DETAILED ACTION
This communication is in response to the Arguments and Remarks filed on 3/20/2026. Claims 1-23 are pending and have been examined. Hence, this Action has been made FINAL.
Any previous objection/rejection not mentioned in this Office Action has been withdrawn by the examiner.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 13, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 12/23/2026, 2/5/2026, and 3/20/2026 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements are being considered by the examiner.
Priority
Applicant claims the benefit of US Provisional Application No. 63/497,091, filed April 19, 2023. Claims 1-23 have been afforded the benefit of this filing date.
Response to Arguments
Applicant's arguments filed 3/20/2026 have been fully considered but they are not persuasive.
With respect to the 35 U.S.C. 101 rejections, the applicant asserts that as amended, claim 1 is directed to a specific computer-implemented architecture for reducing computational overhead in language data processing through a multi-stage pipeline implemented on a centralized platform. Now-amended claim 1 does not merely recite the generalized concept of sending a message, reviewing a response, and deciding what to do next. Rather, now-amended claim 1 requires the limitations described above. These limitations are technological in nature, i.e., computer-specific operations performed on machine-represented language data, not the type of observation, evaluation, or judgment that can practically be performed in the human mind. In this way, the claim recites a coordinated, machine-executed pipeline in which intermediate data generated at one stage is shared and reused at a later stage through a centralized platform.
Examiner respectfully disagrees, a phrase such as “reducing computational overhead” is arguing the affect of the invention rather than making arguments about the individual steps. Also, “implemented on a centralized platform” merely states that the mental process is being applied via a computing device, it does not show that it is not a mental process. The limitations are not inherently technological in nature as things such as creating a message, receiving feedback, labelling, and making determinations on next steps can all be performed by the human mind. The technological aspects of the claim language serve as additional elements to apply the method to a computing device, rather than things that are essential to the system/method. The intermediate data is not enough to say the trained generator is implemented throughout the system as a human mind is capable of creating additional relevant information that is useful in classification. The train generator/classifier are merely outputting things that the human mind is capable of doing itself.
The applicant asserts that in rejecting independent claims 1, 12, and 13, the Office Action analogized the claims to a receptionist reviewing background information, drafting a message, evaluating a response, and deciding a next step. See 12/23/25 Office Action at p. 3. That analogy does not fit the language of the now-amended claims as a human receptionist does not, and cannot, perform each and every limitation in the now-amended independent claims. More specifically, the human mind does not generate and share machine intermediate data through a centralized platform between a trained generator and a trained classifier, and does not perform classification based on collected interaction data together with intermediate data generated by a trained generator and shared across stages of a machine pipeline. Therefore, the amended claims are directed to a specific technical workflow in which multiple trained components operate on language data through inter- stage data sharing on a centralized platform, not to an abstraction that can be carried out in the human mind or by a generic "receptionist". Accordingly, the claims do not recite an abstract idea
Examiner respectfully disagrees, the human mind does not need to do these things using a centralized platform, trained generator, and trained classifier. The point of the analysis is to show that the human mind can perform the method and then evaluate the above-mentioned things as additional components to determine if they are sufficiently implemented with claim. The additional components, as currently claimed, are only serving the purpose of replacing the human mind for an action that the human mind is capable of performing. Overall, if you were to replace the additional components with a human mind the method would still be performed the exact same way and produce the same outcome.
With respect to the 35 U.S.C. 102 rejections, the applicant asserts that although Champaneria generally states that its system may "...increase recruiter productivity by automating the communication process between the recruiter, candidate, and hiring man-ager, and by providing automatic responses to common questions" and "reduce the time and expense required to train a recruiter in a particular field..." of which is said to be performed using a laundry list of machine learning techniques, those disclosures do not teach the presently claimed invention as in now-amended claims 1, 12, and 13. Champaneria at col. 8, Ins. 63-67 through col. 9, Ins 1-7; See col. 10, Ins. 11- 35. Champaneria's disclosure regarding reducing recruiter training time and expense concerns a human-facing business benefit, namely, that a recruiter may rely less on personal experience or specialized recruiting knowledge. That is materially different from the presently claimed reduction of "computational overhead in language data processing through a multi-stage pipeline implemented on a centralized platform". The amended independent claims are not directly merely to making recruiting easier or faster for a human user. Rather, the now-amended claims require a centralized platform in which intermediate data generated by a trained generator is shared and alter used by a trained classifier in determining a label for collected interaction data. Champaneria does not disclose that kind of inter-stage technical arrangement
Examiner respectfully disagrees, once again, “reducing computation overhead” is arguing the affect of the invention rather than steps of the system/method. When showing novelty, the steps are what matters regardless of if the intention of the inventions are different. Furthermore, Champaneria does teach a “centralized platform” around an AI Bot (Col. 56, Lines 36-45) and intermediate data is generated in the form of time windows for user interaction data (Col. 24, Lines 9-16). Further details on why the Champaneria teaches the amended claims can be found below.
The applicant further asserts that Champaneria may describe sending job-related messages to candidates, analyzing candidate responses, and, in some instances, using Al or NLP techniques in connection with those activities. However, Champaneria does not disclose that the classifier determines a label based on both: (1) the collected interaction data; and (2) "...at least a portion of intermediate data generated by the trained generator and shared via the centralized platform", as in now-amended independent claims 1, 12, and 13. Nor does Champaneria disclose that such sharing is part of a multi-stage pipeline implemented on a centralized platform for reducing computational overhead in language data processing. At most, Champaneria discloses generalized recruiting automation and productivity improvements. Champaneria does not disclose the specific reuse of intermediate generator-stage data by a later classifier-stage operation, as in the now- amended independent claims.
Examiner respectfully disagrees, Champaneria creates labels based on interaction data (response from users) and intermediate data (timers for responses that can alter how the user interaction is treated) on a central platform (AI bot) (Col. 24, Lines 9-16) (Col. 56, Lines 36-45). Champaneria also teaches the multi-stage platform as a step wise process of creating a message, receiving a response, and performing actions based on the response data is a multi-stage process (Figs. 5-6). The intermediate data is reused for classifying as a time limit being reached will cause the system to label it “no reply” and take specific action it would not have otherwise taken (Col. 24, Lines 9-16)
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-23 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claims 1, 12, and 13 recite A method for reducing computational overhead in language data processing through a multi-stage pipeline implemented on a centralized platform, comprising: generating, based on received input data, a targeted message for a lead using a [trained generator]; causing projection of the generated targeted message via a [user device] associated with the lead; collecting interaction data generated in response to the projection of the targeted message to the lead: determining, using a [trained classifier], based on the collected interaction data and at least a portion of intermediate data generated by the trained generator and shared via the centralized platform, at least a label for the interaction data; determining a next step based on the determined at least a label, wherein the next step is determined with respect to the lead; and performing the next step upon determination.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. This can be explained as being a mental process using an example of a receptionist. A receptionist could receive input data in the form of their employer asking them to contact clients. The receptions could create a targeted message to send to them using their prior knowledge and any additional information their employer told them. They could then send this message to the clients. Based on the reply received, the receptionist could categorize the types of responses (Ex: interested vs. not interested). In this case the interaction data is represented by a reply message received from the clients. Then, based on the category, the receptionist could perform a next step (Ex: if interested, schedule a phone call). If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. In particular, claims 1, 12, and 13 recite the additional components of a trained generator, a user device, and a classifier. The trained generator is detailed in paragraph 59 of the specification and is described generically with potential architectures of a DNN, RNN, or LLM such as GPT. The user device is described on paragraph 37 of the specification with generic example devices listed. The classifier is detailed in paragraph 48 of the specification and generically states taking the form of a neural network, gradient-boosted algorithm, and/or a supervised machine learning algorithm. Claim 12 specifically lists the additional component of a non-transitory computer readable storage medium. The non-transitory computer readable storage medium is detailed in paragraph 113 of the specification with a generic description of the component. Claim 13 specifically lists the additional component of a memory. The memory is detailed in paragraph 111 of the specification with a generic description of the component. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
As to claims 12 and 13, method claim 1 is related as method using same, with each claimed element's step corresponding to the claimed apparatus function. Accordingly claims 12 and 13 are similarly rejected under the same rationale as applied above with respect to method claim 1.
Claims 2 and 14 recite receiving a list of a plurality of potential leads, wherein the list of the plurality of potential leads includes a subset of potential leads that are ranked based on scores, wherein each potential lead in the list of the plurality of potential leads has a score above a predetermined threshold value; and selecting lead from the list of the plurality of potential leads.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. From the independent claim example, the receptionist could get a list of potential clients/leads and information about them and rank the clients based on that information. For example, they could have criteria such as age, gender, and proximity and if the lead is in the predetermined range for all of them, they can be selected to receive a message. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 3 and 15 recite iteratively repeating creating, causing projection, and processing the collected interaction data in near real-time.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The receptionist from the independent claim could repeat this process for as many potentially clients as they need. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 4 and 16 recite wherein the trained generator is a customized language model that is trained for at least one of: a company, an entity, an industry, and a topic.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind is capable of making a design decision such as the type of data to use in training a model. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 5 and 17 recite wherein the topic is a context of a subject matter in the language data, wherein the language data includes at least one topic.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind is capable of making the design decision to train a language model using language data associated with a topic. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 6 and 18 recite wherein the creating the targeted message further comprises: extracting relevant data using a trained language model from the input data, wherein the input data is expressed as vector embeddings; formatting the extracted relevant data to create a unified data format, wherein formatting includes splitting data into fixed-size data chunks; creating a prompt for the trained generator, wherein the prompt includes a command, background details, and textual data of the formatted relevant data; and feeding the prompt into the trained generator.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. From the independent claim example, the receptionist would be capable of extracting relevant data by interpreting the instructions their employer gave them and/or reading the list of potential clients. Furthermore, the human mind is capable of creating a prompt using information from various sources. A human could understand input data in a vector embedding by formulaically converting it to a human understandable form. They could then format the data into predefined chunk sizes by manually grouping them into that size. Then a human could take these groups and use background knowledge of the input and text to create a prompt. For example, creating a template for responses that include spaces for all the relevant information found in the input and text. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 7 and 19 recite wherein the targeted message is caused to be projected as at least one of: a text, an audio, a video, an image, a multimedia, and a virtual form.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The receptionist from the independent claim example would be capable of formatting their message to a text, an audio, a video, an image, a multimedia, or a virtual form. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 8 and 20 recite collecting feedback data from at least one stage of the multi-stage pipeline; and applying the feedback data to the trained generator to update the trained generator.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. From the independent claim example, the receptionist would be capable of identifying if they made any mistakes or received any negative feedback about the process. They could then learn from this in order to perform it better in the future. Furthermore, the human mind is capable of making a design decision of using feedback data as the data used to update a model. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 9 and 21 recite wherein the classifier is a multi-label classifier that applies at least one of: a neural network, a gradient-based algorithm, and a supervised machine learning algorithm.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind is capable of making the design decision to use either a neural network, a gradient-based algorithm, or a supervised machine learning algorithm for a classifier. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 10 and 22 recite wherein the next step includes scheduling a meeting with the lead, wherein scheduling further comprises: retrieving a lead calendar and a user calendar; identifying a potential meeting time slot by applying an algorithm to the retrieved lead calendar, retrieved user calendar, and the lead data; and causing a display of a reminder, wherein the reminder is generated based on the identified potential meeting time slot.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. From the independent claim example, a receptionist would be capable of checking their employers schedule, a client’s schedule, and any information they know about the client to find a meeting time where everyone is available. The receptionist could then write a message inviting the client to a meeting as well as follow up messages to remind the client and their employer about the meeting. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 11 and 23 recite wherein the display is presented as a part of a sales pipeline that indicates an engagement progress.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. From the independent claim example, when the receptionist sends a meeting reminder to their employer, they could include how far along in the sales process the client currently is. For example, if they are currently ready to submit an offer. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional elements that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-3, 7-8, 12-15, and 19-20 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by US Patent Publication US 10318927 B2 (Champaneria).
Regarding Claims 1, 12, and 13, Champaneria teaches A method for reducing computational overhead in language data processing through a multi-stage pipeline implemented on a centralized platform, comprising:
(Turning now to exemplary FIG. 1, FIG. 1 displays an exemplary embodiment of a flowchart depicting a method of automating a recruiting process. Such a method may include several steps, including a job entry step 1, a sourcing step 2, a matching step 3, a contact step 4, a response step 5, a submission step 6, a review step 7, and an offer and acceptance step 8.) (Col. 9, Lines 8-14).
(In an exemplary embodiment, an AI BOT 1004 may be a central part of the system, and may manage the other components of the system, such as a natural language processing engine 1006 (which may be, for example, a NLP engine wholly incorporated into the AI BOT, or may be, for example, an NLP engine API or another connection to an external NLP engine). The AI BOT 1004 may also be linked to a database 1002.) (Col. 56, Lines 36-45).
The multi-stage pipeline is represented by the steps of the method as presented in the flowchart in Fig. 1. The centralized platform is represented by the AI Bot and can be seen in Fig. 13
generating, based on received input data, a targeted message for a lead using a trained generator;
(in a first step 11 of a job entry step 1, a job description may be input into the system. In some embodiments, the text of the job description, and any other information that has been provided by the hiring manager that may be outside of the job description, may be parsed by the system (for example, by a system configured to use natural language processing) and key points, concepts, and requirements of the job may be identified from the job description.) (Col. 9, Lines 25-33).
(The system may first perform an initial contact step 41, in which information about the job is sent to a candidate, along with a link enabling the candidate to apply for the job, via some form of electronic media or otherwise.) (Col. 23, Lines 43-46).
The system receives input data in the form of a job description and then creates a targeted message to send to a lead (candidate in this case). This message can contain information from the job description and a link to apply to the job.
causing projection of the generated targeted message via a user device associated with the lead;
(In a next step 13 of a job entry step 1, the job description may be loaded into the system based on the parsing of step 11 (and optionally the manual review of step 12). A data point matrix may then be created in the system for subsequent matching of the information in the job description to information of one or more potential candidates for the position) (Col. 9, Lines 43-49).
(The system may first perform an initial contact step 41, in which information about the job is sent to a candidate, along with a link enabling the candidate to apply for the job, via some form of electronic media or otherwise.) (Col. 23, Lines 43-46).
The targeted messaged is sent to the candidate, this a projection is made to the user’s device. Fig. 5 shows the initial contact step in the process as well.
collecting interaction data generated in response to the projection of the targeted message to the lead:
(When a candidate has been contacted by some communication channel, and when the candidate responds by some communication channel (which may or may not be the same communication channel as was used to initially contact the candidate, if desired; for example, the candidate may initially be contacted by email and may elect to respond to the contact by calling a telephone number provided in the email) the system may be configured to handle and respond to the candidate's response 50.) (Col. 29, Lines 56-64).
The response (interaction data) is in direct reply to the initial targeted message as can be visualized in Figs. 5 and 6.
determining, using a trained classifier, based on the collected interaction data and at least a portion of intermediate data generated by the trained generator and shared via the centralized platform, at least a label for the interaction data;
(the system may determine from the candidate's response (for example, by parsing an email of the candidate, or otherwise analyzing their response) whether the candidate has responded to the initial contact message provided by the system by accepting an offer provided in the contact message (such as, for example, an offer to apply for a position), by rejecting an offer provided in the contact message (for example, by sending a message declining an offer to apply for a position), or by asking a question or requesting more information.) (Col. 29, Lines 65-67 to Col. 30, Lines 1-7).
(According to an exemplary embodiment, a follow-up rule 411 may incorporate a predefined timeframe for follow-up or one or more other checks in order to ensure that conversations do not become “stuck in limbo.” For example, according to an exemplary embodiment, a response time of 24 hours may be predefined, or some other response time may be predefined, after which the system will consider the candidate not to have responded to the inquiry.) (Col. 24, Lines 9-16).
(For example, according to an exemplary embodiment, a path-based system for an AI BOT that is constructed around a map of a conversation process (such as is described above) may operate as follows. A candidate may first ask the AI BOT, “Hey, this opportunity sounds great; who is the client?” This may equate to an intent configured as #RequestInfoClient (which may be a business intent), for which the entity may be the #Client. This may trigger the AI BOT to retrieve information about the client, and provide it in the form of an answer to the candidate.) (Col. 37, Lines 21-28).
The interaction data takes the form of a response message (or lack thereof) from the candidate. The response is then labeled as accepting the offer, rejecting the offer, or asking further questions. The logic behind this classifying and the next steps can be seen in Figs. 5 and 6. This process also includes intermediate information generated by the system such as response times that can alter the label by marking it as no reply which can necessitate a follow up message. Furthermore, the AI Bot in this scenario is performing both the tasks of the trained classifier and trained generator as it is creating messages to send to the user and interpreting their responses to determine a next action.
determining a next step based on the determined at least a label, wherein the next step is determined with respect to the lead;
(If the response of the candidate is to indicate interest in a position or an offer, the system may be configured to take action to finalize the candidate 51. In an exemplary embodiment, should the system receive a communication from the candidate indicating that the candidate is interested in the position, the system may send a request for any other remaining required information to the candidate.) (Col. 32, Lines 6-12).
Fig. 6 shows a flow chart of a next step being determined based on the label that was assigned to the response. It can be seen in Fig. 6 that when finalizing a candidate if they are still a match you “Go to 6”. 6 is shown in Fig. 7 as a review of the candidate and an interview being scheduled.
and performing the next step upon determination.
Once more, Figs. 6 and 7 depict the next steps that can be performed such as scheduling an interview, evaluating a question and responding, or stopping the processing of the candidate.
Regarding Claims 2 and 14, Champaneria teaches the method of claims 1 and 13 and, receiving a list of a plurality of potential leads,
(in an automatic sourcing step 211, candidates and résumés of candidates may be sourced automatically from job boards and from social media, as well as any other appropriate sources (such as, for example, résumés submitted by a candidate to an employment page of a company website, which may, for example, be forwarded to the recruiter for review).) (Col. 12, Lines 31-37).
A list of potential candidates is sourced for the next steps in the method.
wherein the list of the plurality of potential leads includes a subset of potential leads that are ranked based on scores,
(In an exemplary embodiment, a data-point matrix may then be created for each résumé. According to an exemplary embodiment, a data-point matrix may be created for each résumé for each subsequent matching and ranking of that résumé.) (Col. 19, Lines 1-5).
(In an exemplary embodiment, résumés may be scored by a semantic engine utilizing machine learning concepts. The semantic engine may apply weight to certain requirements, which may be specified by a hiring organization or recruiter or may be derived from the job description.) (Col. 21, Lines 39-43).
The candidates are ranked based on their resumes and criteria set by the organization in the input job description.
wherein each potential lead in the list of the plurality of potential leads has a score above a predetermined threshold value;
(For example, these requirements may include (but may not be limited to) a title search (i.e. a search of job titles), the date on which the résumé was last updated, particular skills (or synonyms of those skills) that may be listed in the résumé, a number of required years of experience in a particular field, a number of required years of experience in a particular industry, a number of years of experience associated with a skill or with a particular set of skills, employment continuity, salary history, salary requirements, geographical proximity, social footprint (for example, the connections of the user on one or more social media websites), activity of social media (for example, the postings of the user on one or more social media websites), willingness to relocate, and any additional requirements that have been provided by the hiring company or which have been deemed relevant to the process.) (Col. 21, Lines 43-58).
These requirements listed as things the candidates will be ranked on show threshold values such as minimum/maximum years of experience, a minimum proximity, and/or minimum salary requirements.
and selecting lead from the list of the plurality of potential leads.
(The semantic engine may then determine which candidates have the highest scores, and may select candidates accordingly. For example, in an exemplary embodiment, a semantic engine may rank candidates based on the data point matrix scores of the candidates in each of the above areas (or in each of the above areas that are actually considered) and based on any other criteria, as desired. The semantic engine may then use the ranked list of candidates to select one or more candidates to progress to a next stage of hiring; for example, in an exemplary embodiment, the semantic engine may take the top X most highly ranked candidates.) (Col. 22, Lines 1-11).
The top ranked candidates are selected to go to the following steps of the method.
Regarding Claims 3 and 15, Champaneria teaches the method of claims 1 and 13 and, iteratively repeating creating, causing projection, and processing the collected interaction data in near real-time.
(If it is determined that the candidate is unsuitable based on what they have disclosed, the system may proceed to a removal step 53. However, if it is unclear whether the candidate is suitable or not, then an auto-response 54 may be generated and may be sent to the candidate requesting more information) (Col. 33, Lines 19-24).
It can be seen in Fig. 6 that the system repeats itself after responding to questions or if it identified that additional information is needed. In these cases, another targeted message is sent to the user and a response is received and classified.
Regarding Claims 7 and 19, Champaneria teaches the method of claims 1 and 13 and, wherein the targeted message is caused to be projected as at least one of: a text, an audio, a video, an image, a multimedia, and a virtual form.
(In a next step, an AI BOT may send a campaign message 92 to the candidate. In an exemplary embodiment, this campaign message 92 may be sent over a communication channel that has been associated with a created ConversationID 91; for example, in a first exemplary embodiment, when a ConversationID has been created for a text messaging protocol 911, a campaign text message 921 may be sent to the candidate. This may likewise be the case for campaign emails 922 to be sent to the candidate, campaign social media messages 923 to be sent to the candidate, or any other campaign messages 924, 925 to be sent to the candidate.) (Col. 54, Lines 31-41).
The targeted message can be sent via text, email, or social media. This shows a projected targeted message form in at least text or virtual form.
Regarding Claims 8 and 20, Champaneria teaches the method of claims 1 and 13 and, collecting feedback data from at least one stage of the multi-stage pipeline; and applying the feedback data to the trained generator to update the trained generator.
(In an exemplary embodiment wherein, the candidate rejects the job offer, a reason may be requested from the candidate as to why the job offer was rejected 83. In an exemplary embodiment, such a reason may be used to better train an AI bot used in previous steps.) (Col. 48, Lines 38-42).
When a candidate rejects a job offer in one of their responses throughout the method, the system can use this to train the AI to better select/message candidates in the future.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 4-6 and 16-18 are rejected under 35 U.S.C. 103 as being unpatentable over US Patent Publication US 10318927 B2 (Champaneria) in view of US Patent Publication US 12229522 B2 (Gajek et al.).
Regarding Claims 4 and 16, Champaneria teaches the system of claims 1 and 13.
While Champaneria does teach using a trained AI Bot to generate targeted messages, it does not explicitly teach: wherein the trained generator is a customized language model that is trained for at least one of: a company, an entity, an industry, and a topic.
However, Gajek et al. teaches wherein the trained generator is a customized language model that is trained for at least one of: a company, an entity, an industry, and a topic.
(Large language models (LLMs) such as OpenAI's ChatGPT can exhibit human-level abilities to answer questions and perform tasks. One such important application is answering a question based on information contained in a potentially very large set of documents.) (Col. 1, Lines 27-31).
(The text generation interface system may then extract the training data from the completed query division prompt, for instance by parsing JSON included in the completed request. An example of a prompt template for generating training data in the context of legal contracts is as follows: The task is to generate queries and <N> variations thereof that would retrieve a specific contract clause in a retrieval system comprised of a large collection of contracts. Given the following clause for clause type <clause_type>: <clause> queries: <|endofprompt|>) (Col. 31, Lines 62-67 to Col. 32, Lines 1-7).
Gajek et al. teaches a system that creates targeted messages where industry and topic training data is used (in this case, legal contracts).
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the multistage communication pipeline as taught by Champaneria to use a use a language model trained on industry/topic data for generated messages as taught by Gajek et al. This would have been an obvious substitution as Champaneria is already using an AI Bot to generate responses to a candidate’s questions and Gajek et al. presents an alternative method of answering these questions (Gajek et al., Col. 1, Lines 27-31).
Regarding Claims 5 and 17, Champaneria in view of Gajek teaches the system of claims 4 and 16.
Furthermore, Gajek et al. teaches wherein the topic is a context of a subject matter in the language data, wherein the language data includes at least one topic.
(The text generation interface system may then extract the training data from the completed query division prompt, for instance by parsing JSON included in the completed request. An example of a prompt template for generating training data in the context of legal contracts is as follows: The task is to generate queries and <N> variations thereof that would retrieve a specific contract clause in a retrieval system comprised of a large collection of contracts. Given the following clause for clause type <clause_type>: <clause> queries: <|endofprompt|>) (Col. 31, Lines 62-67 to Col. 32, Lines 1-7).
In Gajek et al. the training data topic is a subject matter (above example shows legal contracts). This topic is included in the language data as the data comes from a collection of legal contracts.
Regarding Claims 6 and 18, Champaneria teaches the system of claims 1 and 13.
While Champaneria does teach extracting relevant data from the input and vectorizing it for an AI bot to generate targeted messages, it does not explicitly teach: wherein the creating the targeted message further comprises: extracting relevant data using a trained language model from the input data, wherein the input data is expressed as vector embeddings; formatting the extracted relevant data to create a unified data format, wherein formatting includes splitting data into fixed-size data chunks; creating a prompt for the trained generator, wherein the prompt includes a command, background details, and textual data of the formatted relevant data; and feeding the prompt into the trained generator.
However, Gajek et al. teaches wherein the creating the targeted message further comprises: extracting relevant data using a trained language model from the input data,
(A text generation interface system may take as input one or more arbitrary documents, process them via optical text recognition, segment them into portions, and process the segmented text via various tasks based on need. Different workflows are provided for different tasks, and this application describes a number of examples of such workflows. In many workflows, an input document is divided into chunks via a chunking technique. Then, chunks are inserted into prompt templates for processing by a large language model such as the GPT-3 or GPT-4 available from OpenAI.) (Col. 3, Lines 37-47).
Input is entered into the text generation interface and extracted via optical text recognition.
wherein the input data is expressed as vector embeddings;
(A trained classification model is determined at 1316 based on the training data. According to various embodiments, any of a variety of classification models may be used. For instance, the classification model may include a text embedding model that positions text in a vector space.) (Col. 32, Lines 17-22).
A text embedding model converts the input text into a vector space.
formatting the extracted relevant data to create a unified data format,
(A text generation interface system may take as input one or more arbitrary documents, process them via optical text recognition, segment them into portions, and process the segmented text via various tasks based on need. Different workflows are provided for different tasks, and this application describes a number of examples of such workflows. In many workflows, an input document is divided into chunks via a chunking technique. Then, chunks are inserted into prompt templates for processing by a large language model such as the GPT-3 or GPT-4 available from OpenAI.) (Col. 3, Lines 37-47).
The input data is divided into chunks that fit prompt templates designed to be input to the language model used for creating targeted messages.
wherein formatting includes splitting data into fixed-size data chunks;
(A text generation interface system may take as input one or more arbitrary documents, process them via optical text recognition, segment them into portions, and process the segmented text via various tasks based on need. Different workflows are provided for different tasks, and this application describes a number of examples of such workflows. In many workflows, an input document is divided into chunks via a chunking technique. Then, chunks are inserted into prompt templates for processing by a large language model such as the GPT-3 or GPT-4 available from OpenAI.) (Col. 3, Lines 37-47).
(According to various embodiments, techniques and mechanisms described herein provide for the division of text into chunks, and the incorporation of those chunks into prompts that can be provided to a large language model. For instance, a large language model may impose a limit of, for instance, 8,193 tokens on a task, including text input, text output, and task instructions. In order to process longer documents, the system may split them.) (Col. 4, Lines 34-41).
The input data is separated into chunks. The size/number of chunks can be predetermined by the language model being used.
creating a prompt for the trained generator,
(In some embodiments, the second subset of the text portions may be identified by providing some or all of the first subset of the text portions to the text generation modeling system in one or more prompts. The prompts may instruct the text generation modeling system to identify which, if any, of the text portions are relevant to determining an answer to the query.) (Col. 6, Lines 30-36).
Prompts are created to be used as input to the language model (trained generator).
wherein the prompt includes a command, background details, and textual data of the formatted relevant data;
(The prompts may instruct the text generation modeling system to identify which, if any, of the text portions are relevant to determining an answer to the query.) (Col. 6, Lines 34-36).
(In some embodiments, a skill may be associated with one or more prompts. For instance, the skill 234 is associated with the prompt templates 236 and 238. A prompt template may include information such as instructions that may be provided to the text generation modeling system 270. A prompt template may also include one or more fillable portions that may be filled based on information determined by the orchestrator 230. For instance, a prompt template may be filled based on information received from a client machine, information returned by a search query, or another information source. Additional details regarding prompt templates are provided with reference to FIGS. 8-10.) (Col. 7, Lines 65-67 to Col. 8, Lines ).
The prompts act as a command to the model as they provide instruction to it. In addition to the fixed sized chunks of textual data used in the prompt, background details such as information from the client’s machine, information from their search query, or other information sources.
and feeding the prompt into the trained generator.
(In some embodiments, the second subset of the text portions may be identified by providing some or all of the first subset of the text portions to the text generation modeling system in one or more prompts. The prompts may instruct the text generation modeling system to identify which, if any, of the text portions are relevant to determining an answer to the query.) (Col. 6, Lines 30-36).
The prompts are fed to the language model.
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the multistage communication pipeline as taught by Champaneria to format the input data in a unified manner and use it to create a prompt for a language model to create targeted messages as taught by Gajek et al. This would have been an obvious improvement as the data can be formatted in a manner that fits the requirements of the language model being used (Gajek et al., Col. 4, Lines 34-47).
Claims 9 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over US Patent Publication US 10318927 B2 (Champaneria) in view of US Patent Application Publication US 20220245340 A1 (Yang et al.).
Regarding Claims 9 and 21, Champaneria teaches the system of claims 1 and 13.
While Champaneria does teach using an AI Bot for the classification, it does not explicitly teach: wherein the classifier is a multi-label classifier that applies at least one of: a neural network, a gradient-based algorithm, and a supervised machine learning algorithm.
However, Yang et al. teaches wherein the classifier is a multi-label classifier that applies at least one of: a neural network, a gradient-based algorithm, and a supervised machine learning algorithm.
(The NLU model may be a deep neural network model trained through supervised learning that applies each of embedding vectors for a plurality of training inquiry messages previously obtained as an input and applies a label value for the intent as an output.) (Paragraph 9).
(The representative inquiry 430 refers to an inquiry message having a largest probability (confidence) to be analyzed and classified as a specific intent. For example, the representative inquiry 430 that may be classified into an intent of ‘Inquiry about a Good Lock function’ may be “What is a Good Lock function?”. The representative inquiry 430 mapped with the label value 420 of the intent may be pre-stored in the response message DB) (Paragraph 124)
Yang et al. uses a neural network in order to classify messages received from a user inquiry.
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the multistage communication pipeline as taught by Champaneria to use a neural network for classification of user interaction data as taught by Yang et al. This would have been an obvious substitution as this is an alternative method of applying labels to received messages (Yang et al., Paragraph 9).
Claims 10 and 22 are rejected under 35 U.S.C. 103 as being unpatentable over US Patent Publication US 10318927 B2 (Champaneria) in view of US Patent Publication US 6101480 A (Conmy et al.).
Regarding Claims 10 and 22, Champaneria teaches the system of claims 1 and 13.
wherein the next step includes scheduling a meeting with the lead,
(In a next part of a review step 7, it may be determined whether one or more interviews should be scheduled with the candidate 711. If, based on the review of the hiring manager, recruiter, or other similarly-situated party, the candidate warrants an interview, one or more interviews may be arranged with the candidate in order to further evaluate the candidate.) (Col. 44, Lines 64-67 to Col. 45, Lines 1-3).
In Champaneria et al. a next step after the user interaction data has been labeled can be to schedules a meeting with the candidate.
and causing a display of a reminder,
(when a candidate has accepted an offer to perform an interview at a particular time, a schedule may be created 73. In an exemplary embodiment, a schedule entry may be created in a machine-based calendar or multiple machine-based calendars (such as a MICROSOFT OUTLOOK calendar) for either or both of the candidate and the interviewer. For example, according to an exemplary embodiment, a candidate may be emailed a calendar entry that may be integrated, from their email inbox, into a calendar.) (Col. 46, Lines 9-18).
Champaneria displays a reminder of the meeting by updating a virtual calendar and/or sending an email to the user and candidate.
While Champaneria does teach scheduling a meeting and sending reminders, it does not explicitly teach: wherein scheduling further comprises: retrieving a lead calendar and a user calendar; identifying a potential meeting time slot by applying an algorithm to the retrieved lead calendar, retrieved user calendar, and the lead data; wherein the reminder is generated based on the identified potential meeting time slot.
However, Conmy et al. teaches wherein scheduling further comprises: retrieving a lead calendar and a user calendar;
(According to an embodiment of the present invention, the first step in scheduling an event is the generation of a request for an event, as in step 100. In the request, the coordinator may select desired invitees for the event, including any rooms or other resources that are needed for the event as well as a preferred date, time, duration, and location for the event. After all invitees have been selected, the system accesses availability information from database 200 and if necessary, through calendar connect unit 310 for invitees that may not be stored in database 200. As discussed above, invitee profiles and the calendar files that may be stored in and/or with each invitee's mail file, for example, (in step 102) for each invitee and stored in database 200. That availability information may then be used, in step 104, to determine the busy time events for the user for a certain time period requested.) (Col. 5, Lines 36-50).
Conmy et al. considers the users preferred data and time and accesses invitees’ calendars in order to schedule a meeting. Fig. 3 depicts a flowchart of this method.
identifying a potential meeting time slot by applying an algorithm to the retrieved lead calendar, retrieved user calendar, and the lead data;
(If there is no such time interval during which all invitees are available, the system proceeds to determine a "best fit" in step 108. The process of step 108 is depicted in FIG. 4. In FIG. 4, the first step is that the coordinator is requested to assign a weight for each invitee in step 112. That step may be performed at the time the coordinator is asked to invite the resources or persons or may be delayed until a determination is made as to whether free time for all invitees may be located. Alternatively, default values may be assigned to types of resources. For example, the chairman and the conference room may be assigned a high weighting whereas other individuals may be assigned lower weighting.) (Col. 6, Lines 34-45).
Conmy et al. applies an algorithmic approach to identifying a meeting time slot. This includes assigning priority values to important attendees.
wherein the reminder is generated based on the identified potential meeting time slot.
(If all invitees can attend, in step 110, the system sends an invitation to the invitees to attend the event, for example, by electronic mail using the address stored for each invitee. The invitees may then either accept the invitation, at which point the system would update their respective calendar files with the new event, or if the invitation is declined, the system notifies the requester.) (Col. 5, Lines 57-62).
Reminders are generated in the form invitations being sent out and calendars being updated.
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the multistage communication pipeline as taught by Champaneria to apply an algorithm to calendar information to automatically schedule meetings as taught by Conmy et al. This would have been an obvious improvement to add another automated step for scheduling that enhances the efficiency and productivity of the user (Conmy et al., Col. 1, Lines 26-32).
Claims 11 and 23 are rejected under 35 U.S.C. 103 as being unpatentable over US Patent Publication US 10318927 B2 (Champaneria) in view of US Patent Publication US 6101480 A (Conmy et al.) and further in view of US Patent Publication US 10282759 B1 (King et al.).
Regarding Claims 11 and 23, Champaneria in view of Conmy et al. teaches the system of claims 10 and 22.
Champaneria in view of Conmy et al. et al. does not explicitly teach: wherein the display is presented as a part of a sales pipeline that indicates an engagement progress.
However, King et al. teaches wherein the display is presented as a part of a sales pipeline that indicates an engagement progress.
(The category III. Sales Process is entered in step 460, in which the management system displays the scheduled appointment on the advisor's dashboard and calendar.) (Col. 11, Lines 9-11).
(The process flows to step 492, in which the system updates the category of the potential buyer to “stalled”. The process continues to step 494, in which the advisor support coordinator reviews the stalls on the dashboard with the advisor in a regularly scheduled meeting.) (Col. 11, Lines 55-59).
In King et al., a dashboard is implemented for the entire sales pipeline. Scheduled meeting reminders are displayed on this dashboard as well as a current progress for that particular buyer such as “stalled”. Figs. 7A-B and 8A-D show this dashboard and various statuses that can be displayed for a buyer.
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the multistage communication pipeline featuring automatic meeting scheduling as taught by Champaneria in view of Conmy et al. to display the meeting reminder and sales pipeline engagement progress to the user as taught by King et al. This would have been an obvious improvement as this would apply the method of Champaneria to a user interface element similar to a client relationship management (CRM) system. There is need automated steps in CRM like systems as it reduces the amount of training required by the user (King et al., Col. 1, Lines 32-48).
Conclusion
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 nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICHOLAS DANIEL LOWEN whose telephone number is (571)272-5828. The examiner can normally be reached Mon-Fri 8:00am - 4:00pm.
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, Paras D Shah can be reached at (571) 270-1650. 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.
/NICHOLAS D LOWEN/Examiner, Art Unit 2653
/Paras D Shah/Supervisory Patent Examiner, Art Unit 2653
06/26/2026
Prosecution Insights