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 .
This communication is in response to Applicant’s Arguments filed 05/29/2026 regarding application 19/451,378 filed 01/16/2026. Claims 1, 3, 7, 14, and 24 are amended and hereby entered. Claims 26-28 are added and hereby entered. No claims are allowed
Response to Arguments
Applicant's arguments filed 5/29/2026 are fully considered but they are not persuasive.
Regarding 35 USC 101:
The applicant submits the claims recite a technical improvement under MPEP 2106.05(a) citing the claims are analogous to Desjardin, in which the claims resulted in a technical improvement to underlying technology outlined in the specification. However, in Desjardin, the improvement was directed to the underlying technology of machine learning, in which the problem of catastrophic forgetting was solved, improving performance of the machine learning technology itself. The instant application does not improve the underlying computers or machine learning technology as outlined in MPEP section 2106.05(a). Rather, it uses the computers and multiple machine learning models orchestrated together to perform the commonplace business practice of recruitment. See specification paragraph 5 stating, “The system thereby delivers a unified framework that improves throughput, consistency, and traceability across HR processes while minimizing human intervention”. This supports the analysis the improvement is related to the identified abstract idea, not the underlying computer or machine learning technology itself. The claims simply recite the use of general-purpose computing components and machine learning models to perform the abstract ideas, see MPEP 2106.05(f).
Further, the applicant submits the claims integrate the abstract idea into practical application because the claims recite additional limitations that reflect technological improvements. Specifically, the applicant cites three limitations regarding the ingestion of data, invoking agents and subsystems according to a predetermined orchestration sequence, and generation of a report, (see pages 18 and 19 of applicant’s remarks). However, these are not additional elements for consideration in Step 2A prong Two. They are part of the abstract ideas involving mental processes and certain methods of organizing human activity. For example, the ingestion of data, invoking of agents (wherein the outputs are evaluated against thresholds to determine next stages), and the generation of a report recite a mental process consistent with MPEP 2106.04(a)(2)(III) as a claim to “collecting information, analyzing it, and displaying results of collection and analysis”. Further, the orchestration sequencing and decision-making process based on threshold and logic recited in the limitations additionally describe the commonplace business method of recruiting, consistent with certain methods of organizing human activity outlined in MPEP section 2106.04(a)(2)(II)(B)&(C). Therefore, the applicant described additional limitations do not reflect technological improvements because they are part of the abstract idea, not additional elements for consideration in Step 2A Prong Two.
Further, the applicant submits the claims recite significantly more. Specifically, the applicant submits the claims recite a technological solution to a technological problem in which the specification recites, “Conventional hiring and human resource (HR) management systems involve several manual or semi-automated steps… Each stage typically operates in a siloed manner, requiring human intervention between transitions. Existing systems lack unified orchestration across these stages… This fragmentation leads to delays, inconsistent evaluations, and limited traceability and auditability of decisions”. However, this again is directed to a common place business method being applied on a general-purpose computer and not directed to improvements to the underlying computer components or machine learning technology. The technology is used as a tool to implement the abstract idea, see MPEP 2106.05(f). This also applies to the dependent claims that recite further embellishments of the abstract ideas.
Regarding 35 USC 103:
The applicant submits Colter discloses a sequential workflow rather than a parallel operation. However, see Colter paragraph 125 stating, “while processes or blocks are at times shown as being performed in series, these processes or blocks can instead be performed or implemented in parallel, or can be performed at different times”.
Further, the applicant submits Kaushik’s teaching of input analysis does not meet the claims because the claims require a fundamentally different architecture. However, the claims do not recite different architecture. They recite broad functional limitations of ingesting through different pipelines, time aligning outputs producing an anomaly score, normalizing outputs and associating timestamps evidence links with scores. For example, the breadth of “associating” encompasses a wide range of limitations in which almost any time, evidence, and scores being even remotely related to each other (i.e. “associated”) meets the claim. This broad recitation in of claims also applies to applicant’s arguments on page 29 of 31 regarding similar dependent claims 12 and 23.
Further, the applicant submits Colter does not teach “generating a unique interview link comprising a time-limited authentication token to initiate a candidate interview session. However, the functional limitation is very broad. In Colter, the meeting confirmation agent sends a calendar invite once a time is agreed upon. This meets the broad recitation of a unique interview link and “time-limited” authentication token to initiate an interview session.
Further, the applicant submits Colter does not teach evaluating configurable thresholds to determine next stage activation or termination. However, “threshold” and “next stage activation or termination” are interpreted broadly. This is taught by at least Colter’s AI agents that perform functions based on other outputs of other entities or agents. For example, the scheduler agent proposes meeting slots, and in the next stage, the meeting confirmation agent sends calendar invites once a time is agreed upon. The “threshold” could be completion of a previous stage or agreeing on a time. The “next stage activation or termination” could be scheduling the meeting.
Further, the applicant submits the examiner did not meet the requirement for motivation to combine. In response to applicant’s argument that there is no teaching, suggestion, or motivation to combine the references, the examiner recognizes that obviousness may be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so found either in the references themselves or in the knowledge generally available to one of ordinary skill in the art. See In re Fine, 837 F.2d 1071, 5 USPQ2d 1596 (Fed. Cir. 1988), In re Jones, 958 F.2d 347, 21 USPQ2d 1941 (Fed. Cir. 1992), and KSR International Co. v. Teleflex, Inc., 550 U.S. 398, 82 USPQ2d 1385 (2007). In this case, the examiner states, “one of ordinary skill would have been recognized the benefits of using different agents in a multistage recruiting system to assist in different portions of the process”, yielding predictable results, see Office action dated 3/23/2026. Therefore, obviousness is established by motivation to do so found in the knowledge available to one of ordinary skill. Additionally, in response to applicant's argument that Kaushik and Colter have incompatible architecture, the test for obviousness is not whether the features of a secondary reference may be bodily incorporated into the structure of the primary reference; nor is it that the claimed invention must be expressly suggested in any one or all of the references. Rather, the test is what the combined teachings of the references would have suggested to those of ordinary skill in the art. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981).
Regarding dependent claim arguments under 35 USC 103:
The applicant submits Thompson does not teach subtask failure detection and retry because Thompson’s error handling is fundamentally different (Claims 6, 20, and 25). However, the claims broadly recite detecting failed subtasks and repeating only detected failed subtasks. Although the applicant believes the architecture is different, the argument is not persuasive because regardless of architecture Thompson shows examples of performing these broad limitations.
The applicant submits Thompson does not teach monitoring workload metrics because Thompson teaches monitoring “the status of the plan” (Claim 9). However, the breadth of “workload metrics” is larger than “status of the plan” (i.e. a status is a specific metric)
The applicant submits Thompson does not teach the encryption features because Thompson recites a generic reference to a secure environment (Claim 8 and 21). However, the environment in Thompson includes a system that preforms encryption which is functionally required by the claims. Additionally, enforcing role-based control for data retrieval is already taught by Colter not Thompson. The encrypted data being Personally Identifiable information (PII) is nonfunctional, yet Colter still teaches PII.
The applicant submits Wong does not teach interview features because Wong teaches a selection of questions from a bank (Claims 3 and 17). However, Kaushik already teaches establishing a producer-consumer queue that stores question-response entries, not Wong. Further, selecting a question from a question bank is a form of “producing acknowledgements responsive to candidate inputs” as well as “generating, when the que is empty, a next interview question conditioned on prior candidate responses and the JD schema”. See at least Wong, Column 8 Lines 23-29, stating “The candidate's answer is evaluated in real time and the results is used to select the next question. With the adaptive selection of each question, the interview is more efficient and accurate”.
The applicant submits Colter does not teach the audit trail described by the limitations (Claims 10, 15, and 22). However, the blockchain technology taught by Colter achieves the generation of an audit trail described by the claims. Further, the applicant states the claims recite functional use of the unified profile and associated evidence. However, the applicant is arguing limitations that are not claimed. The functional limitations in the referenced claims simply recite maintaining a record with data elements and generating an interpretable audit trail linking decisions to data. See claims 10, 15, and 22 stating, “further comprising maintaining, for each report, a record comprising model versions, configuration snapshots, and processing steps used to generate the report, thereby enabling auditability and reproducibility” and “wherein the one or more hardware processors are configured to generate an interpretable audit trail linking each automated decision to its contributing data sources, timestamps, agent outputs, and confidence metrics, thereby enabling human reviewers to trace and validate end-to-end decision logic”.
The applicant submits Colter does not teach dynamic security policies, and references encryption keys and data retention durations (Claim 11). However, those limitations are intended results of the functional limitations in the claim, see MPEP 2111.04.
The applicant submits Kaushik does not teach parallel operation of agents (claims 2 and 16). However, it is Colter cited in independent claim 1 that teaches agents operating in parallel, not Kaushik. The dependent claim 2 is further describing the one or more type of agents, which is taught by Kaushik.
The applicant submits generic audio analysis or speech-to-text transcription is not “speaker diarization” (claims 4 and 18). However, “speaker diarization” is not explicitly defined in the specification. The specification discusses using “speaker diarization” as a part of processing audio, see paragraph 21 of the specification, “In some embodiments, processing the audio stream comprises voice activity detection, speaker diarization, transcript alignment to diarized segments, prosody extraction, and fluency metrics (e.g., words-per-minute, pause density, and burstiness)”. One of ordinary skill would interpret “speaker diarization” in this context as the speech to text taught in Kaushik. Further, the applicant provides similar arguments for the limitation of “gaze and face tracking”. Similarly, the broad recitation of “gaze and face tracking” is interpreted as the tracked facial analysis taught in Kaushik.
The applicant submits Kaushik does not teach storing evidentiary artifacts associated with each autonomous decision (claims 5 and 19). However, the data type being stored is nonfunctional descriptive material and does not carry patentable weight in the claims.
The applicant submits Kaushik does not disclose coordinating workflow stages (claim 7). However, Kaushik teaches all of the standard recruiting stages listed, (see at least Kaushik’s figures and paragraphs cited in the nonfinal rejection dated 3/23/2026). Additionally, paragraph 72 of Kaushik teaches each block can be combined, divided, or eliminated showing that each block can be independently deployable as recited in the claims.
The applicant submits Kaushik and Colter do not teach the same training mechanism as the claims because the claims require adjusting orchestration-level parameters while leaving model weights unchanged (claim 13). The applicant submits that Kaushik retrains the model itself which inherently alters model weights. However, the claims recite “ingesting recruiter feedback and hiring-outcome data to adjust rule weights, decision thresholds, or orchestration parameters over time without altering underlying model weights”. The positively recited step involves the ingestion of data to adjust various parameters. This is shown by the reference in which a hiring manager can provide feedback on the AI model decisions.
In response to applicant's argument that the examiner has combined an excessive number of references, reliance on a large number of references in a rejection does not, without more, weigh against the obviousness of the claimed invention. See In re Gorman, 933 F.2d 982, 18 USPQ2d 1885 (Fed. Cir. 1991). Additionally, in response to applicant’s argument that the examiner has not articulated a persuasive rationale for why one of ordinary skill would combine the references, the examiner recognizes that obviousness may be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so found either in the references themselves or in the knowledge generally available to one of ordinary skill in the art. See In re Fine, 837 F.2d 1071, 5 USPQ2d 1596 (Fed. Cir. 1988), In re Jones, 958 F.2d 347, 21 USPQ2d 1941 (Fed. Cir. 1992), and KSR International Co. v. Teleflex, Inc., 550 U.S. 398, 82 USPQ2d 1385 (2007). In this case, the examiner states, “one of ordinary skill would have recognized utilizing encryption would provide a strong security measure to help protect sensitive data in a recruiting system”, “one of ordinary skill would have recognized the benefits of monitoring metrics to promote a more robust system”, and “one of ordinary skill would have recognized the benefits using adaptive interview questions to better assess candidate skills”, yielding predictable results, see Office action dated 3/23/2026. Therefore, obviousness is established by motivation to do so found in the knowledge available to one of ordinary skill. Further, the examiner additionally cites KSR rationale A: Merely a combination of old elements.
Therefore, for the aforementioned reasons, the applicant’s arguments are unpersuasive, and the rejections under 35 USC 101 and 103 are maintained.
Claim Objections
Claim 7, 9, and 11 are objected to because of the following informalities:
The applicant deletes portions of claim 7 without using strikethrough notation. It is unclear whether the applicant intended to remove the limitation “dynamically adapting an execution order, retry logic, or data-flow routing of said stages based on prior stage outcomes, feedback signals, or processing latency thresholds”. For the purposes of compact prosecution, the limitation is included in claim 7. Claims 9 and 11 are objected to due to their dependency on claims 7. Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-28 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention. The limitations “state-aware execution”, “parent and child agents”, and “orchestration graphs” are not explicitly recited in the specification. Therefore, the claims fail to comply with the written description requirement. However, the specification does discuss various agent types as well as sequential and parallel decision nodes. For the purpose of compact prosecution, the applicant labeled limitation “(ii)” is interpreted to mean “coordinating sequential or parallel execution of agents or requirement nodes”.
Claims 2-13, 15-23, and 25-28 are rejected due to their dependency on the independent claims.
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 1-28 are rejected under 35 U.S.C. 101 because the claimed invention is directed to judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) with no practical application and without significantly more.
The claimed invention is directed to an abstract idea in that the instant application is directed to a mental process (See MPEP 2106.04(a)(2)(III)). The independent claims (1, 14 and 24) recite a method and systems to evaluate job and candidate data to make hiring recommendations based on the processed data. These claim elements are being interpreted as concepts performed in the human mind (including observation, evaluation, judgement, and opinion). Using job and candidate data to evaluate candidates during a hiring process can equivalently be achieved by human observation and evaluation of data. The claims recite an abstract idea consistent with the “mental process” grouping set forth in the MPEP 2106.04(a)(2)(III).
Additionally, the claimed invention is directed to an abstract idea in that the instant application is directed to certain methods of organizing human activity (see MPEP 2106.04(a)(2)(II)). The independent claims (1, 14 and 24) recite a method and systems to interview candidates for job roles in a recruiting process. These claim elements are being interpreted as managing personal behavior or relationships or interactions between people (including social activities, teaching, and following rules or instructions). Making decisions and conducting interviews in a hiring process recite an abstract idea consistent with the certain methods of organizing human activity grouping set forth in the MPEP 2106.04(a)(2)(II).
The instant application fails to integrate the judicial exception into a practical application because the instant application merely recites an “apply it” (or an equivalent) with the judicial exception, or merely includes instructions to implement an abstract idea. The instant application is directed towards a method and systems to implement the identified abstract idea of receiving information, processing information, and displaying the result of the analysis (i.e. processing job and candidate data to assess potential hires) and managing relationships between people (i.e. conducting interviews in a hiring process and the like) in a general computer environment. The claims do not include additional elements that integrate the judicial exception into practical application or amount to significantly more than the judicial exception. The independent claims recite the additional elements “a computer”, “hardware processors”, “machine learning models” and “analysis agents” These claim elements are recited at a high level of generality such that it amounts to no more than mere instructions to apply the exception using a general computer environment. The machines merely act as a modality to implement the abstract idea and are not indicative of integration into a practical application (i.e., the additional elements are simply used as a tool to perform the abstract idea), see MPEP 2106.05(f).
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed in Step 2A Prong Two analysis, the additional elements in the claims amount to no more than mere instructions to apply the exception using generic computer components. The same analysis applies here in 2B and does not provide an inventive concept.
In regards to the dependent claims, claims 2-13, 15-23, and 25-28 recite no new additional elements or new abstract ideas and do not impact analysis under 35 USC 101. The dependent claims simply further define the existing abstract ideas or use the additional elements to perform the abstract ideas.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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 1-2, 4-5, 7, 10-16, 18-19, and 22-24 are rejected under 35 U.S.C. 103 as being unpatentable over Kaushik (US 20250252403 A1) in view of Colter (US 20250053928 A1).
Regarding Claims 1, 14, and 24, Kaushik teaches:
A computer-implemented method for orchestrating and automating a multi-stage orchestration workflow associated with talent evaluation and using one or more machine learning models, wherein the method improves distributed execution and orchestration of computer infrastructure by efficiently orchestrating operations of multiple analysis agents while maintaining a shared context and unified schema across these agents, the method comprising, by one or more hardware processors: [see at least Kaushik: (Para 0015) “The computer program product a non-transitory computer readable medium having stored thereon computer executable instructions, which when executed by one or more processors, cause the one or more processors to conduct operations”, (Para 0059) “The AI model 110 may employ Natural Language Processing (NLP) techniques and machine learning models”]
receiving, via a network interface, job-requirement data for an open role; [see at least Kaushik: (Para 0035) “While hiring a candidate for an organization, a shortlisting process needs to be focused on various criteria outlined in a job description. The first step of the shortlisting process includes a thorough review of resumes of the candidates based on predefined criteria such as education, experience, and skills outlined in the job description”, (Para 0057) “In operation, the system 102 may be configured to retrieve a given job description, as described in FIG. 3B.”]
parsing the job-requirement data, wherein the job-requirement data is received in a non-standardized format, to generate a structured job description (JD) schema comprising prioritized skills, competency weightings, and decision thresholds, wherein the JD schema conforms to a standardized format regardless of the non-standardized format of the job-requirement data; [see at least Kaushik: (Para 0057) “The system 102 may be further configured to analyze, using the AI model 110, the job description to generate assessment data. The assessment data comprises at least one of: a plurality of assessment parameters, and a weightage score 314c associated with each of the plurality of assessment parameters, as described in FIG. 3B”, (Para 0059) “In an example, the system 102 may employ the AI model 110 to match the skills and keywords mentioned in the job description with those found in the candidate's profile or resume”, (Para 0061) “In an example, if the initial score for the resume of the candidate is above a threshold value, then the resume may be shortlisted”]
decomposing the JD schema into a plurality of requirement nodes representing different skills, competency dimensions, eligibility conditions, workflow parameters, or evaluation criteria; [see at least Kaushik: (Para 0058) “In an example, if the given job description includes requirements for “Python programming,” the AI model 110 will scan the resume data 314a from the database 314 to detect for the presence of the keyword “Python” in the resumes skills section. In another example, if the given job description includes requirements for “Data Analyst” role the AI model 110 may scan the resume data 314a to detect for the presence of the keyword like, but not limited to “data analysis,” and “data visualization.”]
receiving candidate resume data associated with a candidate; [see at least Kaushik: (Para 0055) “At 302, resume data may be retrieved”]
parsing the candidate resume data, wherein the candidate resume data is received in a non-standardized format, to extract structured fields including skills, employment history, education, and certifications, wherein the structured fields conform to the standardized format regardless of the non-standardized format of the candidate resume data; [see at least Kaushik: (Para 0035) “The first step of the shortlisting process includes a thorough review of resumes of the candidates based on predefined criteria such as education, experience, and skills outlined in the job description”, (Para 0056) “In an example, the resume of the candidate may include information, such as, but not limited to, personal details, educational background, professional experience, skills, achievements, and the like.”]
dispatching, by an orchestration engine implemented as a stateless microservices deployed on a container orchestration platform, the structured fields of resume data to [the limitation recites dispatching structured fields or resume data with software that processes each request on its own; see at least Kaushik: (Para 0010) “According to additional system embodiments, the processor is configured to retrieve a resume of the candidate, analyze, using the trained AI model, the resume of the candidate to generate an initial score; and shortlist the candidate for the virtual interaction based on the initial score”, (Para 0035) “The first step of the shortlisting process includes a thorough review of resumes of the candidates based on predefined criteria such as education, experience, and skills outlined in the job description”, (Para 0059) “In operation, the system 102 may be configured to provide, as an input, the resume data”]
aggregating, by a result aggregator, outputs from the plurality of analysis agents into a unified candidate profile dataset conforming to a shared schema, wherein the shared schema enables consistency across the plurality of analysis agents; [Wherein clause describes intended results, see MPEP 2111.04, see at least Kaushik: (Para 0036) “The disclosed system 102 may be configured to generate, using a trained AI model 110, an aggregated score for assessing skills of candidates, thereby providing optimum candidate shortlisting, and mitigating bias”]
generating a structured candidate summary based on the unified candidate profile dataset and the JD schema; [see at least Kaushik: (Para 0010) “According to additional system embodiments, the processor is configured to retrieve a resume of the candidate, analyze, using the trained AI model, the resume of the candidate to generate an initial score”, (Para 0036) “The disclosed system 102 may be configured to generate, using a trained AI model 110, an aggregated score for assessing skills of candidates, thereby providing optimum candidate shortlisting, and mitigating bias”, (Para 0053) “Further, the system 102 may be configured to display the skills of the candidate on the online platform 104, thereby employing a visual representation of the score of the candidate to identify patterns or trends at a glance.”]
conducting, by a plurality of interview subsystems using one or more machine learning models, a dynamic automated interview with the candidate; [see at least Kaushik: (Para 0061) “In an operation, the system 102 may employ the virtual interaction to assess the skills of the candidate from the list of shortlisted candidates. The virtual interaction may refer to virtual interview that takes place on the online platform 104”, (Para 0003) “During video interviews, the candidate may or may not interact with a live interviewer”, (Para 0091) “In an example, the system 102 may retrieve the text data. Further, the trained AI model 110 may employ automatic speech recognition system to transcribe the virtual interaction. The system 102 may be further configured to apply NLP techniques and machine learning models to analyze the transcription text for correctness, relevance, and communication skills.”]
ingesting, during the dynamic automated interview, audio and video streams and processing the audio and video streams in parallel pipelines, [see at least Kaushik: (Para 0013) “According to additional system embodiments, the response data further comprises at least one of: user input data, video data, or audio data”, (Para 0062) “The user device 106 may have internal built in sensors, or communicatively coupled to one or more sensors (such as, but not limited to cameras and microphones) to receive the response data (such as the video data, and the audio data)”]
comprising at least a speaker diarization pipeline and lip-audio synchronization analysis pipeline, and time-aligning outputs of the parallel pipelines on a shared timeline to produce an anomaly score, [see at least Kaushik: (Para 0085) “Further, the system 102 may generate, using AI model 110, a transcript for the audio portion corresponding to each video frame using speech to text techniques”, (Para 0090) “For example, if a candidate's video interview includes moments of smiling and laughter (positive facial sentiment), a confident and expressive tone in their voice (positive audio sentiment), and coherent and relevant responses in the transcription (positive text analysis) such positive factors may contribute to higher skill scores in their respective categories”]
wherein time-aligning the outputs comprises normalizing outputs from the audio and video pipelines into a queryable timeline and [see at least Kaushik: (Para 0091) “In an example, the system 102 may retrieve the text data. Further, the trained AI model 110 may employ automatic speech recognition system to transcribe the virtual interaction. The system 102 may be further configured to apply NLP techniques and machine learning models to analyze the transcription text for correctness, relevance, and communication skills”]
associating timestamped evidence links to the anomaly score [see at least Kaushik: (Para 0084) “For example, if the response data is video data, the response segment may correspond to a sequence of video frames, and audio portion corresponding to each video frame. In such a scenario, the AI model 110 may be configured to analyze facial features in the sequence of video frames, and corresponding audio features. In an example, the facial analysis may include, but is not limited to smile and facial expression”, (Para 0085) “Further, the system 102 may generate, using AI model 110, a transcript for the audio portion corresponding to each video frame using speech to text techniques”, (Para 0086) “At 406, a score may be generated. In an embodiment, the system 102 may be configured to generate, using the trained AI model 110, the score for the response data based on the assessment data. In an embodiment the system 102 is further configured to analyze, using the AI model 110, each of the one or more response segments based on the corresponding response features. The system 102 is further configured to generate, using the AI model 110, a segment score for each of the one or more response segments based on the analysis. Thereafter, the system 102 is further configured to generate, using the trained AI model, the score for the response data based on the segment score for each of the one or more response segments. In an example, the system 102 may analyze the response data to generate the score. Such an analysis may include, but is not limited to facial analysis, audio analysis, and text analysis. The segment score may refer to score calculated for each category (such as, but not limited to, facial sentiment score, audio sentiment score, and text analysis score)”, (Para 0088) “The AI model 110 may be employed to extract facial landmarks from each frame of the video data. Such facial landmarks include for example, not limited to facial expressions, head movements, eye movements, eyebrow positions, mouth shape, and the like. Thereafter, the system 102 may estimate, using the AI model 110, an emotional state of the candidate based on the extracted facial feature and generate the facial sentiment score associated therewith.”]
fusing the unified candidate profile dataset, the structured candidate summary, interview transcripts, and the anomaly score to compute one or more evaluation scores relative to the competency weightings and decision thresholds; [see at least Kaushik: (Para 0010) “According to additional system embodiments, the processor is configured to retrieve a resume of the candidate, analyze, using the trained AI model, the resume of the candidate to generate an initial score; and shortlist the candidate for the virtual interaction based on the initial score”, (Para 0012) “According to additional system embodiments, the processor is further configured to obtain the user input based on the list of the shortlisted candidates and the resume of the candidate”, (Para 0015) “ The operation comprises retrieving assessment data, wherein the assessment data comprises at least one of: a plurality of assessment parameters, and a weightage score associated with each of the plurality of assessment parameters. The operation further comprises retrieving response data of a candidate associated with the assessment data. Further, the operation comprises generating, using a trained AI model, a score for the response data based on the assessment data”, (Para 0064) “ In an example, if the score of the candidate is above a threshold value, then the candidate may be considered suitable for hiring”]
However, Kaushik does not teach but Colter does teach:
a plurality of analysis agents operating in parallel; [see at least Colter: (Para 0055) “We can automate major recruiting workflows using AI agents that collaborate to accomplish tasks”]
wherein the orchestration engine: (i) provides structured inputs to the plurality of analysis agents; [see at least Colter: (Para 0059) “ Job FoW Classification Agent: Takes the parsed JD, and leverages the LLM to determine the FoW for the JD. Job Onboarder Agent: This agent would enter structured job data into a database, ensuring that job postings are properly stored and accessible”]
(ii) coordinates state-aware execution across multiple parent and child agents operating within an orchestration graph comprising interconnected execution nodes representing workflow stages and requirement nodes, wherein the orchestration graph is stored as a directed dependency graph defining parent-child execution dependencies and downstream activation conditions between workflow nodes; [The limitation recites coordinating sequential or parallel execution of dependent agents or requirement nodes, see 112a rejection; see at least Colter: (Figures 6-8), (Para 0047) “In a given example, the AI core is Perry Assistant, a collection of AI agents powered by Microsoft's AutoGen framework. Other examples such as AutoGPT, OpenAgents, OpenAl, GPTs, BabyAGI, Llama, Mistral, Bard, and other AI agents are similarly suitable. These agents automate key steps in the recruiting workflow”, (Para 0106) “For example, the event stream 804 may store the event and event related data (e.g., actor identifier, organization universal identifier, job universal identifier, hash value, job state, token universal uniform resource locator, user identifiers, the action, verification status, etc.) in the database associated with the event stream, such as in a table, graph, array, dictionary, or other data structure”, (Para 0125) “Also, while processes or blocks are at times shown as being performed in series, these processes or blocks can instead be performed or implemented in parallel, or can be performed at different times”]
(iii) maintains a shared context and unified schema across the plurality of analysis agents by writing structured outputs generated by upstream agents into a shared orchestration schema accessible to downstream execution nodes, such that downstream agents dynamically adapt execution based on outputs generated by upstream agents without reprocessing previously evaluated data: and [see at least Colter: (Figures 6-9), (Para 0062-0067) “Job Parser and Resume Parser Agents use LLMs to extract and classify key details from job posts and candidate resumes/profiles. Matcher Agent uses semantic search, keywords, and other signals to automatically suggest potential matches between candidates and open roles. Recommender Agent proactively identifies and suggests candidates that may be a good fit for a new job posting based on previous matches and peer candidates…The Candidate Intake Agent would facilitate the initial signup process and profile creation. The Profile Parser would extract key details to populate the profile. The Profile Writer, Skill Assessor and Profile Coach could all help enhance the candidate's profile to best highlight their abilities and experience…]
(iv) time-aligns outputs from asynchronous pipelines into a coherent, queryable record: and [see at least Colter: (Para 0065) “Metrics Collector Agent aggregates data on job posts, candidates, hiring performance, user engagement and more. Stores data in Postgres. Natural Language Query Agent allows users to ask questions in plain English (e.g. “How many developer jobs were posted this month?”) and converts to SQL query to generate reports visually. Insights Agent monitors data and proactively surfaces key trends and patterns to users (e.g. Python skills rising in demand)”]
(v) resolves execution dependencies between interconnected execution nodes prior to downstream agent activation: [see at least Colter: (Para 0063) “Scheduler Agent—interacts with HM and candidate to find a mutually available time. Calendar Agent—checks availability in calendars and proposes meeting slots. Meeting Confirmation Agent—sends calendar invites once time is agreed upon”]
generating a unique interview link comprising a time-limited authentication token to initiate a candidate interview session; [see at least Colter: (Para 0063) “Scheduler Agent—interacts with HM and candidate to find a mutually available time. Calendar Agent—checks availability in calendars and proposes meeting slots. Meeting Confirmation Agent—sends calendar invites once time is agreed upon.”]
invoking, by the orchestration engine, the plurality of analysis agents and the interview subsystems according to a predetermined orchestration sequence, wherein outputs of each interview subsystem are evaluated against configurable thresholds to determine next-stage activation or termination, and wherein the orchestration engine dynamically adapts execution order, retry logic, data-flow routing, and orchestration-state transitions based on prior stage outcomes, feedback signals, dependency-resolution conditions, or processing latency thresholds; [see at least Colter: (Figures 5-8), (Para 0062) “Job Parser and Resume Parser Agents use LLMs to extract and classify key details from job posts and candidate resumes/profiles. Matcher Agent uses semantic search, keywords, and other signals to automatically suggest potential matches between candidates and open roles. Recommender Agent proactively identifies and suggests candidates that may be a good fit for a new job posting based on previous matches and peer candidates”, (Para 0063) “scheduler Agent—interacts with HM and candidate to find a mutually available time. Calendar Agent—checks availability in calendars and proposes meeting slots. Meeting Confirmation Agent sends calendar invites once time is agreed upon”]
and generating a report based on the unified candidate profile dataset and recording a trace record comprising model versions, configuration snapshots, interview subsystem inputs, outputs, decision paths taken, and associated timestamps for each relevant interview subsystem, thereby enabling auditability and reproducibility of each step of the evaluation and decision-making process. [The contents of the dataset and trace record do not functionally impact generating a report and recording a trace record; see at least Colter: (Para 0090) “Data-driven analytics and reports provide insights for better hiring”, (Para 0027) “blockchain technology (or distributed ledger technology) generally refers to a growing list of digital records (e.g., blocks) that are connected by a cryptographic hash of the previous block. Each block may further include a timestamp and other data related to an action that has occurred”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage artificial intelligence recruiting work flow (Kaushik) with multiagent system for different tasks (Colter). One of ordinary skill would have recognized the benefits of using different agents in a multistage recruiting system to assist in different portions of the process. Combining these elements would have yielded predictable results to one of ordinary skill.
Regarding Claims 2 and 16, the combination of Kaushik and Colter teach the limitations of claim 1, Kaushik further teaches:
wherein the analysis agents operated in parallel comprise one or more of: a skill analysis agent, a job stability analysis agent, a certification analysis agent, a gap analysis agent, an experience analysis agent, and a relevance analysis agent. [see at least Kaushik: (Para 0064) “At 312, a score may be generated. In an embodiment, the system 102 may be configured to generate, using the AI model 110, the score to assess the skills”]
Regarding Claims 4 and 18, the combination of Kaushik and Colter teach the limitations of claim 1, Kaushik further teaches:
wherein the parallel pipelines comprise at least two of a speaker diarization pipeline, a lip-audio synchronization analysis pipeline, a gaze and face tracking pipeline, and an object detection pipeline. [see at least Kaushik: (Para 0084) “In an embodiment the system 102 may be further configured to analyze, using the trained AI model 110, each of the one or more response segments based on the corresponding response features. Such an analysis may include, but is not limited to facial analysis, audio analysis, and text analysis based on the type of the response data”, (Para 0085) “Further, the system 102 may generate, using AI model 110, a transcript for the audio portion corresponding to each video frame using speech to text techniques.”]
Regarding Claims 5 and 19, the combination of Kaushik and Colter teach the limitations of claim 1, Kaushik further teaches:
autonomously determining whether to advance, reject, or reassign the candidate to an alternate role based on the one or more evaluation scores and the decision thresholds; [see at least Kaushik: (Para 0064) “In an example, if the score of the candidate is above a threshold value, then the candidate may be considered suitable for hiring. On the contrary, if the score of the candidate is below the threshold value, then the candidate may not be considered suitable for hiring”]
and storing, in a data store, the autonomous determination together with evidentiary artifacts comprising time-coded media references, transcripts, and agent outputs [(Para 0040) “The memory 204 may be further configured to store the assessment data, response data, and resume data”, (Para 0053) “Further, the system 102 may be configured to display the skills of the candidate on the online platform 104, thereby employing a visual representation of the score of the candidate to identify patterns or trends at a glance”]
Regarding Claim 7, the combination of Kaushik and Colter teach the limitations of claim 1, Kaushik further teaches:
further comprising: coordinating a plurality of human resources workflow stages comprising job-description generation, resume screening, interview scheduling, automated interview execution, and evaluation reporting, [see at least Kaushik: (Figures 3A-3B and Figure 4), (Para 0059) “the system 102 may employ the AI model 110 to match the skills and keywords mentioned in the job description with those found in the candidate's profile or resume”, (Para 0062) “ At 308, one or more questions may be displayed. Once the virtual interaction is initiated for the shortlisted candidates, the system 102 may be configured to display the one or more questions on the web page to assess the skills of the candidate”, (Para 0064) “At 312, a score may be generated. In an embodiment, the system 102 may be configured to generate, using the AI model 110, the score to assess the skills of the candidate based on the response data, as described in FIG. 4”]
each stage being implemented as an independently deployable service communicating through standardized orchestration messages; [see at least Kaushik: (Figures 3A-3B and Figure 4), (Para 0072) “The exemplary operations illustrated in block diagram 400 may start at 402 and may be performed by any computing system, apparatus, or device, such as by the system 102 of FIG. 1 or the processor 202 of FIG. 2A. Although illustrated with discrete blocks, the exemplary operations associated with one or more blocks of the block diagram 300b may be divided into additional blocks, combined into fewer blocks, or eliminated, depending on the particular implementation”]
dynamically adapting an execution order, retry logic, or data-flow routing of said stages based on prior stage outcomes, feedback signals, or processing latency thresholds. [see at least Kaushik: (Figures 3A-3B and Figure 4), (Para 0061) “In an example, if the initial score for the resume of the candidate is above a threshold value, then the resume may be shortlisted. Thereafter, the system 102 may be configured to add the resume of the candidate to the list of shortlisted candidates and initiate the virtual interaction”]
Regarding Claims 10 and 22, the combination of Kaushik and Colter teach the limitations of claim 1
While Kaushik teaches a multistage artificial intelligence recruiting workflow, it does not explicitly teach but Colter does teach:
further comprising maintaining, for each report, a record comprising model versions, configuration snapshots, and processing steps used to generate the report, thereby enabling auditability and reproducibility. [The limitations recite maintain a report and record. The data within the report is nonfunctional descriptive material that does not carry patentable weight in the claims; see at least Colter: (Para 0065) “Metrics Collector Agent aggregates data on job posts, candidates, hiring performance, user engagement and more. Stores data in Postgres. Natural Language Query Agent allows users to ask questions in plain English (e.g. “How many developer jobs were posted this month?”) and converts to SQL query to generate reports visually”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage artificial intelligence recruiting work flow (Kaushik) with a report record (Colter). The invention is merely a combination of old elements, and one of ordinary skill would have recognized keeping a reportable record of data would allow for insights into decision making. One of ordinary skill would have recognized the results of the combination predictable.
Regarding Claim 11, the combination of Kaushik and Colter teach the limitations of claim 7
While Kaushik teaches a multistage artificial intelligence recruiting workflow, it does not explicitly teach but Colter does teach:
further comprising enforcing dynamic security and compliance policies based on context, such that access permissions, encryption keys, or data-retention durations are automatically varied according to workflow stage, data sensitivity, or jurisdictional regulations. [see at least Colter: (Para 0021) “The mutable data may be stored as metadata associated with the NFT, such as qualifications, certifications, current employment information, past employment information, a user's contact information, or other user information. In some implementations, the metadata of the NFT may include a link to a uniform resource identifier (URI) or uniform resource locator (URL) pointing to the mutable data, enabling the mutable data to be stored off-chain in a permissioned database (e.g., where only certain, permissioned actors have access to such information) to preserve data security of Personally Identifiable Information (PII).”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage artificial intelligence recruiting work flow (Kaushik) with security measures (Colter). The invention is merely a combination of old elements, and one of ordinary skill would have recognized including security measures would help protect sensitive data in a recruiting system. One of ordinary skill would have recognized the results of the combination predictable.
Regarding Claims 12 and 23, the combination of Kaushik and Colter teach the limitations of claim 1, Kaushik further teaches:
further comprising: normalizing outputs from the audio and video pipelines, resume analytics, and JD parsing into a queryable timeline; [see at least Kaushik: (Para 0069) “At 322, output may be displayed. In an embodiment, the system 102 may be configured to display the list of shortlisted candidates and the initial score, as described in FIG. 2C. In an example, the system 102 may be configured to display score of the shortlisted candidates and resume of the candidates on the user interface for easy access. Such visual representation may be helpful for the hiring manager or professional to determine whether or not a deserving candidate is assessed incorrectly due to the assigned initial score”]
However, Kaushik does not each but Colter does teach:
and associating timestamped evidence links to each evaluation score and decision. [See at least Colter: (Para 0027) “blockchain technology (or distributed ledger technology) generally refers to a growing list of digital records (e.g., blocks) that are connected by a cryptographic hash of the previous block. Each block may further include a timestamp and other data related to an action that has occurred”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage artificial intelligence recruiting work flow (Kaushik) timestamped links (Colter). The invention is merely a combination of old elements, and one of ordinary skill would have recognized including timestamped records would allow for traceability in the system. One of ordinary skill would have recognized the results of the combination predictable.
Regarding Claim 13, the combination of Kaushik and Colter teach the limitations of claim 1, Kaushik further teaches:
further comprising ingesting recruiter feedback and hiring-outcome data as reinforcement signals to adjust rule weights, decision thresholds, or orchestration parameters over time without altering underlying model weights, thereby improving consistency and alignment with organizational hiring goals. [see at least Kaushik: (Para 0070) “In such an example, the hiring manager may provide a feedback using the user input to identify areas where the AI model 110 might have misjudged candidates. In an example, the system 102 may receive feedback on candidates who were shortlisted and rejected based on the generated initial score. The system 102 may be configured to retrain the AI model 110 based on the received feedback, thereby optimizing the assessment process to evaluate the skills of the candidates in an efficient manner”]
Regarding Claim 15, the combination of Kaushik and Colter teach the limitations of claim 14,
While Kaushik teaches a multistage artificial intelligence recruiting workflow, it does not explicitly teach but Colter does teach:
wherein the one or more hardware processors are configured to generate an interpretable audit trail linking each automated decision to its contributing data sources, timestamps, agent outputs, and confidence metrics, thereby enabling human reviewers to trace and validate end-to-end decision logic. [See at least Colter: (Para 0027) “blockchain technology (or distributed ledger technology) generally refers to a growing list of digital records (e.g., blocks) that are connected by a cryptographic hash of the previous block. Each block may further include a timestamp and other data related to an action that has occurred”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage artificial intelligence recruiting work flow (Kaushik) with an audit trail (Colter). The invention is merely a combination of old elements, and one of ordinary skill would have recognized keeping a linked trail of data would allow for traceable insights into decision making and actions. One of ordinary skill would have recognized the results of the combination predictable.
Regarding Claim 26, the combination of Kaushik and Colter teach the limitations of claim 1.
While Kaushik teaches a multistage artificial intelligence recruiting workflow, it does not explicitly teach but Colter does teach:
wherein downstream execution nodes are activated only after dependency-resolution conditions associated with upstream orchestration graph nodes are satisfied. [see at least Colter: (Figures 5-8), “Meeting Confirmation Agent—sends calendar invites once time is agreed upon”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage artificial intelligence recruiting work flow (Kaushik) with dependent workflows (Colter). The invention is merely a combination of old elements, and one of ordinary skill would have recognized activating downstream executions after upstream conditions are satisfied would result in a consistent traceable workflow, yielding predictable results. Additionally, one of ordinary skill would have recognized the benefits of dependent workflows for efficient automation of tasks, see at least Kaushik paragraph 36, “To overcome these challenges, the present disclosure provides a comprehensive method and system associated with an automated assessment process that evaluates the skills of the candidates in an efficient manner”.
Regarding Claim 27, the combination of Kaushik and Colter teach the limitations of claim 1.
While Kaushik teaches a multistage artificial intelligence recruiting workflow, it does not explicitly teach but Colter does teach:
wherein the orchestration engine updates a shared orchestration-state store using outputs generated by upstream analytical agents and provides the updated orchestration-state store to downstream analytical agents. [see at least Colter: (Para 0059) “Job Onboarder Agent: This agent would enter structured job data into a database, ensuring that job postings are properly stored and accessible. Job Board Agent: This agent would post jobs to job boards”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage artificial intelligence recruiting work flow (Kaushik) with shared data for workflows (Colter). The invention is merely a combination of old elements, and one of ordinary skill would have recognized updating a data and state in a workflow would result in a consistent traceable workflow, yielding predictable results. Additionally, one of ordinary skill would have recognized the benefits of workflows with shared states for efficient automation of tasks, see at least Kaushik paragraph 36, “To overcome these challenges, the present disclosure provides a comprehensive method and system associated with an automated assessment process that evaluates the skills of the candidates in an efficient manner”.
Regarding Claim 28, the combination of Kaushik and Colter teach the limitations of claim 16. Kaushik further teaches:
wherein outputs generated by the parallel audio and video pipelines are normalized [see at least Kaushik: (Para 0085) “Further, the system 102 may generate, using AI model 110, a transcript for the audio portion corresponding to each video frame using speech to text techniques”]
However, Kaushik does not teach but Colter does teach:
into a timestamp-indexed orchestration timeline prior to anomaly-score generation. [see at least Colter: (Para 0027) “blockchain technology (or distributed ledger technology) generally refers to a growing list of digital records (e.g., blocks) that are connected by a cryptographic hash of the previous block. Each block may further include a timestamp and other data related to an action that has occurred”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine normalization of audio and video data (Kaushik) with normalization into to a timestamp indexed timeline (Colter). It would have been obvious to one of ordinary skill to store the normalized audio and video using a timestamp-indexed orchestration timeline (blockchain) for verifiable, queryable, and timestamped record keeping, See at least Colter Paragraph 27, “Each block may further include a timestamp and other data related to an action that has occurred. The timestamp may indicate (or otherwise prove) that data included in the block was present or otherwise created at that moment in time”.
Claims 6, 8-9, 20-21, and 25 are rejected under 35 U.S.C. 103 as being unpatentable over Kaushik (US 20250252403 A1) in view of Colter (US 20250053928 A1) in further view of Thompson (US 20250371317 A1).
Regarding Claims 6, 20, and 25, the combination of Kaushik and Colter teach the limitations of claim 1
While the combination teaches a multistage multiagent artificial intelligence recruiting system, it does not explicitly teach but Thompson does teach:
further comprising: detecting failed sub-tasks while dispatching the structured fields of resume data to the plurality of analysis agents; [see at least Thompson: (Para 0314-0315) “The task completion sub-process 1010 can verify, or confirm that the task or the sub-tasks have been successfully performed, achieved, or satisfied by the automated agent or other agents using, e.g., a completion criterion, a verification criterion, or other criteria…. The error handling function can detect, identify, or classify the errors, the exceptions, or the failures”]
and repeating only the detected failed sub-tasks, thereby improving throughput and reducing latency of the analysis agents. [see at least Thompson: (Para 0315) “For example, if an output of the execution 1008 or result aggregation deviates from an expected output by more than a threshold amount, the error handling function can skip the response formulation or modify the response formulation to, e.g., request more input from the user and/or obtain more context data”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage multiagent artificial intelligence recruiting work flow (Kaushik and Colter) with the detection and execution of failed sub-tasks (Thompson). The invention is merely a combination of old elements, and one of ordinary skill would have recognized detecting and repeating failed sub-tasks would be essential to execute the workflow of a multistage multiagent system. One of ordinary skill would have recognized the results of the combination predictable.
Regarding Claims 8 and 21, the combination of Kaushik and Colter teach the limitations of claim 1,
While Kaushik teaches a multistage artificial intelligence recruiting workflow, it does not explicitly teach but Colter does teach:
enforcing role-based access control for data retrieval. [see at least Colter: (Para 0021) “enabling the mutable data to be stored off-chain in a permissioned database (e.g., where only certain, permissioned actors have access to such information) to preserve data security of Personally Identifiable Information (PII)”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage multiagent artificial intelligence recruiting work flow (Kaushik) with security measures (Colter). The invention is merely a combination of old elements, and one of ordinary skill would have recognized including role-based access would help protect sensitive data in a recruiting system. One of ordinary skill would have recognized the results of the combination predictable.
While the combination of Kaushik and Colter teach enforcing role-based access, they do not explicitly teach encryption. However, Thompson does teach:
further comprising encrypting personally identifiable information and [see at least Thompson: (Para 0069) “In the example of FIG. 1, the components of the computing system 100 are implemented using at least one application server or server cluster, which can include a secure environment (e.g., secure enclave, encryption system, etc.) for the processing of data”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage artificial intelligence recruiting work flow (Kaushik and Colter) with encryption (Thompson). The invention is merely a combination of old elements, and one of ordinary skill would have recognized utilizing encryption would provide a strong security measure to help protect sensitive data in a recruiting system. One of ordinary skill would have recognized the results of the combination predictable.
Regarding Claim 9, the combination of Kaushik and Colter teach the limitations of claim 7.
While the combination teaches a multistage multiagent artificial intelligence recruiting system, it does not explicitly teach but Thompson does teach:
further comprising monitoring workload metrics across the plurality of workflow stages and [see at least Thompson: (Para 0131) “At block 310, the adaptive machine learning-based orchestrator 216 can monitor the status of the plan generated or updated at block 308”]
automatically provisioning, scaling, or re-routing processing nodes through container orchestration policies to maintain throughput and latency targets. [see at least Thompson: (Para 0199) “Embodiments of the architecture 400 are data-driven. For example, the architecture 400 can include a small service that is deployed and acts as a factory for agent instances. Since the agents maintain their state in memory storage, any agent instance can be spawned to take over, which helps for horizontal scaling and fault tolerance”, (Para 0231) “The orchestration observer agent 508 can be an agent that can monitor, evaluate, or validate the orchestration of the actions performed by the orchestration agent 506. The orchestration observer agent 508 can use one or more AI services or tools to query policies or otherwise perform the observation functions. The orchestration observer agent 508 can communicate with the orchestration agent 506 via, e.g., the asynchronous distributed coordination 460”, (Para 0506) “AI model service 1590 can include a monitoring service that periodically generates, publishes, or broadcasts latency and/or other performance metrics associated with the models. For example, AI model service 1590 can provide a set of APIs that can be used by an agent or agent system to obtain performance metrics for large language models and/or other machine learning models”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage multiagent artificial intelligence recruiting work flow (Kaushik and Colter) with monitoring workloads and making adjustments (Thompson). The invention is merely a combination of old elements, and one of ordinary skill would have recognized the benefits of monitoring metrics to promote a more robust system. One of ordinary skill would have recognized the results of the combination predictable.
Claims 3 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Kaushik (US 20250252403 A1) in view of Colter (US 20250053928 A1) in further view of Wong (US 11238411 B1).
Regarding Claims 3 and 17, the combination of Kaushik and Colter teach the limitations of claim 1. Kaushik further teaches:
further comprising conducting the dynamic automated interview with the candidate by: establishing a producer-consumer queue that stores question-acknowledgement-response entries; [see at least Kaushik: (Para 0061) “At 306, a virtual interaction may be initiated. In an embodiment, the system 102 may be configured to initiate the virtual interaction for the candidate”, (Para 0063) “At 310, response data may be received. In an embodiment, the system 102 may be configured to retrieve the response data of a candidate associated with the one or more questions”]
While the combination of Kaushik and Colter teach a multistage multiagent artificial intelligence recruiting workflow, it does not explicitly teach but Wong does teach:
producing acknowledgements responsive to candidate inputs; and generating, when the queue is empty, a next interview question conditioned on prior candidate responses and the JD schema. [see at least Wong: (Column 8, lines 23-29) “At step 332, an adaptive interview is conducted. In one embodiment, the adaptive interview selects each question from the question bank adaptively. The candidate's answer is evaluated in real time and the results is used to select the next question. With the adaptive selection of each question, the interview is more efficient and accurate”]
Further, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the multistage multiagent artificial intelligence recruiting work flow (Kaushik and Colter) with an adaptive interview process (Wong). The invention is merely a combination of old elements, and one of ordinary skill would have recognized the benefits using adaptive interview questions to better assess candidate skills. One of ordinary skill would have recognized the results of the combination predictable.
Conclusion
Pertinent art not relied upon:
NPL Wallaroo.ai: Working Smarter with Machine Learing Model Chains. Wallaroo.ai teaches coordinated orchestration of multiple machine learning models.
Bolton, (US 20130226578 A1): Asynchronous video interview system. Bolton discusses an adaptive analytical virtual interview system used in recruiting processes
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. 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 nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Examiner Benjamin Truong, whose telephone number is 703-756-5883. The examiner can normally be reached on Monday-Friday from 9 am to 5 pm (EST).
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Nathan Uber SPE can be reached on 571-270-3923. 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.
/B.L.T. /Examiner, Art Unit 3687
/NATHAN C UBER/Supervisory Patent Examiner, Art Unit 3626
Prosecution Insights