DETAILED ACTION
This office action is in response to claims filed 21 May 2026.
Claims 1-30 are pending.
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 .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 21 May 2026 has been entered.
Response to Arguments
Applicant's arguments filed 21 May 2026 regarding the claim objections, drawing objections, and rejections under 35 U.S.C. 112 have been fully considered and are persuasive. They have been withdrawn.
On page 9 of the remarks, the applicant argues:
“Claims 1-30 were rejected under 35 U.S.C. § 101 for the asserted reasons the claimed invention is directed to an abstract idea (mental process) without significantly more. In particular, the instant action states that the claimed invention "under the broadest reasonable interpretation, covers a mental process but for recitation of generic computer components." See, instant action, page 4. Without conceding the foregoing rejections, in the interest of accelerating prosecution the independent claims have been amended. Applicant respectfully submits that the independent claims, as amended herein, are not directed toward an abstract idea. Further, even if the independent claims could be considered to include an abstract idea, the amendments herein integrate any such abstract idea into a practical application thereof. Accordingly, withdrawal of this rejection is respectfully requested.”
The examiner respectfully disagrees. The applicant’s argument amounts to no more than a conclusory statement that the amended claims recite patent eligible subject matter, and do not explain why the newly amended claims are not directed to an abstract idea, or do in fact integrate an abstract idea into a practical application. Since the current office action has determined that the amended claims are not patent eligible, and the applicant’s remarks fail to provide any explanation for their position that the claims are patent eligible, the applicant’s arguments are not persuasive.
On pages 9-11 in the remarks, the applicant argues:
“As shown above, independent claim 1 has been amended to recite, in relevant part "defining a pool of available generative artificial intelligence (AI) resources, wherein the pool of available generative Al resources includes a plurality of discrete generative Al resources, thereby maintaining a model repository, wherein the model repository includes the available generative Al model and metadata about the models, including one or more of: type, performance metrics, intended use cases, and information training datasets." Applicant respectfully submits that the combination of cited references is not understood, and have not been asserted, to disclose such features.
“In consideration of the foregoing, Applicant respectfully submits that amended independent claim 1 is in condition for allowance. As independent claims 11 and 21 have been amended to recite similar features as claim 1, claims 11 and 21 are also believed to be in condition for allowance. Because the remaining claims ultimately depend upon one of claims 1, 11, and 21, the remaining claims are also believed to be in condition for allowance. Accordingly, withdrawal of the rejections under 35 U.S.C. § 103 is respectfully requested.”
The examiner respectfully disagrees. The applicant’s argument fails to take into account the prior art as a whole, specifically the newly referenced portions of PAES (cited below) used to address the newly added limitations below. Specifically, PAES [0002], [0025], [0027], [0029], and [0046] are used to reject the limitations at issue, and are not addressed specifically in the remarks. Therefore, applicant’s arguments are not persuasive.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-30 rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Regarding claims 1, 11, and 21 (line number corresponds to claim 1),
In lines 5-6, the claim fails to particularly point out and distinctly claim what is meant by “the available generative AI model” (i.e., there is a lack of antecedent basis for this term). The examiner will interpret this as “the available generative AI resources”.
In line 6, the claim fails to particularly point out and distinctly claim what is meant by “the models” (i.e., there is a lack of antecedent basis for this term). The examiner will interpret this as “the available generative AI resources”.
Regarding claims 2-10, 12-20, and 22-30, they depend on rejected claims and fail to resolve the deficiencies thereof. They are therefore rejected based on their dependency.
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 1-30 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea (mental process) without significantly more.
Regarding claim 1, in step 1 of the 101 analysis set forth in MPEP 2106, the claim recites a method that routes generative AI requests to resources based on monitored utilization. A method is one of the four statutory categories of invention.
In step 2A, prong 1 of the 101 analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a mental process but for recitation of generic computer components:
i. “defining a pool of available generative artificial intelligence (Al) resources…thereby maintaining a model repository” (a person can mentally define a pool of resources by simply observing the resources and making a judgement that certain resources should be grouped in a pool (MPEP 2106)).
ii. “enabling drag and drop design of a generative AI workflow utilizing the generative AI resources” (a person can mentally enable design of a workflow by simply making a judgement design of a workflow should be enabled (MPEP 2106)).
iii. “monitoring the utilization of the plurality of discrete generative Al resources to define utilization statistics” (a person can mentally monitor utilization of resources by simply observing or evaluating resource utilization data and making a judgement of utilization statistics (MPEP 2106)).
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic computer components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
In step 2A, prong 2 of the 101 analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application:
iv. “A computer-implemented method, executed on a computing device” (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))).
v. “the pool of available generative Al resources includes a plurality of discrete generative Al resources” (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))).
vi. “wherein the model repository includes the available generative AI model and metadata about the models, including one or more of: type, performance metrics, intended use cases, and information training datasets” (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))).
vii. “via graphical user interface” (mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f))).
viii. “receiving a request for the pool of available generative Al resources” (insignificant extra-solution activity of mere data storage (MPEP 2106.05(g))).
ix. “routing at least a portion of the request to one of the plurality of discrete generative Al resources” (insignificant extra-solution activity of mere data output (MPEP 2106.05(g))).
x. “routing…based, at least in part, upon the utilization statistics” (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))).
Since the claim does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea.
In step 2B of the 101 analysis set forth in the 2019 PEG, the examiner has determined through reanalysis of the following limitations considered in step 2A prong 2, that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception:
iv. “A computer-implemented method, executed on a computing device” (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))).
v. “the pool of available generative Al resources includes a plurality of discrete generative Al resources” (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))).
vi. “wherein the model repository includes the available generative AI model and metadata about the models, including one or more of: type, performance metrics, intended use cases, and information training datasets” (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))).
vii. “via graphical user interface” (mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f))).
viii. “receiving a request for the pool of available generative Al resources” (a well-understood, routine, and conventional activity of receiving data over a network (MPEP 2106.05(d)(II))).
ix. “routing at least a portion of the request to one of the plurality of discrete generative Al resources” (a well-understood, routine, and conventional activity of transmitting data over a network (MPEP 2106.05(d)(II))).
x. “routing…based, at least in part, upon the utilization statistics” (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))).
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Regarding claim 2, the additional element “the pool of available Al resources spans a single generative Al model” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claim 3, the additional element “the pool of available Al resources spans a plurality of generative Al models” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claim 4, the additional element “the pool of available Al resources spans one or more of a plurality of accounts and a plurality of regions for a single generative Al model” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claim 5, the additional element “the pool of available Al resources spans one or more of a plurality of accounts and a plurality of regions for a plurality of generative Al model” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claim 6, the additional element “the plurality of discrete generative Al resources includes one or more of: a reasoning generative Al resource; a chat generative Al resource; a text completion generative Al resource; a embedding generative Al resource; an image generation generative Al resource; and a reranker generative Al resource” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claim 7, the additional element “the request includes one or more routing restrictions concerning the plurality of discrete generative Al resources” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claim 8, the additional element “routing at least a portion of the request to one of the plurality of discrete generative Al resources based, at least in part upon the utilization statistics includes: routing at least a portion of the request to one of the plurality of discrete generative Al resources” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (insignificant extra-solution activity of mere data output (MPEP 2106.05(g)), and under step 2B it does not amount to significantly more than the judicial exception (well-understood, routine and conventional activity of transmitting data over a network (MPEP 2106.05(d)(II)). Further, the additional element “based, at least in part, upon the utilization statistics and the one or more routing restrictions” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claim 9, the additional element “the one or more routing restrictions define one or more of: a preferred generative Al model; a preferred type of generative Al model; a preferred account for a generative Al model; and a preferred region for a generative Al model. ” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claim 10, the additional element “the utilization statistics define one or more of: a number of requests made to each of the plurality of discrete generative Al resources during a given period of time; one or more of a throughput and a token count for each of the plurality of discrete generative Al resources during a given period of time; a cost count for each of the plurality of discrete generative Al resources during a given period of time; and a compute count for each of the plurality of discrete generative Al resources during a given period of time” does not render the claim patent eligible because under step 2A prong 2, it does not integrate the judicial exception into a practical application (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)), and under step 2B it does not amount to significantly more than the judicial exception (generally links the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)).
Regarding claims 11-30, they comprise limitations similar to those of claims 1-10, and are therefore rejected for at least similar rationale.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 6-8, 10-11, 16-18, 20-21, 26-28, and 30 are rejected under 35 U.S.C. 103 as being unpatentable over PAES et al. Pub. No.: US 2025/0209200 A1 (hereafter PAES), in view of PAPANCEA et al. Pub. No.: US 2025/0356187 A1 (hereafter PAPANCEA), in view of MYSTETSKYI et al. Pub. No.: US 2025/0244964 A1 (hereafter MYSTETSKYI).
PAES, PAPANCEA, and MYSTETSKYI were cited previously.
Regarding claim 1, PAES teaches the invention substantially as claimed, including:
A computer-implemented method, executed on a computing device, comprising:
defining a pool of available generative artificial intelligence (Al) resources, wherein the pool of available generative Al resources includes a plurality of discrete generative Al resources, thereby maintaining a model repository, wherein the model repository includes the available generative AI model ([0020] FIG. 1A depicts a block diagram of a generative AI system integration environment 100A, according to some embodiments. Generative AI system integration environment 100A includes…generative AI systems 150 (i.e., “pool” of “discrete generative AI resources” available for use)), and metadata about the models ([0025] A request may be entered on client system 110 to initiate a request with a query for a generative AI system 150. For example, the request may specify a particular generative AI system 150 (i.e., particular models may be specified based on attributes or characteristics (i.e., “metadata”) of the models)), including one or more of: type ([0002] Many of these factors, e.g. security, operational controls, and governance, may be at least partially controlled by individual generative AI systems. All of which may have different standards and processes in place, depending on vendor of the generative AI (i.e., each generative AI system has its own "type" defined by security, operational control, and governance factors/standards/processes)), performance metrics ([0027] Observability service 132 may collect metadata associated with the requests and responses to determine performance analytics. In some embodiments, observability service 132 may use HTTP headers of the requests and responses to collect data such as usage of individual generative AI systems 150, frequency of requests from client systems 110, request and/or response timing data, and other metrics that may be measured using the metadata associated with the requests and responses (i.e., performance analytics represent “performance metrics”)), intended use cases ([0029] Synchronous controls 134 may access a ruleset stored in controls database 140. The ruleset may include conditions that indicate whether the query may be forwarded to generative AI systems 150. For example, the ruleset may include a rule regarding character limit, which limits the number of characters in the content of the query forwarded to generative AI systems 150. If the content of the query exceeds the character limit, then the query requesting a response from generative AI systems 150 may be denied (i.e., use cases requiring character limits are maintained)), and information training datasets ([0029] Synchronous controls 134 may utilize a ruleset and/or one or more machine learning models. In some embodiments, the ruleset can include one or more conditions for forwarding the query or response. Synchronous controls 134 may access a ruleset stored in controls database 140...The rules administered may depend on client systems 110, the content of the query, generative AI systems 150, the content of the response, and/or a combination of these factors. [0046] Asynchronous controls 136 may use the content of the queries to train the machine learning models stored in controls database 140. As described above, controls database 140 may be a vector database, which uses vector embedding to translate text (e.g., the content of the query) into a vector. This vector indicates information about the content of the query that can be used to efficiently train the machine learning models used for implementing model-based rules (i.e., request content is used to train machine learning models and is therefore "information training datasets"));
monitoring the utilization of the plurality of discrete generative Al resources to define utilization statistics ([0027] observability service 132 may collect metadata associated with the requests and responses to determine performance analytics. In some embodiments, observability service 132 may use HTTP headers of the requests and responses to collect data such as usage (i.e., “utilization”) of individual generative AI systems 150, frequency of requests from client systems 110, request and/or response timing data, and other metrics that may be measured using the metadata associated with the requests and response);
receiving a request for the pool of available generative Al resources ([0057] At 310, request management system 120 may receive, from client system 110, a generative AI query that requests a response from at least one generative AI system 150); and
routing at least a portion of the request to one of the plurality of discrete generative Al resources ([0061] At 340, in response to determining that content of the query satisfies the ruleset, request management system 120 transmits the generative AI query to the one or more generative AI systems 150. In some embodiments, the query and request may have specifically identified a generative AI system 150 to forward the request to for a response).
While PAES discusses utilizing a generative AI workflow to handle requests, PAES does not explicitly teach:
enabling a drag and drop design of a generative AI workflow utilizing the generative AI resources via a graphical user interface.
However, in analogous art that similarly teaches utilizing generative AI workflows to handle requests, PAPANCEA teaches:
enabling a drag and drop design of a generative AI workflow utilizing the generative AI resources via a graphical user interface ([0035] As shown in FIG. 7, the method 700 includes at 710 receiving, through the no-code user interface, structured portion indicators associated with structured workflow portions…These structured portion indicators each can be, for example, selected by a user through a drag-and-drop technique that provides a template on the canvas (displayed background) of the no-code user interface such that the user can then enter specific values into that template of that structured portion indicator. [0036] The method 700 also includes at 720 receiving, through the no-code user interface, an indicator of at least one unstructured workflow portion that is configured to send a task description and a list of structured parameters to a generative artificial intelligence (AI) model (such as an LLM) for execution of the generative AI model using the task description and the list of structured parameters to obtained the desired information from the user through interactions between the user and the generative AI model…This unstructured portion indicator can be, for example, selected by a user through a drag-and-drop technique that provides a template on the canvas (displayed background) of the no-code user interface. [0038] The structured portion indicators and the unstructured portion indicators are related to (and in some instances may match or used to define) the workflow portions of a workflow. The connector indicators can define the manner and order in which the various workflow portions are executed when a workflow is executed (i.e., portions of a generative AI workflow including structured and unstructured portions are dragged and dropped to assemble the generative AI workflow)).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to have combined PAPANCEA’s teaching of assembling a generative AI workflow using drag and drop actions by users, with PAES’s teaching utilizing a generative AI workflow to process AI requests, to realize, with a reasonable expectation of success, a system that utilizes a generative AI workflow to process AI requests, as in PAES, which is assembled using drag and drop design tools, as in PAPANCEA. A person having ordinary skill would have been motivated to make this combination to give users enhanced and intuitive control of, and ability to customize, a generative AI workflow.
While PAES discusses collecting usage data of generative AI systems, PAES and PAPANCEA does not explicitly disclose
routing at least a portion of the request to one of the plurality of discrete generative Al resources based, at least in part, upon the utilization statistics;
However, in analogous art, which similarly discusses generative AI resources executing requests, MYSTETSKYI teaches:
routing at least a portion of the request to one of the plurality of discrete generative Al resources based, at least in part, upon the utilization statistics ([0503] The process treats the plurality of AI agents as limited resources, with multiple instances of the same AI agent available for purchase and assignment to a limited number of items concurrently. Each AI agent instance is configured to be assigned to up to a predetermined number of items concurrently. [0505] When receiving a request to assign a generative AI agent instance to an item or platform element, the process determines whether the assignment would exceed the resource limit for the generative AI agent and allows or denies the assignment (i.e., “routes the request”) based on this determination (i.e., AI agent usage is used to determine whether to allow or deny a request assignment)).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to have combined MYSTETSKYI’s teaching of assigning requests for generative AI resources based on generative AI resource usage, with PAES and PAPANCEA’s teaching of assigning requests for generative AI resources, to realize, with a reasonable expectation of success, a system that assigns requests to generative AI resources, as in PAES and PAPANCEA, based on monitored utilization statistics, as in MYSTETSKYI. A person having ordinary skill would have been motivated to make this combination to ensure that resources are optimally utilized and balanced based on resource utilization (MYSTETSKYI [0376]).
Regarding claim 6, MYSTETSKYI further teaches:
a reasoning generative Al resource; a chat generative Al resource ([0050] Receive, via the user interface, a user query about a specific output generated by the generative AI agent; analyze the user query to identify the specific output being discussed; retrieve the metadata associated with the identified output; generate a natural language response explaining the reasoning behind the output, including references to the platform elements identified in the metadata; present the natural language response to the user via the user interface; engage in an interactive dialogue (i.e., “chat”) with the user to provide further clarification about the output and its underlying reasoning (i.e., “generating reasoning”) based on the platform elements (i.e., generative AI agents generate chats and reasoning))…
a embedding generative Al resource ([0159] Some non-limiting examples of such machine learning algorithms may include…mathematical embedding algorithms);
an image generation generative Al resource ([1034] As used herein a generative AI model is a function trained using machine learning technical to receive inputs such as text, image, audio, video, and code and generate new content into any of defined modalities. For example, it can turn text inputs into an image (i.e., “image generation”));
a reranker generative AI resource (The prior art teaches “one or more of” the list of alternatives)
Regarding claim 7, PAES further teaches:
the request includes one or more routing restrictions concerning the plurality of discrete generative Al resources ([0061] At 340, in response to determining that content of the query satisfies the ruleset, request management system 120 transmits the generative AI query to the one or more generative AI systems 150. In some embodiments, the query and request may have specifically identified a generative AI system 150 to forward the request to for a response (i.e., rulesets represent “routing restrictions”)).
Regarding claim 8, PAES further teaches:
routing at least a portion of the request to one of the plurality of discrete generative Al resources based, at least in part upon the utilization statistics includes: routing at least a portion of the request to one of the plurality of discrete generative Al resources based, at least in part, upon…the one or more routing restrictions ([0061] At 340, in response to determining that content of the query satisfies the ruleset, request management system 120 transmits the generative AI query to the one or more generative AI systems 150. In some embodiments, the query and request may have specifically identified a generative AI system 150 to forward the request to for a response).
MYSTETSKYI further teaches:
routing at least a portion of the request to one of the plurality of discrete generative Al resources based, at least in part, upon the utilization statistics ([0505] When receiving a request to assign a generative AI agent instance to an item or platform element, the process determines whether the assignment would exceed the resource limit for the generative AI agent and allows or denies the assignment (i.e., “routes the request”) based on this determination (i.e., AI agent usage is used to determine whether to allow or deny a request assignment)).
Regarding claim 10, PAES further teaches:
the utilization statistics define one or more of:
a number of requests made to each of the plurality of discrete generative Al resources during a given period of time; one or more of a throughput and a token count for each of the plurality of discrete generative Al resources during a given period of time ([0027] Observability service 132 may collect metadata associated with the requests and responses to determine performance analytics…observability service 132 may use HTTP headers of the requests and responses to collect data such as usage of individual generative AI systems 150, frequency of requests from client systems 110, request and/or response timing data, and other metrics that may be measured using the metadata associated with the requests and responses (i.e., frequency of requests from client systems represents “number of requests made…during a given period of time” for a particular generative AI system, and timing data for requests and responses indicates “throughput” of the request for a particular AI system));
a cost count for each of the plurality of discrete generative Al resources during a given period of time; and a compute count for each of the plurality of discrete generative Al resources during a given period of time (The prior art teaches “one or more of” the list of alternatives)).
Regarding claims 11, 16-18, 20-21, 26-28, and 30, they comprise limitations similar to those of claims 1, 6-8, and 10, and are therefore rejected for similar rationale.
Claims 2-5, 12-15, and 22-25 are rejected under 35 U.S.C. 103 as being unpatentable over PAES, in view of PAPANCEA, in view of MYSTETSKYI, as applied to claims 1, 11, and 21 above, and in further view of SULLIVAN Pub. No.: US 2025/0138490 A1 (hereafter SULLIVAN).
SULLIVAN was cited previously.
Regarding claim 2, while PAES, in view of MYSTETSKYI discusses pools of generative AI resources, PAES, in view of PAPANCEA, in view of MYSTETSKYI does not explicitly teach:
the pool of available Al resources spans a single generative Al model.
However, in analogous art that similarly discusses a pool of computational resources used to support artificial intelligence workloads, SULLIVAN teaches:
the pool of available Al resources spans a single generative Al model ([0166] A pool of computational resources that can be operated on one or more devices to implement one or more machine learning models. [0004] The machine learning model can include various machine learning model architectures (e.g., networks, backbones, algorithms, etc.), including but not limited to language models, LLMs, attention-based neural networks, transformer-based neural networks, generative pretrained transformer (GPT) models, bidirectional encoder representations from transformers (BERT) models, encoder/decoder models, sequence to sequence models, autoencoder models, generative adversarial networks (GANs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), diffusion models (e.g., denoising diffusion probabilistic models (DDPMs)), or various combinations thereof (i.e., resource pool supports, or spans a single generative AI model like GPT or GAN)).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to have simply substituted SULLIVAN’s teaching of a pool of generative AI resources supporting a single generative AI model, with the combination of PAES, PAPANCEA, and MYSTETSKYI’s teaching of a pool of generative AI resources used to perform requests, because 1) PAES, PAPANCEA, and MYSTETSKYI contain a system that differs from the claimed device in that it is not disclosed that a pool of generative AI resources spans a single generative AI model, 2) SULLIVAN teaches a pool of generative AI resources spans a single generative AI model, and 3) one of ordinary skill could have substituted the pool of generative AI resources in SULLIVAN with the pool of generative AI resources in PAES, PAPANCEA, and MYSTETSKYI to yield the predictable result of a pool of generative AI resources that is used to perform requests and which spans a single generative AI model.
Regarding claim 3, while PAES, PAPANCEA, in view of MYSTETSKYI discusses pools of generative AI resources, PAES, in view of PAPANCEA, in view of MYSTETSKYI does not explicitly teach:
the pool of available Al resources spans a plurality of generative Al models.
However, in analogous art that similarly discusses a pool of computational resources used to support artificial intelligence workloads, SULLIVAN teaches:
the pool of available Al resources spans a plurality of generative Al model ([0166] A pool of computational resources that can be operated on one or more devices to implement one or more machine learning models. [0004] The machine learning model can include various machine learning model architectures (e.g., networks, backbones, algorithms, etc.), including but not limited to language models, LLMs, attention-based neural networks, transformer-based neural networks, generative pretrained transformer (GPT) models, bidirectional encoder representations from transformers (BERT) models, encoder/decoder models, sequence to sequence models, autoencoder models, generative adversarial networks (GANs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), diffusion models (e.g., denoising diffusion probabilistic models (DDPMs)), or various combinations thereof (i.e., resource pool supports, or spans more than one generative AI model like GPT or GAN)).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to have simply substituted SULLIVAN’s teaching of a pool of generative AI resources supporting more than one generative AI model, with the combination of PAES, PAPANCEA, and MYSTETSKYI’s teaching of a pool of generative AI resources used to perform requests, because 1) PAES, PAPANCEA, and MYSTETSKYI contain a system that differs from the claimed device in that it is not disclosed that a pool of generative AI resources spans more than one generative AI model, 2) SULLIVAN teaches a pool of generative AI resources spans more than one generative AI model, and 3) one of ordinary skill could have substituted the pool of generative AI resources in SULLIVAN with the pool of generative AI resources in PAES, PAPANCEA, and MYSTETSKYI to yield the predictable result of a pool of generative AI resources that is used to perform requests and which spans more than one generative AI model.
Regarding claim 4, while PAES, in view of PAPANCEA, in view of MYSTETSKYI discusses pools of generative AI resources, PAES, in view of MYSTETSKYI does not explicitly teach:
the pool of available Al resources spans one or more of a plurality of accounts and a plurality of regions for a single generative Al model.
However, in analogous art that similarly discusses pools of generative AI resources, SULLIVAN teaches:
the pool of available Al resources spans one or more of a plurality of accounts and a plurality of regions for a single generative Al model ([0166] A pool of computational resources that can be operated on one or more devices to implement one or more machine learning models. [0004] The machine learning model can include various machine learning model architectures (e.g., networks, backbones, algorithms, etc.), including but not limited to language models, LLMs, attention-based neural networks, transformer-based neural networks, generative pretrained transformer (GPT) models, bidirectional encoder representations from transformers (BERT) models, encoder/decoder models, sequence to sequence models, autoencoder models, generative adversarial networks (GANs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), diffusion models (e.g., denoising diffusion probabilistic models (DDPMs)), or various combinations thereof (i.e., resource pool supports, or spans a single generative AI model like GPT or GAN)). [0085] The virtual assistant application 120 can include a plurality of applications 120 (e.g., variations of interfaces or customizations of interfaces) for a plurality of respective user types. For example, the virtual assistant application 120 can include a first application 120 for a customer user, and a second application 120 for a service technician user (i.e., customer user and service technician user represent different “accounts”)).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to have combined SULLIVAN’s teaching of a pool of AI resources supporting a single generative AI model having a plurality of associated accounts, with PAES, PAPANCEA, and MYSTETSKYI’s teaching of a pool of generative AI resources, to realize, with a reasonable expectation of success, a system that uses a pool of generative AI resources, as in PAES, PAPANCEA, and MYSTETSKYI, associated with different accounts of a single generative AI model, as in SULLIVAN. A person having ordinary skill would have been motivated to make this combination to enable a generative AI system to be more flexible in handling different types of users.
Regarding claim 5, while PAES, in view of PAPANCEA, in view of MYSTETSKYI discusses pools of generative AI resources, PAES, in view of PAPANCEA, in view of MYSTETSKYI does not explicitly teach:
the pool of available Al resources spans one or more of a plurality of accounts and a plurality of regions for a plurality of generative Al model.
However, in analogous art that similarly discusses pools of generative AI resources, SULLIVAN teaches:
the pool of available Al resources spans one or more of a plurality of accounts and a plurality of regions for a plurality of generative Al model ([0166] A pool of computational resources that can be operated on one or more devices to implement one or more machine learning models. [0004] The machine learning model can include various machine learning model architectures (e.g., networks, backbones, algorithms, etc.), including but not limited to language models, LLMs, attention-based neural networks, transformer-based neural networks, generative pretrained transformer (GPT) models, bidirectional encoder representations from transformers (BERT) models, encoder/decoder models, sequence to sequence models, autoencoder models, generative adversarial networks (GANs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), diffusion models (e.g., denoising diffusion probabilistic models (DDPMs)), or various combinations thereof (i.e., resource pool supports, or spans plural generative AI models like GPT or GAN)). [0085] The virtual assistant application 120 can include a plurality of applications 120 (e.g., variations of interfaces or customizations of interfaces) for a plurality of respective user types. For example, the virtual assistant application 120 can include a first application 120 for a customer user, and a second application 120 for a service technician user (i.e., customer user and service technician user represent different “accounts”)).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to have combined SULLIVAN’s teaching of a pool of AI resources supporting plural generative AI models having a plurality of associated accounts, with PAES, PAPANCEA, and MYSTETSKYI’s teaching of a pool of generative AI resources, to realize, with a reasonable expectation of success, a system that uses a pool of generative AI resources, as in PAES, PAPANCEA, and MYSTETSKYI, associated with different accounts of plural generative AI models, as in SULLIVAN. A person having ordinary skill would have been motivated to make this combination to enable a generative AI system to be more flexible in handling different types of users.
Regarding claims 12-15, and 22-25, they comprise limitations similar to those of claims 2-5, and are therefore rejected for similar rationale.
Claims 9, 19, and 29 are rejected under 35 U.S.C. 103 as being unpatentable over PAES, in view of PAPANCEA, in view of MYSTETSKYI, as applied to claims 1, 11, and 21 above, and in further view of DURVASULA et al. Pub. No.: US 2025/0225401 A1 (hereafter DURVASULA).
DURVASULA was cited previously.
Regarding claim 9, while PAES, PAPANCEA, in view of MYSTETSKYI discuss determining generative AI resources associated with generative AI models, (as in MYSTETSKYI [0004]), PAES, in view of PAPANCEA, in view of MYSTETSKYI does not explicitly teach:
the one or more routing restrictions define one or more of:
a preferred generative Al model;
a preferred type of generative Al model;
a preferred account for a generative Al model; and
a preferred region for a generative Al model.
However, in analogous art that similarly teaches determining a generative AI resources associated with generative AI models (as in MYSTETSKYI [0004]), DURVASULA teaches:
the one or more routing restrictions define one or more of: a preferred generative Al model; a preferred type of generative Al model ([0045] The generative AI models repository 400 may further include ML model to select generative AI model 460. In some embodiments, the ML model to select generative AI model 460 may select a generative AI model based on context and type of data (i.e., ML model restricts selection of a generative AI model to a “preferred” generative AI model based on a preferred “type” of data associated with the generative AI model));
a preferred account for a generative Al model; and a preferred region for a generative Al model (The prior art teaches “one or more of” the list of alternatives);
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the invention to have combined DURVASULA’s teaching of a ML model that restricts selection of generative AI models to particular ones associated with a particular type of data, with the combination of PAES, PAPANCEA, and MYSTETSKYI’s teaching of determining generative AI resources associated with AI models, to realize, with a reasonable expectation of success, a system that selects generative AI resources based on routing restrictions, as in PAES, PAPANCEA, and MYSTETSKYI, which further specify a generative AI model and type of data associated with the generative AI model, as in DURVASULA. A person having ordinary skill would have been motivated to make this combination to improve performance of the system by improving selection of the generative AI model (DURVASULA [0046]).
Regarding claims 19, and 29, they comprise limitations similar to those of claim 9, and are therefore rejected for similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
CAMERON et al. Patent No.: US 12,493,772 B1 discloses a repository storing existing machine learning models and metadata of the models including objectives, training data sources, and functional specifications.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MICHAEL W AYERS whose telephone number is (571)272-6420. The examiner can normally be reached M-F 8:30-5 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, Aimee Li can be reached at (571) 272-4169. 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.
/MICHAEL W AYERS/Primary Examiner, Art Unit 2195
Prosecution Insights