Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
The instant application having Application No. 18/673,086 is presented for examination by the examiner. Claims 1, 4, 8, 11, 15 and 18 are amended. Claims 3, 10 and 17 are cancelled. Claims 1-2, 4-9, 11-16 and 18-20 have been examined.
Response to Arguments
Applicant’s arguments filed on 12/16/2025 have been fully considered but they are not persuasive.
Regarding applicant's argument on page 12 that Wang does not teach building a network environment based on the network configuration, the examiner disagrees. Wang teaches generating network configurations (e.g., P4 table entries) and applying those configurations to network devices within an emulated network environment (Wang, p. 6). Wang explains that the generated configuration is automatically installed on enabled switches, which configures forwarding paths and network behavior across the system. The application of these configuration rules to network devices results in instantiating and configuring the network according to the generated configuration. Accordingly, Wang teaches building a network environment based on the network configuration, and applicant’s argument is therefore not persuasive.
Applicant further argues on page 12 that Wang does not teach populating the network environment with the first set of content and the second set of content. The examiner disagrees. Wang teaches that the generated network configuration (e.g., P4 table entries) is derived from prior outputs such as topology information, routing specifications, and previously generated configurations (Wang, p. 6). These prior outputs correspond to the first set of content and are used to generate and implement forwarding rules across the network, while the generated configurations correspond to the second set of content. Wang further explains that these configurations are deployed onto network devices, where they control forwarding paths and network behavior. The deployment necessarily incorporates both the prior content (e.g., routing logic and topology information) and the generated configuration into the operational network environment. Accordingly, Wang teaches populating the network environment with both the first set of content and the second set of content, and applicant’s argument is therefore not persuasive.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness
rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 8-9 and 15-16 is/are rejected under 35 U.S.C. 103 as being unpatentable over N.P.L Jacob “Hey, Lumi! Using Natural Language for Intent-Based Network Management” in view of N.P.L Wang “Making Network Configuration Human Friendly”, on view of AGHAJANYAN (US 20250117410 A1).
Regarding Claim 1
Jacob discloses:
A method comprising: generating a first prompt to an artificial intelligence (Al) model to generate a first output based on an initial input (Jacobs Page 3: “The main building blocks for LUMI’s Information Extraction module are a chatbot interface as the entry point into our system and the use of Named Entity Recognition (NER) [30] to extract and label entities from the operators’ natural language intents.” This teaches the operator’s natural language intent is submitted through a chatbot interface that serves as a prompt to the AI model (NER), which in turn generates a first output by extracting and labeling entities. This directly corresponds to generating a first prompt based on an initial input and receiving a first output from the AI model.);
receiving the first output from the Al model, the first output comprising a first set of content (Jacob Page 4: “First, for the Information Extraction module (described in Section 3), we rely on machine learning to extract and label entities from the operator utterances and implement them using a chatbot-like conversational interface. The extracted entities form the input of the Intent Assembly module (described in Section 4), where they are used to compose a Nile network intent.”. This teaches the system uses machine learning model to extract and label entities from natural-language input, and these extracted entities constitute the first output, which is then received and used as a first set of content in the subsequent module. This directly corresponds to receiving the first output from the AI model, the first output comprising a first set of content.);
storing the first set of content and the network configuration (Jacob Page 8: “LUMI stores every confirmed Nile intent so that it can install or remove device configurations according to times and dates defined by the operator.” This teaches that LUMI/Nile explicitly stores Nile intents (the first set of content derived from natural language inputs), which are later used to determine when to install or remove device configurations (network configurations). This demonstrates storage of both the intermediate outputs (intents) and the final configurations);
Jacob teaches that NLP inputs are processed through a chatbot to generate labeled entities as a first output, which are stores as intents and later used to generate and apply network configuration. However, Jacob does not teach the following limitation “generating, by a processing device, a second prompt to the Al model to generate a second output comprising a network configuration based on the first set of content and the initial input, wherein the second prompt comprises at least a portion of the first set of content and the initial input, and requires the Al model to generate the network configuration to be consistent with both the first set of content and the initial input; receiving the second output from the Al model, the second output comprising the network configuration, building a network environment based on the network configuration; and populating the network environment with the first set of content and the second set of content.”
However, in an analogous art, Wang discloses network configuration system/method that includes:
generating, by a processing device, a second prompt to the Al model to generate a second output comprising a network configuration based on the first set of content and the initial input (Wang Page 3 and 5: “Before generating low-level network configurations, NETBUDDY translates the formal specification into a high-level configuration… Prompting LLMs to develop code to generate the routing information often results in more efficient & correct output… Finally, the generated output script/code may contain syntax and functional errors. To solve this issue, NETBUDDY also embeds a Verifier component… and provide feedback to the LLM iteratively until all the syntax and functional errors are fixed”. This teaches NETBUDDY describes using a subsequent prompt to an LLM to generate further content (high-level routing information, and ultimately low-level configurations). The system explicitly discloses prompting LLMs with the prior outputs (first set of content) and the initial requirements to yield the second output comprising the network configuration, while also incorporating verifier feedback for iterative refinement.);
receiving the second output from the Al model, the second output comprising the network configuration (Wang Page 6: “Step 3 NETBUDDY interacts with GPT-4 three times to sequentially provide (i) the topology, (ii) switches & hosts configurations … and (iii) the already-deployed P4 program. Then, it uses the output of the previous step to generate the P4 table entries. In this example, NETBUDDY generates P4 table entries to configure switches for all the forwarding paths between hosts … Our testbed uses the output of NETBUDDY (i.e., a JSON structure) and automatically installs the generated table rules on the P4-enabled switches running on Kathará”. This teaches that NETBUDDY receives the second output from the AI model in the form of generated P4 table entries (the network configuration). These entries implement the policies and specifications from the prior step (first set of content) and the operator’s original requirements (initial input), ensuring consistency between the generated configuration and the earlier content. The disclosure also confirms the output is structured and deployable, since it is stored as a JSON structure and installed on network devices.).
building a network environment based on the network configuration (Wang page 6: teaches that the generated network configuration (e.g., P4 table entries) is applied to network devices within an emulated network environment (Kathará), thereby instantiating and configuring the network based on the generated configuration. The automatic installation of configuration rules onto switches demonstrates that the system builds a network environment by deploying the generated configuration to control network behavior and forwarding paths.); and
populating the network environment with the first set of content and the second set of content (Wang page 6: teaches that the network environment is populated with both (i) content derived from prior outputs (e.g., routing information and formal specifications generated in earlier steps, corresponding to the first set of content) and (ii) the generated network configuration (e.g., P4 table entries, corresponding to the second set of content). The disclosure shows that these elements are applied to network devices by installing configuration rules that implement forwarding paths and policies derived from earlier outputs, thereby populating the network environment with both the prior content and the generated configuration data.).
Given the teaching of Wang, a person having ordinary skill in the art before the effective filing date of the claimed invention would have recognized the desirability of modifying the teachings of Jacob by issuing successive prompts to an AI model to produce network configurations derived from both intermediate outputs and initial operator requirements. Wang describes a multi-stage pipeline where an LLM generates a formal or high-level specification and is subsequently prompted again with that specification and the original requirements to produce deployable network configurations such as P4 table entries or BGP rules. Wang further teaches using prior outputs as inputs to subsequent prompts and employing a verifier to ensure the resulting configurations are consistent with both the intermediate outputs and the initial input. Wang additionally teaches deploying the generated configurations by installing them on network devices within an emulated network environment, thereby building the network environment and populating it with both the generated configuration data and content derived from prior outputs. It would have been obvious to apply such a multi-prompt approach to achieve consistent and deployable network configurations, as Wang demonstrates this process explicitly.
Jacob and Wang teaches that NLP inputs are processed through a chatbot to generate labeled entities as a first output, which are stores as intents and later used to generate and apply network configuration. However, they do not disclose the following limitation “wherein the second prompt comprises at least a portion of the first set of content and the initial input, and requires the Al model to generate the network configuration to be consistent with both the first set of content and the initial input.”
However, in an analogous art, AGHAJANYAN discloses prompt chaining system/method that includes:
wherein the second prompt comprises at least a portion of the first set of content and the initial input, and requires the Al model to generate the network configuration to be consistent with both the first set of content and the initial input (¶¶0072–0074, 0083, 0153–0155: teaches generating subsequent prompts in a prompt engineering pipeline, where each prompt is constructed using outputs from prior workflow states together with the original query. The reference further discloses chaining prompts in a guided conversational manner to iteratively refine context and questions, and employing validation loops to cross-check and improve results. This ensures that the generated output is aligned with the original query and prior generated content, thereby requiring the AI model to generate output consistent with both the first set of content and the initial input.);
Given the teachings of Aghajanyan, a person having ordinary skill in the art before the effective filing date of the claimed invention would have recognized the desirability of modifying the combined teachings of Jacob and Wang to construct a second prompt that includes both prior generated content and the initial input, and to require the AI model to generate output consistent with both. Aghajanyan teaches a prompt engineering pipeline in which prompts are chained together such that outputs from earlier stages are reused as inputs for subsequent prompts, and further teaches iterative refinement and validation of outputs to ensure alignment with the original query and prior generated content. It would have been obvious to incorporate Aghajanyan’s prompt chaining and validation techniques into the system of Jacob and Wang in order to improve the accuracy, reliability, and consistency of generated network configurations, as Aghajanyan explicitly recognizes the deficiencies of one-off prompting and provides a solution through structured, context aware prompt construction and validation.
Regarding Claim 2
Jacob, Wang and Aghajanyan combined teaches that NLP inputs are processed through a chatbot to generate labeled entities as a first output, which are stores as intents and later used to generate and apply network configuration.
Wang further discloses network configuration system/method that includes:
The method of claim 1, further comprising: generating a third prompt to the Al model to generate a third output based on the first set of content, the network configuration, and the initial input; and receiving the third output from the Al model, the third output comprising a second set of content that is consistent with the first set of content and the initial input (Wang page 5: “Finally, the generated output script/code may contain syntax and functional errors. To solve this issue, NETBUDDY also embeds a Verifier component (see Fig. 1) that can detect such issues (via existing tools, real or symbolic execution engines, and pre-defined test cases) and provide feedback to the LLM iteratively until all the syntax and functional errors are fixed.” teaches the verifier uses the previously generated configuration together with the initial operator requirements and specification to form additional prompts back to the LLM. These feedback prompts constitute a third prompt since they occur after the initial prompt that generated the specification and the second prompt that generated the configuration. In response, the AI produces a third output in the form of corrected or refined content, such as updated code or verification artifacts.);
Given the teaching of Wang, a person having ordinary skill in the art before the effective filing date of the claimed invention would have recognized the desirability of modifying the teachings of Jacob and Aghajanyan by issuing successive prompts to an AI model to refine and validate network configurations based on both intermediate outputs and the original operator requirements. Wang describes that after an initial specification and configuration are generated, a verifier component detects syntax and functional errors and provides feedback to the LLM iteratively until all the syntax and functional errors are fixed. This iterative feedback incorporates the initial input, the first set of content, and the generated configuration into a further prompt, causing the AI to produce a corrected output that is consistent with the earlier specification and operator requirements. It would have been obvious to employ such iterative prompting to generate refined content in order to ensure correctness and reliability of the final network configuration (Wang page 5).
Regarding Claim 8
Claim 8 is directed to a system corresponding to the computer-implemented method in claim 1. Claim 8 is similar in scope to claim 1 and is therefore rejected under similar rationale.
Regarding Claim 9
Claim 9 is directed to a system corresponding to the computer-implemented method in claim 2. Claim 9 is similar in scope to claim 2 and is therefore rejected under similar rationale.
Regarding Claim 15
Claim 15 is directed to a computer readable medium instructions corresponding to the computer-implemented method in claim 1. Claim 15 is similar in scope to claim 1 and is therefore rejected under similar rationale.
Regarding Claim 16
Claim 16 is directed to a computer readable medium instructions corresponding to the computer-implemented method in claim 2. Claim 16 is similar in scope to claim 2 and is therefore rejected under similar rationale.
Claims 4, 11 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over N.P.L Jacob “Hey, Lumi! Using Natural Language for Intent-Based Network Management”, in view of N.P.L Wang “Making Network Configuration Human Friendly”, in view of AGHAJANYAN (US 20250117410 A1) as applied to claim 1 above, and in further view of Andriani (US 2023/0156032 A1).
Regarding Claim 4
Jacob, Wang and Aghajanyan combined teaches that NLP inputs are processed through a chatbot to generate labeled entities as a first output, which are stores as intents and later used to generate and apply network configuration. However, they do not disclose the following limitation “monitoring the network environment for malicious activity; and collecting information associated with the malicious activity.”
However, in an analogous art, Andriani discloses network environment system/method that includes:
The method of claim 1, further comprising: monitoring the network environment for malicious activity; and collecting information associated with the malicious activity (Andriani ¶152: “Coordination agents 222 deployed on production servers … may also produce statistical baselines … The coordination agents 222 report to the coordination device 220 and the coordination device 220 reports back to the monitor controller 200 in as close to real-time as possible. This allows the monitor controller 200 … to make decisions about the production environment’s … stability status.”; “The entire system working together gives an administrator an immediate and real-time overview of how a perimeter security attack passed through each ring of security” These passages show that the prior art monitors the network environment for malicious activity by continuously collecting baselines and health data during simulated or real attacks. They also show that the system collects information associated with the malicious activity, including traffic statistics and attack progression across different security layers, which is reported back for analysis).
Given the teaching of Andriani, a person having ordinary skill in the art before the effective filing date of the claimed invention would have recognized the desirability of modifying the teachings of Jacob, Wang and Aghajanyan by monitoring the network environment for malicious activity and collecting associated information. Andriani discloses continuously monitoring an emulated production environment, detecting malicious activity during simulated attacks, and recording traffic, statistics, and vulnerabilities in real time. It would have been obvious to apply such monitoring and data collection to identify threats and support mitigation (Andriani ¶152–159 and 267).
Regarding Claim 11
Claim 11 is directed to a system corresponding to the computer-implemented method in claim 4. Claim 11 is similar in scope to claim 4 and is therefore rejected under similar rationale.
Regarding Claim 18
Claim 18 is directed to a computer readable medium instructions corresponding to the computer-implemented method in claim 4. Claim 18 is similar in scope to claim 4 and is therefore rejected under similar rationale.
Claims 5, 12 and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over N.P.L Jacob “Hey, Lumi! Using Natural Language for Intent-Based Network Management”, in view of N.P.L Wang “Making Network Configuration Human Friendly”, in view of AGHAJANYAN (US 20250117410 A1) as applied to claim 2 above, and in further view of Shah (US 20240422187 A1).
Regarding Claim 5
Jacob, Wang and Aghajanyan teaches that NLP inputs are processed through a chatbot to generate labeled entities as a first output, which are stores as intents and later used to generate and apply network configuration. However, they do not disclose the following limitation “converting the first output of the AI model from a raw text format to a first file-type format corresponding to a first type of content, wherein the first file-type format comprises at least one of a portable document format (PDF), an mbox format, a slide deck format, or a text editor format; and converting the second output of the AI model from a raw text format to a second file-type format corresponding to a second type of content, wherein the second file-type format comprises at least one of a portable document format (PDF), an mbox format, a slide deck format, or a text editor format”
However, in an analogous art, Shah discloses an AI model system/method that includes:
The method of claim 2, further comprising: converting the first output of the AI model from a raw text format to a first file-type format corresponding to a first type of content, wherein the first file-type format comprises at least one of a portable document format (PDF), an mbox format, a slide deck format, or a text editor format; and converting the second output of the Al model from a raw text format to a second file-type format corresponding to a second type of content, wherein the second file-type format comprises at least one of a portable document format (PDF), an mbox format, a slide deck format, or a text editor format (Shah ¶127 “The final output can be provided as a report, stored in the HDFS 528, displayed on the dashboard 522. For example, the report can be text from the generative AI program 520 which is converted into an HTML format and then into a PDF. Of course, other embodiments are contemplated. The alerts 534 can be push notifications, Slack alerts, etc.” teaches taking raw text produced by a generative AI model and converting it into structured file-type formats such as HTML and PDF, with additional embodiments supporting other output formats. This directly maps to converting the first and second AI outputs into specified file-type formats.).
Given the teaching of Shah, a person having ordinary skill in the art before the effective filing date of the claimed invention would have recognized the desirability of modifying the teachings of Jacob, Wang and Aghajanyan by converting AI-generated raw text into standardized file-type formats for storage, distribution, and presentation. Shah discloses generating reports from generative AI outputs and converting those outputs into HTML and PDF formats, with other embodiments contemplated for alternative formats. It would have been obvious to extend such conversion to additional file formats such as slide decks, mbox, or text editors, since these represent routine and well-known document formats used in common business and communication workflows (Shah ¶127).
Regarding Claim 12
Claim 12 is directed to a system corresponding to the computer-implemented method in claim 5. Claim 12 is similar in scope to claim 5 and is therefore rejected under similar rationale.
Regarding Claim 19
Claim 19 is directed to a computer readable medium instructions corresponding to the computer-implemented method in claim 5. Claim 19 is similar in scope to claim 5 and is therefore rejected under similar rationale.
Claims 6-7, 13-14 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over N.P.L Jacob “Hey, Lumi! Using Natural Language for Intent-Based Network Management”, in view of N.P.L Wang “Making Network Configuration Human Friendly”, in view of AGHAJANYAN (US 20250117410 A1) as applied to claim 1 and 2 above, and in further view of PERINCHERRY (US 2023/0334387 A1).
Regarding Claim 6
Jacob, Wang and Aghajanyan teaches that NLP inputs are processed through a chatbot to generate labeled entities as a first output, which are stores as intents and later used to generate and apply network configuration. However, they do not disclose the following limitation “wherein the second set of content comprises at least one of workstation related information, communications, or pocket litter files that are dependent on the first set of content, the first set of content comprising at least one of employee related information, business-related information, or customer-related information”.
However, in an analogous art, Ranjan discloses a user profile system/method that includes:
The method of claim 2, wherein the second set of content comprises at least one of workstation related information, communications, or pocket litter files that are dependent on the first set of content, the first set of content comprising at least one of employee related information, business-related information, or customer-related information (PERINCHERRY ¶34: “The commitment analyzer 208… can determine focus areas and/or assign commitment information to focus areas for each user associated with an entity. The commitment information can include time, activity type, number of users, and/or the like needed to complete a certain activity or group of activities associated with a focus area.” and PERINCHERRY ¶35: “The activity stream 210 receives digital artifacts 246. The digital artifacts 246 include information associated with at least one user, such as calendar information, communications, activity reports, direct engagement, and/or the like.” teaches collecting employee or business-related information (commitment information tied to users and activities) and generating a second set of content in the form of digital artifacts such as communications, calendar entries, and activity reports, which correspond to workstation-related information and pocket litter files.).
Given the teaching of PERINCHERRY, a person having ordinary skill in the art before the effective filing date would have recognized the desirability of modifying the teachings of Jacob, Wang and Aghajanyan by generating secondary workstation-related content and communications dependent on primary employee or business information. The system discloses collecting commitment information tied to users and activities (¶0034) and producing dependent digital artifacts such as calendar entries, communications, and activity reports (¶0035), which illustrates that workstation data and pocket-litter-type files can be derived from employee, business, or customer information (Ranjan ¶34 and 35).
Regarding Claim 7
Jacob, Wang and Aghajanyan teaches that NLP inputs are processed through a chatbot to generate labeled entities as a first output, which are stores as intents and later used to generate and apply network configuration. However, they do not disclose the following limitation “wherein the initial input comprises a company profile, the company profile comprising a description of the company”.
However, in an analogous art, PERINCHERRY discloses a user profile system/method that includes:
The method of claim 1, wherein the initial input comprises a company profile, the company profile comprising a description of the company (PERINCHERRY ¶25 and 26: “the engagement plan can be organized into focus areas. The focus areas can be associated with business groups of an entity, roles of the user(s) U1, or groups of user(s) U1” and “the activity stream 210 can monitor the user compute device 130 … or other computing systems associated with the entity (e.g., document servers, time management servers, customer relationship management (CRM) servers, etc.) for relevant activity by the user”.).
Given the teaching of PERINCHERRY, a person having ordinary skill in the art would have recognized the desirability of modifying the teachings of Jacob, Wang and Aghajanyan by using a company profile as the initial input. The system discloses receiving business group and role information for entities and pulling data from CRM servers, which provide descriptive company details such as organizational structure, customer data, and operational context, effectively serving as a company profile with a description of the company (PERINCHERRY ¶25 and 26).
Regarding Claim 13
Claim 13 is directed to a system corresponding to the computer-implemented method in claim 6. Claim 13 is similar in scope to claim 6 and is therefore rejected under similar rationale.
Regarding Claim 14
Claim 14 is directed to a system corresponding to the computer-implemented method in claim 7. Claim 14 is similar in scope to claim 7 and is therefore rejected under similar rationale.
Regarding Claim 20
Claim 20 is directed to a computer readable medium instructions corresponding to the computer-implemented method in claim 6. Claim 20 is similar in scope to claim 6 and is therefore rejected under similar rationale.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any 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 date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SAAD A ABDULLAH whose telephone number is (571) 272-1531. The examiner can normally be reached on Monday - Friday, 8:30am - 5:00pm, EST. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Lynn Feild can be reached on (571) 272-2092. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/SAAD AHMAD ABDULLAH/Examiner, Art Unit 2431
/SARAH SU/Primary Examiner, Art Unit 2431
Prosecution Insights