DETAILED ACTION
Notice to Applicant
The following is a FINAL Office action upon examination of application number 18/626,893 filed on 04/04/2024. Claims 1-20 are pending in this application, and have been examined on the merits discussed below.
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Amendment
In the response filed April 20, 2026, Applicant amended claims 1, 7, 11 and 17, and did not cancel any claims. No new claims were presented for examination.
Applicant's amendments to claims 1, 7, 11, and 17 are hereby acknowledged. The amendments are sufficient to overcome the previously issued rejection of claims 1-19 under 35 U.S.C. 112(b); accordingly, this rejection has been withdrawn.
Applicant's amendments to claims 1, 7, 11, and 17 are hereby acknowledged. The amendments are not sufficient to overcome the previously issued claim rejection under 35 U.S.C. 101; accordingly, this rejection has been maintained.
Response to Arguments
Applicant's arguments filed April 20, 2026, have been fully considered.
Applicant submits “The Examiner has alleged that the claimed steps "can be performed in the human mind using pen and paper." However, claims do not recite a mental process when they do not contain limitations that can practically be performed in the human mind, for instance when the human mind is not equipped to perform the claim limitations.” [Applicant’s Remarks, 04/20/2026, page 8]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that the limitations of identifying interaction events, generating a prediction prompt, and estimating futures interactions can be interpreted as tasks that can be performed in the human mind using pen and paper. For instance, a person could manually review interaction data, generate prediction based on observed patters, and estimate future outcomes. The presence of a machine learning model in the claim does not alter the fact that the fundamental process of identifying patterns and generating predictions can be seen as a mental process. Therefore, the claim continues to fall under the “Mental Process” abstract idea grouping. For the reasons above, this argument is found unpersuasive.
Applicant submits “The step of "generating a prediction prompt for estimating one or more future interaction events" as recited by claim 1 requires processing interaction metadata items from interactions assigned to an agent in a manner that produces a structured input suitable for a machine learning model. This is not a mental process as it requires computational processing of data structures that cannot be replicated through human cognition alone.” [Applicant’s Remarks, 04/20/2026, page 8]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that while the inclusion of a machine learning model ay involve technical components, the core process, in this case identifying interaction events and generating a prediction prompt based on those events, remains abstract. Although the “generating” step involves processing data in a structures manner for a machine learning model, this activity is still part of the abstract idea. The generation of a prediction prompt can be viewed as a mental process in which a human could analyze interaction data, identify patterns, and form predictions based on prior experience. The presence of a machine learning model used to automate the process does not transform the underlying abstract concept, as the predicting future events based on past interactions remains an abstract idea. For the reasons above, this argument is found unpersuasive.
Applicant submits “The step of "applying said prediction prompt to a machine learning model to estimate said one or more future interaction events" as recited by claim 1 explicitly requires a machine learning model to perform the estimation. A human cannot mentally execute a machine learning model, which involves mathematical operations across potentially thousands of parameters and weights. This limitation is fundamentally incompatible with mental performance.” [Applicant’s Remarks, 04/20/2026, page 8]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that
while a machine learning model may involve computational complexity that cannot be performed mentally, the core of the claim remains abstract. The underlying process of estimating future interaction events based on identified interaction events can still be viewed as mental process. In particular, the identification of one or more interaction events, generation of a prediction prompt, and estimation of future interaction events are all actions that can be performed by a human using pen and paper. The presence of a machine learning model employed to automate these process does not transform the underlying abstract concept. The reliance on a machine to carry out mathematical operations does not eliminate the fact that the central concept of predicting future events based on prior data is abstract, and such process can be performed mentally with the aid of pen and paper. For the reasons above, this argument is found unpersuasive.
Applicant submits “The specification explains that embodiments provide for "automatically predicting interaction events for interactions, e.g. between agents and customers and distributing workload for agents based on generated predictions" and "more efficiently distribute data such as interaction data in a call center." As-Filed Specification, paragraph [0006]. This context demonstrates that the claims address a technical problem in call center systems and the efficient distribution of interaction data through a technical solution involving machine learning-based prediction.” [Applicant’s Remarks, 04/20/2026, page 9]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that the cited passage describes a business objective rather than a technical problem. “Efficiently distribute data” and “predicting interaction events” are described at a high level and do not identify a specific technological deficiency in prior computer systems. The claimed solution (i.e., using a machine learning model to generate and apply a prediction prompt) is recited without any technical details as to how the model is implemented or how it improves computer functionality. For the reasons above, this argument is found unpersuasive.
Applicant submits Claim 5 further recites that "estimating said one or more future interaction events includes sequential initiation and termination of said one or more interactions, thereby maintaining a concurrent assignment of interaction requests to said agent." This maintenance of concurrent assignment of interaction requests requires real-time computational coordination that cannot be performed mentally.” [Applicant’s Remarks, 04/20/2026, page 9]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that the limitation merely describes “maintaining a concurrent assignment of interaction requests” through sequential initiation and termination of interactions. This reflects scheduling or workload management concepts, which are a form of organizing human activity. The claim does not recite any specific technical mechanism for achieving the alleged “real-time computational coordination.” There are no details regard how concurrency is implemented, managed, or improved at a system level. Moreover, in response to the Applicant’s argument that “This maintenance of concurrent assignment of interaction requests requires real-time computational coordination that cannot be performed mentally,” it is noted that a person can conceptually track interaction start and end times and adjust assignments to maintain concurrency. For the reasons above, this argument is found unpersuasive.
Applicant submits “claim 4 recites that "estimating said one or more future interaction events includes determining a latency in responses of said agent to one or more interactions." Determining response latency across multiple concurrent interactions requires computational measurement and analysis of timing data, which is beyond human mental capability.” [Applicant’s Remarks, 04/20/2026, page 9]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that determining “a latency in response” amounts to evaluating timing information. For instance, observing when a response is received and comparing it to when an interaction occurred. This is a type of data analysis that is conceptually analogous to a human tracking response times (i.e., noting delays with a clock or log). The claim does not recite any specific technical manner for measuring or computing latency. Instead, it broadly recites the result of “determining latency: without technical detail. For the reasons above, this argument is found unpersuasive.
Applicant submits “Claim 11 recites a system comprising "a computing device; a memory; and a processor, the processor configured to: identify one or more interaction events from interaction metadata items located in one or more interactions assigned to an agent; generate a prediction prompt for estimating one or more future interaction events for said one or more interactions based on said identified interaction events; and apply said prediction prompt to a machine learning model to estimate said one or more future interaction events for said one or more interactions." The processor is specifically configured to perform these operations, demonstrating that the claims are directed to a particular technological implementation rather than an abstract concept.” [Applicant’s Remarks, 04/20/2026, page 9]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that
the claim has not been shown to recite or require any non-generic computer hardware/software to implement any of the claims steps. The Examiner further emphasizes that neither Applicant's claims nor the Specification support Applicant’s assertion that specialized computer hardware and software are required to implement the invention. Notably, Applicant's Specification suggests that virtually any computing device under the sun can be used to implement the claims, including general purpose computers. See, e.g., paragraphs [0059]: “In practice, an LLM or NN, or NN learning, can be simulated by one or more computing nodes or cores, such as generic central processing units (CPUs, e.g., as embodied in personal computers) or graphics processing units (GPUs such as provided by Nvidia Corporation), which can be connected by a data network.” and [0062]: “Computing device 100 may include a controller or processor 105 that may be, for example, a central processing unit processor (CPU), a chip or any suitable computing or computational device, an operating system 115, a memory 120, a storage 130, input devices 135 and output devices 140 such as a computer display or monitor displaying for example a computer desktop system.” Accordingly, the generic computing elements recited in the is similar to simply adding the words “apply it,” which is not sufficient to amount to a practical application or add significantly more, as noted in MPEP 2106.
Furthermore, in so far as Applicant is suggesting that a "specialized computer hardware and software" mandates a finding of eligibility, the Examiner respectfully points out that the Federal Circuit’s opinion in In re Alappat, 33 F.3d 1526 (Fed. Cir. 1994) (which once supported the notion that a special purpose computer is §101-eligible) has been superseded by at least the Supreme Court’s Bilski, Mayo, and Alice opinions, such that eligibility does not hinge on the presence or absence of a special purpose computer to perform the claims, but instead hinges on the outcome of the subject matter eligibility inquiry adhering to more recent and authoritative guidance gleaned from the Supreme Court’s Mayo/Alice decisions, and MPEP 2106 that sets forth the procedure for determining subject matter eligibility in compliance with the controlling law.
Accordingly, even assuming arguendo that the claimed general purpose computer that has programmed/configured to perform Applicant’s invention were considered as “specialized computer hardware and software,” this alone would not be sufficient to mandate a finding of §101 eligibility because, when an abstract idea is recited in the claims, evaluation of any additional elements, including a “special purpose computer,” must be conducted to determine whether the additional limitations (e.g., the alleged special purpose computer) amount to a practical application or add significantly more beyond the abstract idea. Although neither Applicant's claims nor Specification indicate that a special purpose computer is required to implement the system, even if a special purpose computer was required, this alone would not be sufficient to render the claims eligible. Furthermore, with respect to exemplary claim 11, none of the steps of identify, generate, and apply, individually or in combination, have been shown to yield an improvement to a computer or to any technology. Notably, the claims have not been shown to modify, reconfigure, manipulate, or transform the computing device, memory, processor, or any technology in any discernible manner, much less yield an improvement thereto. There is simply no support to show that implementing the claim steps with a processor amounts to an improvement to the computer or to any other technology. Therefore, regardless of whether or not the claimed generic computing elements amount to "specialized computer hardware and software," these additional elements have been shown to integrate the abstract idea into a practical application or add significantly more to the claims. For the reasons above, this argument is found unpersuasive.
Applicant submits The claims do not merely recite generic computer components performing abstract functions. Rather, "generating a prediction prompt" as recited by claims 1, 11, and 20 represents a specific technical step that transforms identified interaction events into a format suitable for machine learning processing. This is a concrete data transformation step, and not an abstract mental process.” [Applicant’s Remarks, 04/20/2026, pages 9-10]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that
Characterizing “generating a prediction prompt” as a “specific technical step” is not supported by the claim language itself. The step is recited solely In terms of its inputs and intended output, without reciting any particular algorithm, data structure, encoding scheme, or other technical implementation details. The claim does not explain how the alleged “transformation” is performed in a technically specific manner or how it improves the operation of the computer or the machine learning model itself. Instead, it broadly describes formatting data for use in a model, which remains a form of data preparation. For the reasons above, this argument is found unpersuasive.
Applicant submits “The Examiner's characterization of the claims as "managing interactions between people by following rules" mischaracterizes the claimed subject matter. The claims are directed to a technical method of predicting future interaction events using machine learning, and not to the management of human interactions themselves.” [Applicant’s Remarks, 04/20/2026, page 10]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that the claimed limitations are framed in terms of organizing and forecasting interaction handling within a workflow, rather than improving a specific technical processes or system operation. Although the claims recite “machine learning,” this is recited at a high level of generality without reciting a specific technical improvement to the machines learning model itself or to the underlying computing functionality. The use of prediction is applied to the administrative task of managing and allocating interactions, which remains a form of organizing human activity. For the reasons above, this argument is found unpersuasive.
Applicant submits “The prediction of when interactions will terminate or when new interactions should be initiated, as recited by claims 2-3 and 12-13, represents a technical improvement in how interaction data is distributed to agents.” [Applicant’s Remarks, 04/20/2026, page 10]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that “the prediction of when interactions will terminate or when new interactions should be initiated” remains a forecasting and scheduling function applied to interaction data. It describes determining timing and assignment decisions, but does not identify any specific technical improvement. The claim language does not explain how the prediction mechanism improves the functioning of the computer or machine learning model itself, nor does it recite any particular technical method for achieving improved prediction accuracy or computational efficiency. Instead, it uses predicted timing results to support allocation of interactions to agents, which is and administrative outcome. Therefore, this limitation reflects abstract prediction ad resource scheduling applied in a business context, rather than a technical improvement to computer technology. For the reasons above, this argument is found unpersuasive.
Applicant submits “Claims 6 and 16 further recite "identifying an interaction capacity of said agent from said interaction metadata items; and evaluating, using machine learning, whether said agent has capacity to receive a new interaction request." This machine learning-based evaluation of agent capacity represents a specific technical application that improves the functioning of interaction distribution systems.” [Applicant’s Remarks, 04/20/2026, page 10]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that “identifying an interaction capacity of said agent from said interaction metadata items; and evaluating, using machine learning, whether said agent has capacity to receive a new interaction request” describe determining workload availability and making an allocation decision based on that determination. This is fundamentally a resource scheduling or assignment tasks. The use of “machine learning” is recited only at a high level and does not recite any specific model structure training technique, or technical improvement to the machine learning process itself. Instead, it describes using a generic machine learning model as a tool to perform a capacity assessment based on input metadata. The claim does not explain how this evaluation improves the operation of the computing system or the machine learning model, as opposed to merely automating a decision making step in interaction routing. For the reasons above, this argument is found unpersuasive.
Applicant submits “The ordered combination of elements: identifying interaction events from metadata, generating a prediction prompt based on those events, and applying that prompt to a machine learning model represents a specific technical process for predicting future interaction events. This combination provides a technological improvement in distributing interaction data to agents, and not merely an automation of a mental process.” [Applicant’s Remarks, 04/20/2026, page 10]
The Examiner respectfully disagrees. In response to Applicant’s argument, it is noted that the argument relies on characterizing the claim as “a specific technical process,” but the ordered combination still amounts to generic data handling steps related to extracting information from metadata, formatting it for input, and applying a machine learning model to generate a prediction. The claim does not identify any particular technical mechanism by which these steps interact to improve computer functionality or machine learning performance. The sequence describe a logical workflow of data preparation and analysis, but not a technical improvement in the way data is processed, stored, or computed. Merely arranging abstract steps in a sequence does not convert them into a technological solution, absent a demonstrated improvement in computer or model operation. Here, the purported benefit is still directed to better distributing interactions to agents, which is an administrative outcome rather than a technical improvement. Therefore, the ordered combination does not amount to a technical improvement. For the reasons above, this argument is found unpersuasive.
Applicant submits “that the combination of Traba and Low fails to teach or suggest at least "generating a prediction prompt for estimating one or more future interaction events for said one or more interactions based on said identified interaction events" and "applying said prediction prompt to a machine learning model to estimate said one or more future interaction events for said one or more interactions," as recited in the independent claims.” [Applicant’s Remarks, 04/20/2026, page 11]
In response to the Applicant’s argument that “the combination of Traba and Low fails to teach or suggest at least "generating a prediction prompt for estimating one or more future interaction events for said one or more interactions based on said identified interaction events" and "applying said prediction prompt to a machine learning model to estimate said one or more future interaction events for said one or more interactions," as recited in the independent claims,” it is noted that this argument is a mere allegation of patentability by the Applicant with no supporting rationale or explanation. Merely stating that the claims do not teach a feature does not offer any insight as to why the specific sections of the prior art relied upon by the Examiner fail to disclose the claimed features. Applicant's arguments amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references. For the reasons above, this argument is found unpersuasive.
Applicant submits “Traba does not teach or suggest at least “generating a prediction prompt based on identified interaction events” and “applying that prompt to a machine learning model to estimate future interaction events.”” [Applicant’s Remarks, 04/20/2026, pages 11-12]
With respect to the §103 rejection of independent claim 1, Applicant argues that “Traba does not teach or suggest at least “generating a prediction prompt based on identified interaction events” and “applying that prompt to a machine learning model to estimate future interaction events,” in response it is noted that Applicant’s argument is not commensurate with the rejection. Traba was not relied upon to disclose “generating a prediction prompt based on identified interaction events” and “applying that prompt to a machine learning model to estimate future interaction events. Accordingly, this argument is deemed moot.
Applicant submits “The prompts in Low are designed to elicit responses that emulate human users for testing purposes, not to estimate future interaction events in an agent interaction distribution system.” [Applicant’s Remarks, 04/20/2026, page 12]
With respect to the §103 rejection of independent claim 1, as best understood by Examiner, Applicant argues that Low does not teach “estimating one or more future interaction events.” However, in at least paragraphs 0142, 0148, 0221, Low teaches the instant limitation by constructing prompts from logged interaction data (i.e., subsets of user interaction logs and contextual information) and inputting those prompts to a machine learning model to obtain generated outputs representing predicted user actions. These generated outputs are based on patterns from prior interaction data. This corresponds to the claimed “generating a prediction prompt” from interaction events and applying that prompt to a machine learning model to obtain predicted outcomes. The claim does not limit the “prediction prompt” to a particular application domain. The use of such prompts to elicit model outputs derived from interaction data satisfies the recited functional relationship between input construction and prediction. Thus, given the broadest reasonable interpretation consistent with the specification in construing the claimed invention, it is Examiner’s position that the disclosure of Low teaches the disputed limitation. Accordingly, this argument is found unpersuasive.
Applicant submits “Low does not teach or suggest using prompts to estimate when an interaction will terminate or when a new interaction should be initiated.” [Applicant’s Remarks, 04/20/2026, page 12]
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., using prompts to estimate when an interaction will terminate or when a new interaction should be initiated) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicant submits “There is no teaching or suggestion in either reference that would lead one of ordinary skill in the art to generate a prediction prompt based on interaction events from agent-assigned interactions and apply that prompt to a machine learning model to estimate future interaction events for those interactions. The references address fundamentally different technical problems and employ different technical approaches.” [Applicant’s Remarks, 04/20/2026, page 13]
In response to Applicant’s argument that “there is no teaching or suggestion in either reference that would lead one of ordinary skill in the art to generate a prediction prompt based on interaction events from agent-assigned interactions and apply that prompt to a machine learning model to estimate future interaction events for those interactions,” 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, it is noted that the examiner has provided reasoning articulating why it would have been obvious to combine the references as proposed. The Examiner notes that a prior art reference must either be in the field of applicant’s endeavor or, if not, then be reasonably pertinent to the particular problem with which the applicant was concerned, in order to be relied upon as a basis for rejection of the claimed invention. See In re Oetiker, 977 F.2d 1443, 24 USPQ2d 1443 (Fed. Cir. 1992). The Examiner points out that the rejection of claim 1 provides an articulated line of reasoning based on the teachings of the prior art, the knowledge of one skilled in the art, and the motivation to modify the prior art to arrive at the conclusion of obviousness of claimed invention, which is a permissible means to support the legal conclusion of the obviousness of the claimed subject matter. See KSR Int'l Co. v. Teleflex Inc., 550 U.S. at 418, 82 USPQ2d at 1396 (quoting In re Kahn, 441 F.3d 977, 988, 78 USPQ2d 1329, 1336 (Fed. Cir. 2006)).
As discussed in the Office action, the rejection relied on the combined teachings of Traba, which discloses generating predictive inputs from interaction related data and applying machine learning models to estimate future interaction outcomes, and Low, which discloses constructing prompts from interaction log data and applying them to machine learning models to generate predicted actions. Together, these teachings would have suggested using interaction derived inputs in prompt form to drive machine learning based predictions of future interaction events, which is a predictable combination of known techniques. The argument regarding different technical problems is not persuasive, as both references relate to predicting user or interaction outcomes using machine learning based on interaction data.
As the claims have been given their "broadest reasonable interpretation consistent with the specification", the Examiner asserts that the scope and contents of the prior art have been determined, thereby satisfying the first factual inquiry set forth by Graham v. John Deere Co. The Examiner applied the teachings of Traba and Low, and determined the deficiencies, thereby ascertaining the differences between the prior art and the claims at issue. The Examiner has fulfilled the role of factfinder while resolving the Graham inquiries, as per MPEP 2141, and determined that the level of ordinary skill in the art is reflected by the prior art itself, thereby resolving the level of ordinary skill in the pertinent art. The Examiner asserts that the Graham factual inquiries have been properly resolved, resulting in a proper prima facie case of obviousness. Accordingly, this argument is found unpersuasive.
Applicant submits “the addition of McGann does not render obvious the subject matter of claims 2, 8-10, 12, 18, and 19.” [Applicant’s Remarks, 04/20/2026, page 13]
In response to the Applicant’s argument that “the addition of McGann does not render obvious the subject matter of claims 2, 8-10, 12, 18, and 19,” it is noted that this argument is a mere allegation of patentability by the Applicant with no supporting rationale or explanation. Merely stating that the claims do not teach a feature does not offer any insight as to why the specific sections of the prior art relied upon by the Examiner fail to disclose the claimed features. Applicant's arguments amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references. For the reasons above, this argument is found unpersuasive.
Applicant submits “the addition of Ristock does not render obvious the subject matter of claims 5 and 15.” [Applicant’s Remarks, 04/20/2026, page]
In response to the Applicant’s argument that “the addition of Ristock does not render obvious the subject matter of claims 5 and 15,” it is noted that this argument is a mere allegation of patentability by the Applicant with no supporting rationale or explanation. Merely stating that the claims do not teach a feature does not offer any insight as to why the specific sections of the prior art relied upon by the Examiner fail to disclose the claimed features. Applicant's arguments amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references. For the reasons above, this argument is found unpersuasive.
Applicant’s remaining arguments either logically depend from the above-rejected arguments, in which case they too are unpersuasive for the reasons set forth above, or they are directed to features which have been newly added via amendment. Therefore, this is now the Examiner's first opportunity to consider these limitations and as such any arguments regarding these limitations would be inappropriate since they have not yet been examined. A full rejection of these limitations will be presented later in this Office Action.
Claim Rejections - 35 USC § 101
27. 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.
28. Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The eligibility analysis in support of these findings is provided below, in accordance with MPEP 2106.
With respect to Step 1 of the eligibility inquiry (as explained in MPEP 2106), it is first noted that the method (claims 1-10), device (claims 11-19), and method (claim 20), are directed to at least one potentially eligible category of subject matter (i.e., process, machine, and process, respectively). Thus, Step 1 of the Subject Matter Eligibility test for claims 1-20 is satisfied.
With respect to Step 2A Prong One, it is next noted that the claims recite an abstract idea that falls into the “Certain Methods of Organizing Human Activity” abstract idea set forth in MPEP 2106 because the claims recite steps for distributing interaction data to agents, which encompasses activity for managing personal behavior or relationships or interactions (e.g., following rules or instructions), and steps that can be performed in the human mind (including observation, evaluation, judgment, opinion), and therefore fall under the “Mental Processes” abstract idea grouping. With respect to independent claim 1, the limitations reciting the abstract idea are indicated in bold below: identifying one or more interaction events from interaction metadata items located in one or more interactions assigned to an agent; generating a prediction prompt for estimating one or more future interaction events for said one or more interactions based on said one or more identified interaction events; and applying said prediction prompt to a machine learning model to estimate said one or more future interaction events for said one or more interactions. These steps are organizing human activity by managing interactions between people by following rules, or instructions, and may also be accomplished mentally such as via human observation and perhaps with the aid of pen and paper. The claim recites limitations that fall under the “Certain Methods of Organizing Human Activity” abstract idea grouping because the limitations focus on collecting, analyzing, and distributing interaction data among agents, which are activities that involve managing human interactions and assignment, which are abstract processes related to human task coordination. The claim also falls under the “Mental Processes” abstract idea grouping because the limitations of identifying interaction events, generating a prediction prompt, and estimating futures interactions can be performed in the human mind using pen and paper.
Therefore, because the limitations above set forth activities falling within the “Certain methods of organizing human activity” and “Mental Processes” abstract idea grouping described in MPEP 2106, the additional elements recited in the claims are further evaluated, individually and in combination, under Step 2A Prong Two and Step 2B below. Independent claims 11 and 20 recite similar limitations as those discussed above and are therefore found to recite the same or substantially the same abstract idea as claim 1.
With respect to Step 2A Prong Two, the judicial exception is not integrated into a practical application. With respect to the independent claims, the additional elements are: a machine learning model (claim 1), a computing device, a memory, a processor, and a machine learning model (claim 11), a machine learning model (claim 20). These additional elements have been evaluated, but fail to integrate the abstract idea into a practical application because they amount to using generic computing elements or computer-executable instructions (software) to perform the abstract idea, similar to adding the words “apply it” (or an equivalent), and merely serve to link the use of the judicial exception to a particular technological environment. See MPEP 2106.05(f) and 2106.05(h). In addition, these limitations fail to provide an improvement to the functioning of a computer or to any other technology or technical field, fail to apply the exception with a particular machine, fail to apply the judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition, fail to effect a transformation of a particular article to a different state or thing, and fail to apply/use the abstract idea in a meaningful way beyond generally linking the use of the judicial exception to a particular technological environment.
Accordingly, because the Step 2A Prong One and Prong Two analysis resulted in the conclusion that the claims are directed to an abstract idea, additional analysis under Step 2B of the eligibility inquiry must be conducted in order to determine whether any claim element or combination of elements amount to significantly more than the judicial exception.
With respect to Step 2B of the eligibility inquiry, it has been determined that the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. With respect to the independent claims, the additional elements are: a machine learning model (claim 1), a computing device, a memory, a processor, and a machine learning model (claim 11), a machine learning model (claim 20). These elements have been considered individually and in combination, but fail to add significantly more to the claims because they amount to using generic computing elements or instructions (software) to perform the abstract idea, similar to adding the words “apply it” (or an equivalent), and merely serve to link the use of the judicial exception to a particular technological environment and does not amount to significantly more than the abstract idea itself. Notably, Applicant’s Specification suggests that virtually any type of computing device under the sun can be used to implement the claimed invention (Specification at paragraph [0062]). Accordingly, the generic computer involvement in performing the claim steps merely serves to generally link the use of the judicial exception to a particular technological environment, which does not add significantly more to the claim. See, e.g., Alice Corp., 134 S. Ct. 2347, 110 USPQ2d 1976.).
Even if the machine learning was evaluated as an element beyond software/code for a generic computer to execute, it is noted that that the claimed use of machine learning is recited at a high level of generality these elements amount to well-understood, routine, and conventional activity in the art, which fails to add significantly more to the claims. See, e.g., Magdon-Ismail et al., US 2009/0055270 (paragraph 39: “Both local and central engines may incorporate analysis techniques, such as artificial intelligence, machine learning and other techniques, which are well known in the art”). See also, Anders et al., US 2020/0020015 (paragraph 101: “inferences may be performed by any combination of means known in the art, such as by pattern-matching, text analytics, semantic analytics, statistical methods, artificial intelligence, Bayesian analysis, machine learning, or keyword searching”).
In addition, when taken as an ordered combination, the ordered combination adds nothing that is not already present as when the elements are taken individually. There is no indication that the combination of elements integrate the abstract idea into a practical application. Their collective functions merely provide generic computer implementation. Therefore, when viewed as a whole, these additional claim elements do not provide meaningful limitations to transform the abstract idea into a practical application of the abstract idea or that, as an ordered combination, amount to significantly more than the abstract idea itself.
Dependent claims 2-10 and 12-19 recite the same abstract idea as recited in the independent claims, and when evaluated under Step 2A Prong One are found to merely recite details that serve to narrow the same abstract idea recited in the independent claims accompanied by the same generic computing elements or software as those addressed above in the discussion of the independent claims, which is not sufficient to amount to a practical application or add significantly more, or other additional elements that fail to amount to a practical application or add significantly more, as noted above. In particular, dependent claims 2-9 recite “wherein said one or more future interaction events comprise interaction termination,” “wherein said one or more future interaction events comprise initiating a new interaction,” “wherein estimating said one or more future interaction events comprises determining a latency in responses of said agent to one or more interactions,” “wherein estimating said one or more future interaction events comprises sequential initiation and termination of said one or more interactions, thereby maintaining a concurrent assignment of interaction requests to said agent,” “comprising identifying an interaction capacity of said agent from said interaction metadata items; and evaluating whether said agent has capacity to receive a new interaction request,” “wherein said interaction capacity is identified based on an evaluation of agent data items,” “wherein evaluating said interaction capacity of said agent comprises comparing an interaction latency of an agent to a threshold value,” “wherein said agent is available for receiving said new interaction request when said interaction latency is below said threshold value and wherein said agent is unavailable for receiving said new interaction request when said latency is above said threshold value,” “wherein when said agent is unavailable for receiving an interaction request, identifying another agent for receiving said interaction request”, however these limitations cover activity for managing personal behavior or relationships or interactions (e.g., following rules or instructions), which is part of the same abstract idea as addressed in the independent claims that falls within the “Certain Methods of Organizing Human Activity” abstract idea grouping and also recite steps that may also be accomplished mentally such as via human observation and perhaps with the aid of pen and paper. Dependent claims 6 and 16 recite additional elements of: machine learning. However, when evaluated under Step 2A Prong Two and Step 2B, these additional elements do not amount to a practical application or significantly more since they merely require generic computing devices (or computer-implemented instructions/code) which as noted in the discussion of the independent claims above is not enough to render the claims as eligible.
The ordered combination of elements in the dependent claims (including the limitations inherited from the parent claim(s)) add nothing that is not already present as when the elements are taken individually. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide generic computer implementation. Accordingly, the subject matter encompassed by the dependent claims fails to amount to a practical application or significantly more than the abstract idea itself.
For more information, see MPEP 2106.
Claim Rejections - 35 USC § 103
29. 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 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.
30. The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
31. The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
32. This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
33. Claims 1, 3-4, 6-7, 11, 13-14, 16-17, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Traba et al., Pub. No.: US 2021/0377392 A1, [hereinafter Traba], in view of Low et al., Pub. No.: US 2025/0165678 A1, [hereinafter Low].
As per claim 1, Traba teaches a method of distributing interaction data to agents, the method (paragraph 0002: “methods, apparatus, and systems for routing customer communications…)” comprising:
identifying one or more interaction events from interaction metadata items located in one or more interactions assigned to an agent (paragraph 0100, discussing that once the customer is identified, a predictive calculator accesses a profile of the identified customer to determine customer data. For example, the predictive calculator may request customer data for the ANI (automatic number identification) associated with the identified customer from behavior database to determine customer data including customer attributes and customer interaction history. Examples of customer attributes include personality style or type, communication patterns, preferred mode of communication, or a combination thereof. Examples of customer interaction history include interaction sentiment history, distress history, interaction outcomes, or a combination thereof. In one embodiment, behavior database returns the customer data, including interaction history and communication preferences, to predictive calculator; paragraph 0107, discussing that building the predictive models include collecting interaction data, customer data and agent data. Interaction data can include unstructured and structured data from a plurality of different communication channels utilized by an agent to interact with a customer. For example, interaction data may include a transcription of a previous telephone call or video chat between a customer and an agent, the text of an email exchange between the customer and agent, a written essay or other text unilaterally submitted by a customer, an applicant's enrollment application, or a pre-recorded video clip submitted by a customer. Further, structured telephony data such as call length, call origination, hold time, interaction outcome data, and similar data associated with customer interactions may also be collected. The customer data includes biographical and identification information, and the agent data collected can include training level, personality type, and other data. In some embodiments, the input data collected and/or identified may be derived from customer interactions occurring within the contact center and stored in the database, however, in other embodiments, the data may be imported from external sources, such as one or more third-party databases operated by data collection companies; paragraph 0115, discussing that the predictive model may then be utilized to determine the identified outcome or the likelihood of the identified outcome occurring in association with the current interaction. This interaction may be a telephone call, video chat, instant message conversation, email exchange, or other communication session as described herein. The interaction can be real-time, near real-time (i.e., within 5 minutes, preferably within 2 minutes, and more preferably within 1 minute of capture), previously captured, or a combination thereof. In certain preferred embodiments, it is real-time);
estimating one or more future interaction events for said one or more interactions based on said identified interaction events (paragraph 0094, discussing a predictive calculator that leverages agent performance, ACD skill, and customer data to predict interaction outcome metric values for every available agent if the interaction were routed to that agent. Each model is specific to a customer metric and the predictive calculator uses each model to make a metric-specific prediction; paragraph 0106, discussing a predictive model for AHT (average handle time) that outputs a predicted AHT for the customer and each available agent, a predictive model for CSAT (Customer Satisfaction) that outputs a predicted CSAT for the customer and each available agent, a predictive model for FCR that outputs the predicted likelihood that the communication will be resolved in the first communication for the customer and each available agent, a predictive model for RR that predicts the likelihood that the customer will stay with the company for the customer and each available agent...For example, if there were three agents available and the optimization customer metric was handle time, the predictive calculator would select the handle time model and the model would output the AHT for each of the three agents; paragraph 0112, discussing that once an outcome to be predicted is identified, a predictive model operable to predict the identified outcome or the likelihood of the identified outcome occurring is built using the input data as standardized…As an example, the model may indicate that whether a customer will cancel his or her service is correlated to the customer's personality, the number of distress events during a call, the agent's experience, and the customer's tenure, and assign a coefficient to each of the four variables); and
a model to estimate said one or more future interaction events for said one or more interactions (paragraph 0114, discussing that after a predictive model has identified variables relevant to the identified outcome, a benchmark data set is selected for each identified variable. Specifically, to accurately apply the predictive model to incoming customer interactions, data values related to the relevant variables collected during the incoming customer interactions are standardized before being fed into the model. As discussed above, benchmark data sets define the particular data against which a data value is compared for the generation of its z-score. In other words, selecting a different benchmark data set may generate a different z-score, which, in turn, may result in a different outcome prediction. Thus, selection of benchmark data sets may be utilized to customize prediction results. For example, it may be desired to determine the likelihood of a customer purchasing a product in view of customer interactions recorded in the past six months, rather than all customer interactions ever recorded. To achieve such a prediction result, the benchmark data sets selected would include data associated with customer interactions occurring in the past six months. For example, if the number of distress events per call is deemed relevant to predict an outcome, the number of distress events during a current call may be compared against a benchmark data set that only includes calls recorded in the past six months. Additionally, benchmark data sets may be based on other criteria besides time periods. In one example embodiment, a benchmark data set associated with agent tenure may be selected that includes agent tenure data for different subsets of agents, for example, agents located within a specific contact center or region of the country…; paragraph 0115, discussing that the predictive model may then be utilized to determine the identified outcome or the likelihood of the identified outcome occurring in association with the current interaction).
While Traba describes estimating one or more future interaction events for said one or more interactions based on said identified interaction events and a model to estimate said one or more future interaction events for said one or more interactions, Traba does not explicitly teach generating a prediction prompt for estimating one or more future interaction events for said one or more interactions based on said one or more identified interaction events; and applying said prediction prompt to a machine learning model to estimate said one or more future interaction events for said one or more interactions. However, Low in the analogous art of predictive modeling systems teaches these concepts. Low teaches:
generating a prediction prompt for estimating one or more future interaction events for said one or more interactions based on said one or more identified interaction events (paragraph 0142, discussing that synthetic user memories of N different users may be built using user interactivity data logged during an experiment performed on N different real human users, e.g., an A/B feature test, or A/B feature test that is not yet completed. N different human users may be separated into multiple groups exposed to different features. Using the processes described in FIGS. 10-11, it is possible to simulate and create N different synthetic user memories for the N different real human users based on user interactivity data logged for these N different real human users. Subsets of log entries in the memory log may be used in a prompt, e.g., a prompt chain, to obtain generated responses from a model. The generated responses may include predicted actions based on the contextual information presented in the prompt [i.e., prediction prompt]; paragraph 0148, discussing that in some cases, in addition to a request to generate a predicted action, the request may include one or additional instruction(s) to prompt the model to form higher-level abstract memories and reasoning about the predicted action and/or past user interactivity data. The higher-level abstract memories and/or reasoning may give explainable insights about the predicted actions and/or behavior of users. For example, the first request may further include a first instruction to generate a first reasoning for the first predicted action. The second request may further include a second instruction to generate a second reasoning for the second predicted action. In another example, the first request may further include a third instruction to identify one or more first natural language log entries in the first subset that led to the first predicted action. The second request may further include a fourth instruction to identify one or more second natural language log entries in the second subset that led to the second predicted action; paragraph 0221, discussing a method, including transforming user data into training data, the user data including data collected from a content streaming platform, and user communication data, and the training data including prompts and responses to the prompts; updating parameters of a model using the training data; inputting a first prompt to the model, the first prompt including a first description of a first persona, a context description, and a question; receiving a first response from the model in response to the first prompt; inputting a second prompt to the model, the second prompt including a second description of a second persona different from the first persona, the context description, and the question; receiving a second response from the model in response to the second prompt; and analyzing the first response and the second response); and
applying said prediction prompt to a machine learning model to estimate said one or more future interaction events for said one or more interactions (paragraph 0036, discussing that the one or more models may include machine learning models, which can learn through supervised learning or unsupervised learning. With supervised learning, machine learning models can learn from training data and find patterns or insights from the training data. With unsupervised learning, machine learning models can find patterns or insights directly from the input data; paragraph 0149, discussing that in some cases, to ensure the model generates responses consistently, the model may be prompted to summarize the factors or considerations the synthetic user may take into account when performing a given action. The model may output a set of factors or considerations in a response. Then, the model may be prompted to generate a predicted action [i.e., estimate said one or more future interaction events] for the synthetic user and a reasoning behind the predicted action in view of the factors or considerations that the model produced in the earlier response. The model may generate a response having a predicted action that would be consistent with the factors or considerations that the model produced in the earlier response; paragraph 0203, discussing inputting, into a model, a first subset of the first natural language log entries in the first synthetic user memory log and a first request to generate a first response representing a first predicted action of the first user based on the first subset of the first natural language log entries).
Traba is directed towards a system and method for predictive behavioral routing. Low is directed towards a system and method for prediction of user actions. Therefore they are deemed to be analogous as they both are directed towards prediction systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Traba with Low because the references are analogous art because they are both directed to solutions for predictive modeling and interaction routing, which falls within applicant’s field of endeavor (system and method for distributing interaction data to agents), and because modifying Traba to include Low’s features for generating a prediction prompt for estimating one or more future interaction events for said one or more interactions based on said one or more identified interaction events; and applying said prediction prompt to a machine learning model to estimate said one or more future interaction events for said one or more interactions, in the manner claimed, would serve the motivation of better understanding the behavior of various users and evaluating whether a prompt accurately captures the user and the user's behaviors, and whether the model would generate responses that represent an accurate prediction of the user's actions (Low, paragraphs 0120, 0121); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 3, the Traba-Low combination teaches a method according to claim 1. Traba further teaches wherein said one or more future interaction events comprise initiating a new interaction (paragraph 0111, discussing that an outcome associated with a customer interaction is identified as a target of a predictive model. In more detail, for a contact center it may be desirable to predict an actual outcome or the likelihood of some specific outcome occurring in association with a current customer interaction, be it a telephone-based interaction, web-based interaction, or other type of electronic-assisted interaction. For example, it may be useful for a company to predict a customer satisfaction score, an average handling time, or amount of sales during a customer interaction, taking into account the activities, outcomes, and experiences from prior interactions. Further examples of outcomes associated with a customer include whether a customer will terminate his or her account, whether the customer will purchase a product, whether a customer will pay an outstanding bill, whether a customer is a fraudster, and whether a customer will initiate additional subsequent interaction sessions regarding the same issue, or a combination thereof. This is a non-exhaustive list and additional and/or different outcomes related to a customer or customer interaction may be identified).
As per claim 4, the Traba-Low combination teaches a method according to claim 1. Traba further teaches wherein estimating said one or more future interaction events comprises determining a latency in responses of said agent to one or more interactions (paragraphs 0120-0123, discussing that a customer communication is received from a customer, and there are three available agents to handle the communication. The predictive calculator uses the predictive handling time model to output handle times for each of the three available agents. For example, the predictive handle time model may output: Agent 1, Predicted Handle Time=80 seconds, Agent 2, Predicted Handle Time=100 seconds, Agent 3, Predicted Handle Time=120 seconds).
As per claim 6, the Traba-Low combination teaches a method according to claim 1. Traba further teaches comprising identifying an interaction capacity of said agent from said interaction metadata items (paragraph 0101, discussing that the predictive calculator identifies available agents for handling the customer communication. In several embodiments, predictive calculator determines the available agents by reviewing the occupancy level of agents, e.g., by obtaining agent data from the contact center. ACD (Automatic Call Distributor) dynamically monitors occupancy level of the agents to determine availability and addresses the real-time performance metrics of the agent. This real-time (or near-real time) dynamic data is typically used to select a destination for the customer communication); and
evaluating, using machine learning, whether said agent has capacity to receive a new interaction request (paragraph 0018, discussing using dynamic metric optimization to leverage caller and agent information along with advanced analytics to determine, in real-time, the best customer metric(s) to optimize the routing of a customer to a suitable, available agent. Based on data available from interaction analytics (IA)…, the optimization customer metric for the distribution method is determined in real time by artificial intelligence...For example, a specific customer communication may be routed to optimize average handle time (AHT), first call resolution (FCR), customer satisfaction (CSAT), revenue retention (RR), and/or sales. Dynamic metric optimization enables the ACD to have multi-metric improvement capabilities within a given agent skill; paragraph 0101, discussing that the predictive calculator identifies available agents for handling the customer communication. In several embodiments, predictive calculator determines the available agents by reviewing the occupancy level of agents, e.g., by obtaining agent data from the contact center. ACD dynamically monitors occupancy level of the agents to determine availability and addresses the real-time performance metrics of the agent. This real-time (or near-real time) dynamic data is typically used to select a destination for the customer communication).
As per claim 7, the Traba-Low combination teaches a method according to claim 6. Traba further teaches wherein said interaction capacity is identified based on an evaluation of agent data items (paragraph 0102, discussing that once the available agents are identified, the predictive calculator accesses a profile of each available agent to determine agent data. For example, predictive calculator may request agent data from behavior database to determine agent performance history. Agent performance history includes one or more of: agent effectiveness, revenue generating proficiency , customer satisfaction level, speed, efficiency, experience, cross-sell ability, personal satisfaction, proficiency at closing a transaction, and occupancy, or any combination thereof. Other data that can additionally or alternatively be used in the embodiment above or various alternative embodiments to determine agent performance include the transaction or task type, the time-of-day, the result, a self-rating of the servicing agent respecting the agent's proficiency in handling the customer, the rating of the customer of the agent's proficiency in handling the customer, the rating of another party, such as the agent's supervisor or another observer, or how the customer was serviced; paragraph 0109, discussing that after input data, including multi-channel interaction data, customer data, and agent data, has been collected and/or identified, the input data is preferably standardized...As an example, the multi-channel interaction data may include information about the number of distress events occurring during telephone calls between customers and agents; paragraph 0111).
Claim 11 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 1, as discussed above. Further, as per claim 11 the Traba-Low combination teaches a system for distributing interaction data to agents, the system comprising: a computing device (paragraph 0037: “In the illustrated embodiment, the contact center control system 142 is an information handling system such as a computer, server, workstation, mainframe computer, or other suitable computing device. In other embodiments, the control system 142 may be a plurality of communicatively coupled computing devices coordinated to provide the above functionality for the contact center 100. The control system 142 includes a processor 144 that is communicatively coupled to a system memory 146, a mass storage device 148, and a communication module 150.”; paragraph 0145); a memory (paragraph 0037: “The control system 142 includes a processor 144 that is communicatively coupled to a system memory 146, a mass storage device 148, and a communication module 150….The system memory 146 provides the processor 144 with non-transitory, computer-readable storage to facilitate execution of computer instructions by the processor. Examples of system memory may include random access memory (RAM) devices such as dynamic RAM (DRAM), synchronous DRAM (SDRAM), solid state memory devices, and/or a variety of other memory devices known in the art.”; paragraph 0144); and a processor (paragraph 0037: “The control system 142 includes a processor 144 that is communicatively coupled to a system memory 146, a mass storage device 148, and a communication module 150. The processor 144 can be any custom made or commercially available processor, a central processing unit (CPU), an auxiliary processor among several processors associated with the control system 142, a semiconductor-based microprocessor (in the form of a microchip or chip set), a macroprocessor, a collection of communicatively coupled processors, or any device for executing software instructions.”; paragraph 0145).
Claim 13 recites substantially similar limitations that stand rejected via the art citations and
rationale applied to claim 3, as discussed above.
Claim 14 recites substantially similar limitations that stand rejected via the art citations and
rationale applied to claim 4, as discussed above.
Claim 16 recites substantially similar limitations that stand rejected via the art citations and
rationale applied to claim 6, as discussed above.
Claim 17 recites substantially similar limitations that stand rejected via the art citations and
rationale applied to claim 7, as discussed above.
Claim 20 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 1, as discussed above. Further, as per claim 20, the Traba-Low combination teaches a method of predicting interaction events (paragraph 0002: “The present disclosure relates to methods, apparatus, and systems for routing customer communications, and more particularly to determining how to optimize routing across different customer metrics based on employee, customer, and interaction information.”; paragraphs 0018, 0120).
34. Claims 2, 8-10, 12, 18, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Traba in view of Low, in further view of McGann et al., Pub. No.: US 2017/0111509 A1, [hereinafter McGann].
As per claim 2, the Traba-Low combination teaches a method according to claim 1. Traba further teaches wherein said one or more future interaction events comprise termination (paragraph 0106, discussing a predictive model for AHT (average handle time) that outputs a predicted AHT for the customer and each available agent, a predictive model for CSAT (Customer Satisfaction) that outputs a predicted CSAT for the customer and each available agent, a predictive model for FCR that outputs the predicted likelihood that the communication will be resolved in the first communication for the customer and each available agent, a predictive model for RR that predicts the likelihood that the customer will stay with the company for the customer and each available agent...For example, if there were three agents available and the optimization customer metric was handle time, the predictive calculator would select the handle time model and the model would output the AHT for each of the three agents; paragraph 0111, discussing that an outcome associated with a customer interaction is identified as a target of a predictive model. In more detail, for a contact center it may be desirable to predict an actual outcome or the likelihood of some specific outcome occurring in association with a current customer interaction, be it a telephone-based interaction, web-based interaction, or other type of electronic-assisted interaction. For example, it may be useful for a company to predict a customer satisfaction score, an average handling time, or amount of sales during a customer interaction, taking into account the activities, outcomes, and experiences from prior interactions. Further examples of outcomes associated with a customer include whether a customer will terminate his or her account, whether the customer will purchase a product, whether a customer will pay an outstanding bill, whether a customer is a fraudster, and whether a customer will initiate additional subsequent interaction sessions regarding the same issue, or a combination thereof. This is a non-exhaustive list and additional and/or different outcomes related to a customer or customer interaction may be identified; paragraph 0112, discussing that once an outcome to be predicted is identified, a predictive model operable to predict the identified outcome or the likelihood of the identified outcome occurring is built using the input data as standardized. Specifically, in one embodiment, the standardized input data is fed into predictive analytics software that creates a binary logistic regression model based on the input data. The regression model identifies the variables within the input data that correlate to the identified outcome in the context of a customer interaction. Further, a regression coefficient may be assigned to each identified variable to establish the contribution of the variable to the predicted outcome. As an example, the model may indicate that whether a customer will cancel his or her service is correlated to the customer's personality, the number of distress events during a call, the agent's experience, and the customer's tenure, and assign a coefficient to each of the four variables. As will be discussed in detail below, data points associated with each of these four factors may be collected during a current customer interaction, aggregated at the customer level as needed, and multiplied by their respective coefficients to generate a prediction score indicative of the likelihood that a customer will cancel his or her service).
The Traba-Low combination does not explicitly teach wherein said one or more future interaction events comprise interaction termination. However, McGann in the analogous art of o interaction routing system teaches this concept. McGann teaches:
wherein said one or more future interaction events comprise interaction termination (paragraph 0137, discussing a module that proceeds to calculate a predicted wait time associated with each of the candidate agents. According to one embodiment, the predicted wait time is based on the agent's current status and the threshold customer patience for the identified interaction intent type. The agent's current status may include information on whether the agent is available or not to handle the interaction. If the agent is not currently available, the current status may include information on the interaction that be is currently handling, such as interaction type, intent identified for the interaction, handling time, and the like. According to one embodiment, the wait time is set to be 0 if the agent is currently available, and set to be −1 if the agent is currently busy and not expected to be available until after a time that the caller is predicted to abandon the interaction. For other cases, the wait time is a function of the predicted availability of the agent).
The Traba-Low combination describes features related to prediction of user actions and routing. McGann is directed towards optimized routing of interactions to contact center agents based on machine learning. Therefore they are deemed to be analogous as they both are directed towards prediction systems. 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 Traba-Low combination with McGann because the references are analogous art because they are both directed to solutions for predictive modeling and interaction routing, which falls within applicant’s field of endeavor (system and method for distributing interaction data to agents), and because modifying the Traba-Low combination to include McGann’s feature for including wherein said one or more future interaction events comprise interaction termination, in the manner claimed, would serve the motivation of better meeting real-time needs or desires of the contact center and allowing more efficient use of resources of the contact center (McGann, paragraph 0034); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 8, the Traba-Low combination teaches a method according to claim 6. While Traba describes determining an agent predicted handle time (paragraphs 0120-0123), the Traba-Low combination does not explicitly teach wherein evaluating said interaction capacity of said agent comprises comparing an interaction latency of an agent to a threshold value. However, McGann in the analogous art of interaction routing systems teaches this concept. McGann teaches:
wherein evaluating said interaction capacity of said agent comprises comparing an interaction latency of an agent to a threshold value (paragraph 0059, discussing that it may be desirable to route the interaction to the second-best agent if the second-best agent has been idle for a maximum amount of time, or if the occupancy of the optimal agent is higher than a threshold value; paragraph 0125, discussing that in maximizing the total expected reward for multiple interaction in the queue, the alternate reward maximization module leverages information on customer patience and forecast agent availability if certain agents are not currently available to receive an interaction assignment. Based on the information, the alternate reward maximization module determines whether it should hold-off routing an interaction to get a more optimal agent assignment…For example, information on the impact of caller wait time on the final NPS score may be used to calculate the customer patience number of a call intention type. In this regard, a wait time threshold may be identified for each intention type after which the NPS score drops. For instance, assume that for one of the intention types, it is observed that its Average Handling Time is 619 seconds and average caller Wait Time is 40 seconds (with 70% of the calls answered in less than 1 seconds), and the NPS score drops sharply only after 190 seconds. The customer “patience number” for this intention type may be set as 180 seconds. This example illustrates that customers are prepared to wait (for some time) for the right agent rather than settle for a lesser skilled agent. Therefore, given reliable short term forecasts of agent availability, and an estimate of customer's patience or tolerance level for waiting before abandoning or negatively impacting outcomes, that time flexibility may be exploited to do a more optimal interaction-agent match; paragraph 0137, discussing that the module proceeds to calculate a predicted wait time associated with each of the candidate agents. According to one embodiment, the predicted wait time is based on the agent's current status and the threshold customer patience for the identified interaction intent type. The agent's current status may include information on whether the agent is available or not to handle the interaction. If the agent is not currently available, the current status may include information on the interaction that be is currently handling, such as interaction type, intent identified for the interaction, handling time, and the like. According to one embodiment, the wait time is set to be 0 if the agent is currently available, and set to be −1 if the agent is currently busy and not expected to be available until after a time that the caller is predicted to abandon the interaction…; paragraph 0126).
The Traba-Low combination describes features related to prediction of user actions and routing. McGann is directed towards optimized routing of interactions to contact center agents based on machine learning. Therefore they are deemed to be analogous as they both are directed towards prediction systems. 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 Traba-Low combination with McGann because the references are analogous art because they are both directed to solutions for predictive modeling and interaction routing, which falls within applicant’s field of endeavor (system and method for distributing interaction data to agents), and because modifying the Traba-Low combination to include McGann’s feature for including wherein evaluating said interaction capacity of said agent comprises comparing an interaction latency of an agent to a threshold value, in the manner claimed, would serve the motivation of better meeting real-time needs or desires of the contact center and allowing more efficient use of resources of the contact center (McGann, paragraph 0034); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 9, the Traba-Low combination teaches a method according to claim 8. Although not explicitly taught by the Traba-Low combination, McGann in the analogous art of interaction routing systems teaches wherein said agent is available for receiving said new interaction request when said interaction latency is below said threshold value and wherein said agent is unavailable for receiving said new interaction request when said latency is above said threshold value (paragraph 0125, discussing that in maximizing the total expected reward for multiple interaction in the queue, the alternate reward maximization module leverages information on customer patience and forecast agent availability if certain agents are not currently available to receive an interaction assignment. Based on the information, the alternate reward maximization module determines whether it should hold-off routing an interaction to get a more optimal agent assignment. According to one embodiment, the customer patience value is for particular intention type. The patience value may be a value that is derived/calculated by the alternate reward maximization module based on observed abandonment rates, service times, NPS scores, and the like…In this regard, a wait time threshold may be identified for each intention type after which the NPS score drops. For instance, assume that for one of the intention types, it is observed that its Average Handling Time is 619 seconds and average caller Wait Time is 40 seconds, and the NPS score drops sharply only after 190 seconds. The customer “patience number” for this intention type may be set as 180 seconds. This example illustrates that customers are prepared to wait (for some time) for the right agent rather than settle for a lesser skilled agent. Therefore, given reliable short term forecasts of agent availability, and an estimate of customer's patience or tolerance level for waiting before abandoning or negatively impacting outcomes, that time flexibility may be exploited to do a more optimal interaction-agent match; paragraph 0126, discussing a timing diagram showing a window of opportunity that may be exploited by the alternate reward maximization module for a better interaction-agent matching. In the example, although Agent-1 is available when the caller entered the queue, the interaction is not assigned to Agent-1 due to a potential of a low reward compared to the reward that may be achieved by assigning the interaction to Agent 2, who is currently unavailable. Instead, the alternate reward maximization module is configured to wait to assign the interaction to Agent-2 who is expected to fetch higher reward. In the above toy example, suppose A1 is available, and A2 is not available, but would be available in an acceptable time (e.g. within the caller's patience threshold). In that example, the alternate reward maximization module chooses a more optimal assignment with a total reward of 1.3, where the Enquire_Bill interaction is assigned to A1, and the Report_Fault interaction is assigned to A2; paragraph 0134, discussing that in the above toy example, assume that agent A2 is currently on a call, and the alternate reward maximization module predicts that he will be freed up in the next time tick that is within the customer's patience threshold. Such prediction of when the agent is to become available may be based, for example, on analysis of estimated handling times for the call intent type. Other algorithms for forecasting agent availability may also be invoked to predict when the agent it to become available…; paragraph 0138, discussing that the module calculates, for each interaction and each candidate agent, a time-discounted reward that is expected to be achieved by assigning the interaction to the agent…In this regard, the discount factor gives more importance to rewards that are to be achieved earlier rather than later in time. According to one embodiment, the discount factor is a value that is computed as a function of the wait time. For example, the smaller the wait time, the smaller the discount factor value. If the agent is immediately available, the expected reward computed for that agent is not discounted. On the other hand, if the interaction is likely to be abandoned before the agent becomes available (e.g. the time the agent is predicted to become available is longer than the time that the customer is predicted to hold prior to abandoning the interaction, or in other words, the wait time exceeds a predicted customer patience threshold), the expected reward is calculated to be 0).
The Traba-Low combination describes features related to prediction of user actions and routing. McGann is directed towards optimized routing of interactions to contact center agents based on machine learning. Therefore they are deemed to be analogous as they both are directed towards prediction systems. 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 Traba-Low combination with McGann because the references are analogous art because they are both directed to solutions for predictive modeling and interaction routing, which falls within applicant’s field of endeavor (system and method for distributing interaction data to agents), and because modifying the Traba-Low combination to include McGann’s feature for including wherein said agent is available for receiving said new interaction request when said interaction latency is below said threshold value and wherein said agent is unavailable for receiving said new interaction request when said latency is above said threshold value, in the manner claimed, would serve the motivation of better meeting real-time needs or desires of the contact center and allowing more efficient use of resources of the contact center (McGann, paragraph 0034); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 10, the Traba-Low-McGann combination teaches a method according to claim 9. Traba further teaches wherein when said agent is unavailable for receiving an interaction request, identifying another agent for receiving said interaction request (paragraph 0018, discussing using dynamic metric optimization to leverage caller and agent information along with advanced analytics to determine, in real-time, the best customer metric(s) to optimize the routing of a customer to a suitable, available agent; paragraph 0032, discussing that if the agent is not available, a second agent who is presently available and who is almost as suitable could be selected and the ACD could route the communication to that second agent; paragraph 0101, discussing that the predictive calculator identifies available agents for handling the customer communication…).
Claim 12 recites substantially similar limitations that stand rejected via the art citations and
rationale applied to claim 2, as discussed above.
Claim 18 recites substantially similar limitations that stand rejected via the art citations and
rationale applied to claim 8, as discussed above.
Claim 19 recites substantially similar limitations that stand rejected via the art citations and
rationale applied to claim 9, as discussed above.
35. Claims 5 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Traba in view of Low, in further view of Ristock et al., Pub. No.: US 2012/0197678 A1, [hereinafter Ristock].
As per claim 5, the Traba-Low combination teaches a method according to claim 1. Although not explicitly the Traba-Low combination, Ristock in the analogous of interaction processing systems teaches wherein estimating said one or more future interaction events comprises sequential initiation and termination of said one or more interactions, thereby maintaining a concurrent assignment of interaction requests to said agent (paragraph 0013, discussing that the natural capacity value for a worker is specific to multitasking and may be calculated to reflect two or more different task types being processed by the worker concurrently; paragraph 0014, discussing that a method is provided for optimizing the time of a worker in a multitasking environment comprising the steps (a) establishing a natural capacity value defining the ability of the worker to work one or more specific task types concurrently without developing a significant backlog, (b) during a work period, routing or assigning a task to the worker and monitoring the load on the worker's natural capacity value established in step (a), (c) upon observance that the worker is performing significantly under the natural capacity value for that worker, routing or assigning a second task to the worker and monitoring the load on the worker's natural capacity value; paragraph 0044, discussing that the natural capacity is the measure of how many interactions can be process concurrently without delaying the interaction handling or creating a backlog. It is clear in this example that routing a second interaction to the same agent for concurrent processing will reduce the idle time of the agent significantly and hopefully without sacrificing the quality of service (QoS) provided by the agent; paragraph 0055, discussing that if is determined that there is an agent working significantly below the natural capacity value for that agent for the type interaction, a second interaction request of the type is routed to the agent and the agent processes the interactions concurrently; paragraphs 0045, 0047)
The Traba-Low combination describes features related to prediction of user actions and routing. Ristock is directed towards methods and apparatus for managing interaction processing. Therefore they are deemed to be analogous as they both are directed towards interaction management systems. 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 Traba-Low combination with Ristock because the references are analogous art because they are both directed to solutions for interaction routing, which falls within applicant’s field of endeavor (system and method for distributing interaction data to agents), and because modifying the Traba-Low combination to include Ristock’s feature for including wherein estimating said one or more future interaction events comprises sequential initiation and termination of said one or more interactions, thereby maintaining a concurrent assignment of interaction requests to said agent, in the manner claimed, would serve the motivation of effectively utilizing an agent's time (Ristock, paragraph 0036); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Claim 15 recites substantially similar limitations that stand rejected via the art citations and
rationale applied to claim 5, as discussed above.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
K et al., Pub. No.: US 2023/0153340 A1 – describes machine learning for multi-channel interaction workflows.
Jain et al., Pub. No.: US 2019/0087746 A1 – describes methods for automatic and intelligent incident routing using machine learning.
Lee, Seung Man, and Amy R. Pritchett. "Predicting interactions between agents in agent-based modeling and simulation of sociotechnical systems." IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and Humans 38.6 (2008): 1210-1220 – describes providing a timing and prediction mechanism for the accurate modeling of interactions among agents, correspondingly increasing the computational efficiency of agent-based simulations.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DARLENE GARCIA-GUERRA whose telephone number is (571) 270-3339. The examiner can normally be reached M-F 7:30a.m.-5:00p.m. 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, Brian M. Epstein can be reached on (571) 270-5389. 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.
/Darlene Garcia-Guerra/
Primary Examiner, Art Unit 3625
Prosecution Insights