8, and 15 respectively.
Manohar teaches calculating KPIs by creating and implementing an LLM prompt and gathering information from disparate data sources, but does not explicitly teach input-output pairs and output that explicitly links two pieces of data together which is taught by the following citations from Makhija:
wherein the LLM removes a non-numeric character from at least one value (Paragraph Number [0103] teaches the method of training one or more LLM agent and the master controller LLM agent includes the step of 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. The training includes the step of 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. Further, the training method includes 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).
A person of ordinary skill in the art would have been motivated to combine these references as described in regard to claim 1.
As per claims 6, 13, and 20, the combination of Manohar, Makhija, Gromenko and Sheppard-Guzelaydin teaches each of the limitations of claims 1 and 5, 8 and 12, and 15 and 19 respectively.
In addition, Manohar teaches:
wherein the LLM determines a KPI by providing the value to a calculator tool and reading an output from the calculator tool. (Col. 42 line 56 - Col 43 line 4 teaches the first artificial intelligence (AI) model (e.g., the large language model (LLM)) determines whether the one or more actual performance levels of the one or more services associated with the one or more applications, are compliant with the one or more expected performance levels by comparing the one or more actual performance levels with one or more pre-defined key performance indicators (KPIs) and the one or more pre-defined goals of the one or more services defined in at least one of: the service level agreements (SLAs) and the service level objectives (SLOs). The first artificial intelligence (AI) model is trained with the one or more pre-defined rules and criteria to assess the one or more pre-defined key performance indicators (KPIs) and the one or more pre-defined goals defined in at least one of: the service level agreements (SLAs) and the service level objectives (SLOs)).
As per claims 7, 14, and 24 the combination of Manohar, Makhija, Gromenko and Sheppard-Guzelaydin teaches each of the limitations of claims 1, 8, and 15 respectively.
In addition, Manohar teaches:
wherein the LLM determines the data source and key figure pair for at least one variable name based on semantic similarity (Col. 28 lines 22-39 teach a block diagram 900 representing a fine-tuning process of the artificial intelligence (AI) model, in accordance with an embodiment of the present disclosure. The large language model may understand the one or more semantics, structure, and context of the specific language which is trained. Extending the large language models to understand the semantic, structure and context of “SLA, SLO automation” domain and further understanding the real-time data which comes about as a part of monitoring the applications provide the models the ability to (a) understand a current state of the target system, (b) evaluate how compliant is the target system from the defined goals, (c) recommend the one or more actions based on intelligent insights, (d) continuous learning and evaluate new information, and (e) drive the target system towards optimized compliance. In an embodiment, the vector database 802 may have at least one of: specific data, generic and other data types).
As per claims 25 and 26, the combination of Manohar, Makhija, Gromenko and Sheppard-Guzelaydin teaches each of the limitations of claims 1 and 8 respectively.
In addition, Manohar teaches:
wherein the first prompt instructs the LLM to identify exactly one key figure and data source pair for each variable name in the set of variable names (Col. 30 lines 26-38 teach the fine-tuned language model is deployed for inference without the need for few-shot prompting, thereby making the process more efficient. In an embodiment, the fine-tuned agent may be capable of adapting its method based on task complexity, providing strong generalization and robustness because of the diverse learning support that the language model is received during training. In an embodiment, the performance of the fine-tuned language model is evaluated using a set of 500 dev questions from HotpotQA and questions from other datasets. In an embodiment, the evaluation focuses on exact match (EM) scores to determine an accuracy of the fine-tuned language model in answering the questions correctly).
Response to Arguments
Applicant’s arguments filed 5/27/2026 have been fully considered but they are not persuasive.
Applicant argues that the previously cited reference does not teach the newly amended portions including the new limitations recited by the independent claims. (See Applicant’s Remarks, 5/27/2026, pgs. 11-14). Examiner respectfully disagrees. Examiner has added citations from the Makhija reference and the new Sheppard-Guzelaydin reference in response to Applicant’s amendments and provided for additional citations as explanation for the claim language in general as shown in the above new 35 USC 103 rejection. As such, Applicant’s arguments directed towards the previous 35 USC 103 rejection are moot. In response to Applicant’s arguments, Examiner directs Applicant to review the new citations and explanations provided in the new 35 USC 103 rejection presented above.
Conclusion
Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 extension fee 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 date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MATTHEW H DIVELBISS whose telephone number is (571)270-0166. The examiner can normally be reached on 7:30 am - 6:00 PM. 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, Jerry O'Connor can be reached on (571) 272-6787. 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.
/M. H. D./
Examiner, Art Unit 3624
/Jerry O'Connor/Supervisory Patent Examiner,Group Art Unit 3624
Prosecution Insights