DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
Claims 1-20 were previously pending and subject to a non-final rejection dated January 28, 2026. In Response, submitted April 17, 2026, claims 1, 3, 4, 6, 9, 11, 12, 14, 17, and 19 were amended. Therefore, claims 1-20 are currently pending and subject to the following final rejection.
Response to Arguments
Applicant’s remarks on Page 10 of the Response regarding the previous objection of the claims have been fully considered and the objections are withdrawn in light of the amended claims.
Applicant’s remarks on Pages 10-13 of the Response, regarding the previous rejection of the claims under 35 U.S.C. 101, have been fully considered and are not found persuasive.
On Pages 10-11 of the Response, Applicant argues “The technical problem addressed by the amended claims is described in paragraph [0009] of the specification: traditional contact centers employed agents working in physical proximity to supervisors, but pandemic-induced changes moved operations to remote locations (agents' homes, public places). Proper supervision of remotely-operating agents became an ongoing and challenging problem because agents have access to customer data that can be compromised-unauthorized people may gain access to the agent's session, agents may work from public locations, agents may misuse data, or agents may be part of malicious groups. The specification states that ‘[t]imely identifying the above-described and other similar situations and protecting customer data from misuse, e.g., to implement a “zero-trust call center” remains an important and outstanding problem.’ (Specification, par. [0009].) The amended claims address this technical problem by providing a specific ML-based architecture-the cascaded anomaly detection/evaluation pipeline with intermediate feature vectors and security trigger similarity matching-that enables automated real-time security monitoring of remote agents, which was not previously possible through manual supervision alone. This is a technical improvement to the functioning of the contact center's computer security system, not merely organizing human activity. The amended claims are integrated into a practical application under Step 2A Prong Two. The amendments to claims 1, 9, and 17 now recite a specific technical architecture that goes far beyond generic ML processing. Specifically, the amended claims recite a cascaded two-model ML architecture where an anomaly detection ML model first processes feature vectors to detect anomalies, and then an anomaly evaluation ML model processes intermediate feature vectors generated by the anomaly detection ML model-not raw data or final outputs, but the detection model's internal representations. This creates an architectural coupling between the two models that is a specific technical implementation, not merely invoking ML generically.”
Examiner notes, the stated problems with “Proper supervision of remotely-operating agents”, “agents hav[ing] access to customer data that can be compromised-unauthorized people may gain access to the agent's session, agents may work from public locations, agents may misuse data, or agents may be part of malicious groups”, “‘[t]imely identifying the above-described and other similar situations and protecting customer data from misuse, e.g., to implement a ‘zero-trust call center’” are not inherently technical in nature. Further the stated fact that the solution is not found in previously identified “manual supervision alone”, does not preclude the claims from reciting an abstract idea. In the instant case, as discussed further in the detailed rejection below, it is maintained that (though additional elements are recited) the claims recite an abstract idea at Step 2A, Prong One. Therefore, analysis must proceed to evaluating the recited additional elements.
Examiner further notes, as discussed further in the detailed rejection below, insofar as they are claimed the “cascaded anomaly detection/evaluation pipeline with intermediate feature vectors and security trigger similarity matching-that enables … real-time security monitoring of remote agents” are recitations of the abstract ideas performed by additional elements such as the anomaly detection ML model and the anomaly evaluation ML model. The “automated” aspect appears to merely be directed to the recitation of “one or more processing devices” being used as merely a tool (i.e., “apply it”) to perform many of the abstract ideas recited, and does not constitute any technical advancement in and of itself. Regarding the “specific ML-based architecture”, Specification Paras. 29-34 provide plenty of detail on types of abstract data processing and determinations the anomaly detection ML model and the anomaly evaluation ML model are capable of doing, but provides scant detail as to specifically how they accomplish these tasks. This high level generic disclosure supports the findings that these elements serve only to generally link the abstract ideas (such as “process[ing] feature vectors to detect anomalies” and “process[ing] intermediate feature vectors generated by the anomaly detection ML model-not raw data or final outputs, but the detection model's internal representations” which amount to merely processing data) to the field of machine learning modeling. While the instant application delves into a more specific branch of machine learning modeling, i.e., “two-model ML architecture”, even this receives only generic disclosure in Specification Para. 31, stating merely “Anomaly detection model(s) 230 may flag data as anomalous or potentially anomalous (e.g., borderline) but, in some embodiments, need not make a final determination … Such a determination may be made by an anomaly evaluation model 240 … anomaly [evaluation] model 240 may operate on intermediate feature vectors generated by anomaly detection model(s) 230, e.g., feature vectors that are used as inputs into a final classifier of the anomaly detection model(s) 230 or some other (earlier) intermediate outputs.” Here, much like the disclosure of the two models themselves, the disclosure only provides high level generalities that the two models are able to perform the abstract ideas communicating data between them, but is fully devoid of any detail regarding how connectedness of the two models actually works. Therefore, whether evaluated individually or as a whole/ordered combination these additional elements fail to integrate the abstract idea into a practical application or amount to significantly more.
On Page 11 of the Response, Applicant argues “Furthermore, the amended claims recite that the anomaly evaluation ML model generates the risk indication ‘based at least on a similarity of the intermediate feature vectors with feature vectors corresponding to one or more security triggers in a feature space of the anomaly evaluation ML model.’ This is a specific technical mechanism comparing intermediate feature vectors against security trigger clusters using similarity in the evaluation model's feature space. This is not an abstract concept that can be performed in the human mind; it is a defined computational process operating on learned internal representations within a multi-dimensional feature space.”
Examiner notes, as discussed further in the detailed rejection below, insofar as they are claimed “generat[ing] the risk indication ‘based at least on a similarity of the intermediate feature vectors with feature vectors corresponding to one or more security triggers in a feature space’” is a recitation of the abstract idea an unhelpful in bringing the claims to eligibility, much like the discussion above, these abstract ideas are carried out by the anomaly evaluation ML model merely through generally linking them to the field of machine learning. While recited along side the technical aspect of the anomaly evaluation ML model, the “comparing intermediate feature vectors against security trigger clusters using similarity in the evaluation model's feature space” is the abstract idea that is being performed and is not technical itself. Further, while these comparisons and data manipulations cannot be easily performed in the human mind, the abstract ideas are not classified as “mental processes”, and therefore the ability or inability to perform these processes in the human mind is not a valid test for this abstract idea, rendering this aspect of the argument moot.
On Pages 11-12 of the Response, Applicant argues “The amended claims are analogous to the patent-eligible Claim 3 in USPTO Subject Matter Eligibility Example 47. In Example 47, Claim 3 was found eligible because it used an ANN to detect anomalies in network traffic and then took specific remedial actions (dropping malicious packets, blocking future traffic from the source address) that improved network security. … Similarly, amended claims 1, 9, and 17 use ML models to detect anomalies in agent activity data and then take specific remedial actions (remotely causing re-authentication) that improve the security of customer data during live agent-customer interactions. Unlike the ineligible Claim 2 in Example 47-which merely detected and analyzed anomalies in a generic ‘data set’ without any specific remedial action-the amended claims recite (i) a specific two-model architecture with architectural coupling via intermediate feature vectors, (ii) a specific mechanism for risk determination using similarity with security trigger feature vectors in a feature space, and (iii) specific automated remedial actions that improve security by remotely causing an agent security application to generate a re-authentication request. Importantly, the re-authentication requirement recited in the claims-‘remotely causing an agent security application operating on a computing device accessible to the agent, to generate a re-authentication request to the agent’-effectively prevents the agent from continuing to access customer data until the agent satisfies the re-authentication request. This is analogous to the remedial actions in Example 47, Claim 3 (dropping malicious packets and blocking future traffic), because it automatically and proactively prevents continued access to protected data in response to a detected security threat, rather than merely alerting a human to take action. The re-authentication acts as a technical gate that restricts the agent's access to customer data until identity is re-verified, thereby providing a concrete technical improvement to the security of the interaction center's computing system.”
Examiner notes, contrary to Example 47, which provide the explicitly technical background of the dropping of malicious packets and the blocking of future traffic from the source address, no such analogous technical detail is provided regarding the “re-authentication request”. The “re-authentication request” is disclosed in the specification merely at face value as “a single sign-on (SSO) authentication, a two-point authentication, a biometric authentication (e.g., fingerprint/retina/picture authentication), or any other suitable authentication technique” (Para. 35) and under broadest reasonable interpretation this can be interpreted as simply “alerting a human to take action” or merely the abstract verification of data. Further, it is noted that the features upon which applicant relies in this argument (i.e., “effectively prevent[ing] the agent from continuing to access customer data until the agent satisfies the re-authentication request”, “automatically and proactively prevents continued access to protected data in response to a detected security threat, rather than merely alerting a human to take action”, and “act[ing] as a technical gate that restricts the agent's access to customer data until identity is re-verified”) are not recited in the rejected claims or in the Applicant’s specification, and therefore are presented in merely a conclusory manner (i.e., a bare assertion of an improvement without the detail necessary to be apparent to a person of ordinary skill in the art), thus the Examiner cannot determine the claims improve the technology (See MPEP 2106.04(d)(1)).
On Pages 12-13 of the Response, Applicant argues “The Examiner previously characterized the claims as ‘certain methods of organizing human activity’ (commercial interactions). However, the amended claim language-particularly the cascaded ML architecture with intermediate feature vector processing and similarity-based security trigger matching-cannot practically be performed in the human mind or by organizing human activity. These are specific technical processes that impose meaningful limits on any abstract idea. The human mind is not equipped to process intermediate feature vectors generated by an anomaly detection ML model and compare them against security trigger clusters using similarity in a multi- dimensional feature space. The dependent claims further add technical specificity. Claims 3 and 11 recite that the multi-dimensional feature space is associated with both the agent activity data and contextual information about the agent-customer interaction, and that the anomaly detection ML model detects the anomaly based at least on the contextual information. Claims 4, 12, and 19 recite that the anomaly evaluation ML model is trained using at least a training dataset comprising outputs of the anomaly detection ML model that are labeled as one of true security concerns or false security concerns-further deepening the architectural interrelationship between the two models. For at least these reasons, Applicant respectfully submits that the amended claims are integrated into a practical application and are patent-eligible under 35 U.S.C. § 101. Accordingly, Applicant respectfully requests that the rejections of claims 1-20 under 35 U.S.C. § 101 be withdrawn.
Examiner notes, as discussed above, the recited abstract idea is categorized as “certain methods of organizing human activity” and therefore arguments regarding the inability to perform certain steps in the human mind are moot. Further, the determination of the claimed abstract idea as “certain methods of organizing human activity” is made “based on whether the activity itself falls within one of the sub-groupings”. In this case, the activities such as collecting and processing activity data regarding customer/agent interactions in order to detect, evaluate, and respond to customer risk falls within the sub-grouping of “commercial interaction”. While the ML models are used to carry out aspects of these commercial interactions, as discussed at length above and in the detailed rejection below, they fail to integrate the abstract idea into a practical application or amount to significantly more because they merely serve to generally link the abstract idea to the field of machine learning.
Further, as discussed further in the detailed rejection below, “the multi-dimensional feature space [being] associated with both the agent activity data and contextual information about the agent-customer interaction, and … detecting the anomaly based at least on the contextual information” insofar as they are claimed in claims 3 and 11 are recitations of the abstract idea and unhelpful in bringing the claims to eligibility. Similarly, “train[ing] using at least a training dataset comprising outputs … that are labeled as one of true security concerns or false security concerns” insofar as they are claimed in claims 4, 12, and 19 are recitations of the abstract idea and unhelpful in bringing the claims to eligibility.
Applicant’s remarks on Pages 13-17 of the Response, regarding the previous rejection of the claims under 35 U.S.C. 103, have been fully considered and are found persuasive in light of the amended 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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
Claims 1-8 are directed to a server (i.e., a machine); claims 9-16 are directed to a method (i.e., a process); claims 17-20 are directed toa non-transitory storage medium (i.e., a machine). Therefore, claims 1-20 all fall within the one of the four statutory categories of invention.
Step 2A, Prong One
Independent claim 1 substantially recites storing customer data associated with the customer, and
collecting agent activity data comprising one or more of:
one or more stored files viewed by the agent in association with the live agent-customer interaction, or
duration of viewing of the one or more stored files;
processing the agent activity data to generate one or more feature vectors, each of the one or more feature vectors corresponding to a point in a multi-dimensional feature space associated and representative of at least one pattern in the agent activity data;
processing the one or more feature vectors to generate an indication that the agent is placing customer data at risk, wherein processing the one or more feature vectors comprises:
processing the one or more feature vectors to detect an anomaly in the agent activity data; and
processing, responsive to the detected anomaly in the agent activity data, intermediate feature vectors generated to generate the indication that the customer data is at risk based at least on a similarity of the intermediate feature vectors with feature vectors corresponding to one or more security triggers in a feature space; and
automatically causing, responsive to the indication that the agent is placing the customer data at risk, one or more remedial actions to be performed, the one or more remedial actions comprising at least:
remotely causing, accessible to the agent, to generate a re-authentication request to the agent.
Independent claims 9 and 17 substantially recite collecting agent activity data comprising one or more of:
one or more stored files viewed by the agent in association with the live agent- customer interaction, or
duration of viewing of the one or more stored files;
processing the agent activity data to generate one or more feature vectors, each of the one or more feature vectors corresponding to a point in a multi-dimensional feature space associated and representative of at least one pattern in the agent activity data;
processing the one or more feature vectors to generate an indication that the agent is placing the customer data at risk, wherein processing the one or more feature vectors, comprises:
processing the one or more feature vectors to detect an anomaly in the agent activity data; and
processing, responsive to the detected anomaly in the agent activity data, intermediate feature vectors generated to generate the indication that the customer data is at risk based at least on a similarity of the intermediate feature vectors with feature vectors corresponding to one or more security triggers in a feature space; and
responsive to the indication that the agent is placing the customer data at risk, automatically causing one or more remedial actions to be performed, the one or more remedial actions comprising at least:
remotely causing, accessible to the agent, to generate a re-authentication request to the agent.
The limitations stated above are processes/functions that under broadest reasonable interpretation covers “certain methods of organizing human activity” (commercial or legal interactions) of ensuring security of interactions. Therefore, the claim recites an abstract idea.
Step 2A, Prong Two
The judicial exception is not integrated into a practical application. Claims 1, 9, and 17 as a whole amount to: (i) merely invoking generic components as a tool to perform the abstract idea or “apply it” (or an equivalent), and (ii) generally links the use of a judicial exception to a particular technological environment or field of use. The claim recites the additional elements of: (i) a memory device (claim 1), (ii) one or more processing devices/a processing device (claims 1, 9, 17), (iii) one or more machine learning (ML)-readable feature vectors (claims 1, 9, 17), (iv) one or more ML models (claims 1, 9, 17), (v) an agent security application (claims 1, 9, 17), (vi) a computing device (claims 1, 9, 17), (vii) an anomaly detection ML model (claims 1, 9, 17), (viii) an anomaly evaluation ML model (claims 1, 9, 17), (ix) a computing server (claim 1), and (x) a non-transitory computer-readable storage medium storing instructions (claim 17).
The additional elements of (i) a memory device, (ii) one or more processing devices/a processing device, (v) an agent security application, (vi) a computing device, (ix) a computing server, and (x) a non-transitory computer-readable storage medium storing instructions are recited at a high level of generality (see [0054] of the Applicant’s specification discussing the memory, [0046] discussing one or more processing devices/a processing device, [0010] discussing the agent security application, and [0014] discussing the computing device and the computing server, [0058] discussing the non-transitory computer-readable storage medium storing instructions) such that, when viewed as whole/ordered combination, it amounts to no more than mere instruction to apply the judicial exception using generic computer components or “apply it” (See MPEP 2106.05(f)).
The additional element of (iii) one or more machine learning (ML)-readable feature vectors, (iv) one or more ML models, (vii) an anomaly detection ML model (claims 1, 9, 17), and (viii) an anomaly evaluation ML model are recited at a high level of generality (See [0048] of the Applicant’s specification discussing the one or more machine learning (ML)-readable feature vectors, and [0019, 0026, 0029] discussing the one or more ML models, [0026] discussing the anomaly detection ML model, [0033] discussing the anomaly evaluation ML model) such that when viewed as whole/ordered combination, do no more than generally link the use of the judicial exception to a particular technological environment or field of use (i.e., machine learning and mobile applications) (See MPEP 2106.05(h)).
Accordingly, these additional elements, when viewed as a whole/ordered combination [See Figures 1 and 2 showing all the additional elements (i) a memory device, (ii) one or more processing devices/a processing device, (iii) one or more machine learning (ML)-readable feature vectors, (iv) one or more ML models, (v) an agent security application, (vi) a computing device, (vii) an anomaly detection ML model, (viii) an anomaly evaluation ML model, (ix) a computing server, and (x) a non-transitory computer-readable storage medium storing instructions in combination], do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. Thus, the claim is directed to an abstract idea.
Step 2B
As discussed above with respect to Step 2A Prong Two, the additional elements amount to no more than: (i) “apply it” (or an equivalent), and (ii) generally link the use of a judicial exception to a particular technological environment or field of use, and are not a practical application of the abstract idea. The same analysis applies here in Step 2B, i.e., (i) merely invoking the generic components as a tool to perform the abstract idea or “apply it” (See MPEP 2106.05(f)); and (ii) generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)), does not integrate the abstract idea into a practical application at Step 2A or provide an inventive concept at Step 2B. Thus, even when viewed as a whole/ordered combination, nothing in the claims adds significantly more (i.e., an inventive concept) to the abstract idea. Thus, the claims 1, 9, and 17 are ineligible.
Dependent Claims 3-8, 11-16, 19 and 20 merely narrow the previously recited abstract idea limitations. For reasons described above with respect to claims 1, 9, and 17 these judicial exceptions are not meaningfully integrated into a practical application or significantly more than the abstract idea. Thus, claims 3-8, 11-16, 19 and 20 are also ineligible.
Step 2A, Prong Two
Dependent Claims 2, 10, and 18 further narrow the previously recited abstract idea limitations and further recite the additional abstract idea limitations of: wherein the feature vectors comprise a representation of a voice sample of the agent collected during the live agent-customer interaction, and determining that the voice sample of the agent collected during the live agent-customer interaction does not match one or more stored voice samples of the agent.
Claims 2, 10, and 18 also recites the additional elements of a voice recognition ML model, which is recited at a high-level of generality (See [0026] of the Applicants PG Publication disclosing the voice recognition ML model) such that when viewed as whole/ordered combination, the additional elements do no more than generally link the use of the judicial exception to a particular technological environment or field of use (i.e., machine learning) (See MPEP 2106.05(h)).
Accordingly, the additional elements, when viewed individually and as a whole/ordered combination, do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. Thus, the claims are directed to an abstract idea.
Step 2B
As discussed above with respect to Step 2A Prong Two, the additional element amounts to no more than: generally linking the use of a judicial exception to a particular technological environment or field of use, and is not a practical application of the abstract idea. The same analysis applies here in Step 2B, i.e., (i) generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)), does not integrate the abstract idea into a practical application at Step 2A or provide an inventive concept at Step 2B.
Therefore, the additional element of a voice recognition ML model does not integrate the abstract idea into a practical application at Step 2A or provide an inventive concept at Step 2B. Thus, even when viewed as a whole/ordered combination, nothing in the claim adds significantly more (i.e., an inventive concept) to the abstract idea. Thus, claims 2, 10, and 18 are ineligible.
Novel and Non-Obvious Over the Prior Art
Claims 1-20 are novel and non-obvious over the prior art; however, these claims are subject to the above rejections.
The closest prior art is U.S. Patent Application No. 2016/0065732 to Davis et al (hereafter Davis). Davis discloses a server system and method including a computing device accessible to the agent for collecting agent activity data, including files viewed during agent-customer interactions and duration of viewing, processing agent activity to determine patterns, assessing agent activity for customer data risk and cause remedial action to be performed
The next closest prior art is U.S. Patent Application No. 2022/0207506 to Daruna et al (hereafter Daruna). Daruna discloses using machine learning feature vectors in a multi-dimensional feature space representative of patterns in activity data and generates indicator outputs of detected anomalies in the activity data.
The next closest prior art is U.S. Patent Application No. 2021/0194883 to Badhwar et al (hereafter Badhwar). Badhwar discloses remotely causing an agent security application to generate a re-authentication request.
The next closest prior art is U.S. Patent No. 11,895,264 to Phatak et al (hereafter Phatak). Phatak discloses generating indication that the customer data is at risk based at least on a similarity of feature vectors with feature vectors corresponding to one or more security triggers in a feature space.
The next closest prior art is U.S. Patent Application No. 2020/0395123 to Akselrod-Ballin et al (hereafter Akselrod-Ballin). Akselrod-Ballin discloses processing intermediate feature vectors in connection with multiple machine learning models.
While the closest prior art above teaches the various aspects of the claimed invention individually, the combination of these references are not obvious in such a way that they would have been obvious to one of ordinary skill in the art at the time of invention. Specifically Davis in view of Daruna and further in view of Badhwar and even further in view of Phatak and even further in view of Akselrod-Ballin does not explicitly disclose the limitation “process, responsive to the detected anomaly in the agent activity data, using an anomaly evaluation ML model, intermediate feature vectors generated by the anomaly detection ML model to generate the indication that the customer data is at risk based at least on a similarity of the intermediate feature vectors with feature vectors corresponding to one or more security triggers in a feature space of the anomaly evaluation ML model” (emphasis added) as recited in representative claim 1, and similarly recited in independent claims 9 and 17. Therefore, the claims are rendered novel and non-obvious over the prior art.
Conclusion
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 DAVID G GODBOLD whose telephone number is (571)272-5036. The examiner can normally be reached M-F 8-5.
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, Shannon S Campbell can be reached at 571-272-5587. 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.
/DAVID G. GODBOLD/Examiner, Art Unit 3628
/RUPANGINI SINGH/Primary Examiner, Art Unit 3628
Prosecution Insights