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
This action is in response to the communication filed on 05/01/2024.
Claims 1-20 are pending and addressed in the Action.
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.
Claim 1-2, 5-6, 8-12, 14-17, 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over
Toffoli, “Basic Calendar Manager AI Agent”, 2023, Medium, https://Medium.com, 6 pages (hereinafter: Toffoli-Calendar), in view of Toffoli, “Building LLM-powered products - Part 1”, 2023, Medium, https://Medium.com, 6 pages 13 pages (hereinafter: Toffoli-LLM)
As per Claim 1: Toffoli-Calendar discloses the limitations in bold as below:
1. A computer-implemented method comprising:
generating an event generation prompt comprising textual instructions for generating, using a [large language model], a calendar event corresponding to a target objective;
(See Figure in p. 1. refer to “Find a 30 min slot for a run tomorrow morning” as textual instruction prompt, and AI in the Figure as LLM, where calendar event and target objective is described the in the Prompt)
in response to the event generation prompt, generating, via [the large language model] (See Figure in p. 1, The System in far left flow, pointed by user’s Prompt):
a task curriculum comprising a set of executable tasks whose completion accomplishes the target objective; and
(See in Figure in p. 1, the System with “Task”: in far left flow: Task B, to perform Action’ C for execute “check Schedule”, to AI in the middle System to perform Action E for execute “add event” and to the far right system with Observation F, and Final answer G.
The B, C, D, E, F, G are tasks, and in the reference, it is to complete a prompt given to the system: The objective is it to find a final answer: "Morning run scheduled for tomorrow at 9am!")
In light of the specification, and within the functionality of the limitations, the order set of A, B, C, D, E, F, G reads “task curriculum”)
computer code executable by a calendar application (In Figure in p.1: referred to “AI” and “Action” to execute tasks “Execute” to ‘computer code’) to generate a series of calendar events corresponding to the set of executable tasks in the task curriculum;
(See Figure in p. 1, and Action C with Execute “Check schedule”, and Calendar API in the System in far left flow. See C in p. 2 and p.3, “"name": "check_schedule",”: ‘executable tasks’, and “"datetime": "2023-08-26"”: ‘calendar events’.
Also see “(A) System” in p. 2, that give an analogy of task executions )
and
executing, via the calendar application, the computer code to generate the series of calendar events for completing the task curriculum.
(See Figure in p. 1, and Action E with Execute “add event”, and Calendar API in the System in middle flow. See E in p. 3 and p.4, “"name": "add_event",”: completing Tasks, “ "datetime": "2023-08-26"”: series of calendar events. And see ‘Final answer’ in the System in the far right flow with the generation: “"Morning run scheduled for tomorrow at 9am!"” .
[It should be noted that the implementations in the Figure for event and task are only Examples; Users could recognize the “Task” to cover a set of tasks, and event to cover a series of events])
Toffoli-Calendar does not explicitly address [large language model] in the claim, but rather AI and with generation of calendar event with user Prompt.
Toffoli-LLM discloses “large language model” (Toffoli-LLM: See in p. 2, in sec. 1 Basic Prompt and two bold dots: “The most fundamental LLM concept:
. you send a piece of text (called prompt) to the model,
and it responds with another piece of text (often called completion).”).
Therefore, it would be obvious to an ordinary of skills in the art before the effective filing of the application to combine the teaching of Toffoli-LLM in describing the prompt and response of an LLM concept, and the teaching of generating a series of calendar events managed by AI Agent in Toffoli-Calendar; the combination would yield predictable results because of the separated descriptions are of the same teachings but for details of key areas, and the integration would be necessary for a completion.
As per Claim 2: Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations as below:
2. The computer-implemented method of claim 1, wherein generating the task curriculum (P. 1, the Figure with B,C,D,E,F,G ) further comprises:
determining an execution timeframe for completing the target objective;
(See in the Figure, Prompt with “30 min slot” ( In light of the specification [0034] “A target objective can include a goal/outcome to be achieved within a set timeframe”)) 30 min slot requested by prompt and Final answer at task G, meet ‘timeframe’)
and
determining a series of event durations for completing the series of calendar events within the execution timeframe.
(See Toffoli-Calendar: Figure in p. 1, the prompt requests a 30 min slot” and the response is a schedule at 9am. In p. 5: Playground. Other user prompt with observation, such as “- 8:00 > 9:00 Call with Jack - 12:30 > 13:00 Lunch with Tara”> (It should be noted that “determining a series of event durations” is based on user prompts, where user’s prompt is a human determination, it could be anything) )
As per Claim 5: Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations as below:
5. The computer-implemented method of claim 1, further comprising:
determining available time intervals within the calendar application based on a priority of the target objective;
(Toffoli-Calendar: See in p. 1, the complete response to the prompt with “Morning run scheduled for tomorrow at 9am” )
and
executing, via the calendar application, the computer code to generate the series of calendar events by scheduling the series of calendar events within the available time intervals.
(See in Figure in p. 1, the complete response to user at “"Morning run scheduled for tomorrow at 9am!", and see in p. 3 “ (E) Assistant
Thought: Based on the schedule, the available time slots for a 30-minute run tom
- 9:00 to 9:30
- 9:30 to 10:00 ”;
And the complete response: see in p. 4
“"datetime": "2023-08-26 09:00:00"”)
As per Claim 6: Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations as below:
6. The computer-implemented method of claim 1, further comprising:
determining priorities of scheduled calendar events within the calendar application;
generating additional computer code for rearranging the scheduled calendar events within the calendar application according to the priorities; and
(All above read on the user determination to schedule set events based on use’s prompt.
In the Figure in Toffoli-Calendar, the chart shows the user’s prompt “ "Find a 30 min slot
for a run tomorrow morning!”, and the AI’s response by find available events in Calendar API to present Final Answer to meet user prompt’s priority with the schedule shown in the path out from Final answer. It should be noted that “prompt” is human’s determination, and the above limitations read on the response from any prompt given by human)
executing, via the calendar application, the additional computer code for rearranging the scheduled calendar events based on the priorities of the scheduled calendar events within the calendar application.
(See the Figure in p. 1 “"Morning run scheduled for tomorrow at 9am!"”
based on the analogy “(E) Assistant in p. 3, and p. 4, as in the above of claim 5 )
As per claim 8: The claim is directed to a system and recites the limitations having functionality corresponding to the method of claim 1 above. The claim is rejected with the same rationales addressed in claim 1.
As per Claim 9: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations in bold as below:
9. The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to:
determine, utilizing the [large language model], an additional calendar event among the series of calendar events; and
(Toffoli-Calendar: See “(D) User” and “Observation: Events for 2023-08-26
- 8:00 > 9:00 Call with Jack
- 12:30 > 13:00 Lunch with Tara”; and/or “(E) Assistant” “the available time slots for a 30-minute run tom..
- 9:00 to 9:30
- 9:30 to 10:00” )
execute, via the calendar application, additional computer code to generate the additional calendar event ordered within the calendar application based on a relation of the additional calendar event to the series of calendar events.
(Figure in p. 1 task set, A, B, C, D, E, F, G with above “(D) User” and “(E) Assistant”)
Toffoli-Calendar does not explicitly address [large language model] in the claim, but rather AI and with generation of calendar event with user Prompt.
Toffoli-LLM discloses “large language model” (Toffoli-LLM: See in p. 2, in sec. 1 Basic Prompt and two bold dots: “The most fundamental LLM concept:
. you send a piece of text (called prompt) to the model,
and it responds with another piece of text (often called completion).”).
Therefore, it would be obvious to an ordinary of skills in the art before the effective filing of the application to combine the teaching of Toffoli-LLM in describing the prompt and response of an LLM concept, and the teaching of generating a series of calendar events managed by AI Agent in Toffoli-Calendar; the combination would yield predictable results because of the separated descriptions are of the same teachings but for details of key areas, and the integration would be necessary for a completion.
As per Claim 10: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations as below:
10. The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to:
determine background event parameters from one or more connectors integrating application data from one or more external computer applications;
(Toffoli-Calendar: Figure in p. 1 and playground in p. 5, read the claimed limitations because the Assistant is an operation of AI connected with other external computer application and the “Calendar API” is also compring background event parameters. Therefore, the series of observation in the Figure is connected to Calendar API where the action of “add event” is to find the available slots that have no background events in Calendar API. In this manner, in the AI with User Observation finds 2 parameters ((D) User in p. 3 or in playground in p. 5) with the tomorrow morning, and 1 available time slot ((F) User in p. 4 or in playground in p. 5))
and
generate the event generation prompt based on the background event parameters.
(Toffoli-Calendar: In playground in p. 5, e.g.
Observation: Events for 2023-08-26
- 8:00 > 9:00 Call with Jack
- 12:30 > 13:00 Lunch with Tara
And
Observation: Event successfully added for 2023-08-26 at 9:00>9:30)
As per Claim 11: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations in bold as below:
11. The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to:
generate, via [the large language model], additional computer code executable by the calendar application to set an interval of time allocated to the calendar event; and
execute, via the calendar application, the additional computer code to set the interval of time allocated to the calendar event.
(Toffoli-Calendar: The Figure in p. 1, and analogies of (A), (B), (C), (D), (E), (F), (G) in pages 2-4)
Toffoli-Calendar does not explicitly address [large language model] in the claim, but rather AI and with generation of calendar event with user Prompt.
Toffoli-LLM discloses “large language model”
(See the same rationales addressing claim 9 above)
As per Claim 12: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations below:
12. The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to modify, in response to a change in the series of calendar events, the calendar event for completing the task curriculum.
(See Figure in p. 1, and Action E with Execute “add event”, and Calendar API in the System in middle flow. See E in p. 3 and p.4, “"name": "add_event",”: completing Tasks, “ "datetime": "2023-08-26"”: series of calendar events. And see ‘Final answer’ in the System in the far right flow with the generation: “"Morning run scheduled for tomorrow at 9am!"” .
[It should be noted that the implementations in the Figure for event and task are only Examples; Users could recognize the modify, in response to a change in the calendar are the response base on the user prompts and the completion based on the set of tasks in the Figure of p. 1 and of the set of tasks (A), (B), (C), (D), (E), (F), (G) in pages2-4]).
As per Claim 14: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations as below:
14. The system of claim 8, further comprising instructions that, when executed by the at least one processor, cause the system to generate the event generation prompt based on analyzing terminology of an event input to determine the target objective corresponds to an execution timeframe.
(See in the Figure, Prompt with “Find a 3 0 min slot for a run tomorrow morning” , and the system analyzing the user prompt by the AI of the system to generate the event generation prompt for “30 min slot” and “tomorrow morning” with “"Morning run scheduled for tomorrow at 9am!"” with the time slot “- 9:00 to 9:30” as seen in (E) Assistant in p. 3)
As per claim 15: The claim is directed to a non-transitory computer readable medium and recites the limitations having functionality corresponding to the method of claim 1 above. The claim is rejected with the same rationales addressed in claim 1.
As per claim 16: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations in bold as below:
16. The non-transitory computer readable medium of claim 15, further comprising
instructions that, when executed by the at least one processor, cause the at least one processor to:
determine an incomplete status of the target objective based on a status of the set of
executable tasks;
(Figure in p. 1, Read on D - User Observation, and see: “(D) User” with “Observation: Events for 2023-08-26
- 8:00 > 9:00 Call with Jack
- 12:30 > 13:00 Lunch with Tara”
As incomplete status, in compare to prompt “run tomorrow morning”)
generate, via the [large language model], additional computer code executable by the
calendar application to generate an additional calendar event for completing the task curriculum (Toffoli-Calendar: referred to “Execute “add event”);
and
execute, via the calendar application, the additional computer code to generate the
additional calendar event to complete the task curriculum.
(See in Figure in p. 1, and “Execute “add event” and “User
Observation” and “Final answer” with "Morning run scheduled for tomorrow at 9am!" as additional calendar event to complete the order set of tasks A, B, C, D, E, F, G);
Toffoli-Calendar does not explicitly address [large language model] in the claim, but rather AI and with generation of calendar event with user Prompt.
Toffoli-LLM discloses “large language model”
(See the same rationales addressing claim 9 above)
As per claim 17: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations in bold as below:
17. The non-transitory computer readable medium of claim 15, further comprising
instructions that, when executed by the at least one processor, cause the at least one processor to:
compare the task curriculum with a scheduled calendar event to determine that the
scheduled calendar event is associated with accomplishing the target objective;
(Toffoli-Calendar: See in Playground in p. 5, the series of Assistant, based on user “Observation: Events for 2023-08-26
- 8:00 > 9:00 Call with Jack
- 12:30 > 13:00 Lunch with Tara” and
Compared to user prompt “Find a 30 min slot for a run tomorrow morning”
and
generate, via [the large language model], additional computer code to associate the scheduled
calendar event with the set of executable tasks whose completion accomplishes the target
objective.
(See in Figure in p. 1, and “Execute “add event” and “User
Observation” and “Final answer” with "Morning run scheduled for tomorrow at 9am!" as additional calendar event to complete the order set of tasks A, B, C, D, E, F, G);
Toffoli-Calendar does not explicitly address [large language model] in the claim, but rather AI and with generation of calendar event with user Prompt.
Toffoli-LLM discloses “large language model”
(See the same rationales addressing claim 9 above)
As per claim 19: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations in bold as below:
19. The non-transitory computer readable medium of claim 15, further compring
instructions that, when executed by the at least one processor, cause the at least one processor to:
determine an amount of time for completing the target objective;
generate, via [the large language model], additional computer code to allocate the amount of
time for completing the target objective to the series of calendar events;
(See in the Figure, in p.1: The User prompts with “Find…30 min slot…” . The AI system in the Figure determines 30 min slot requested by prompt. In p. 5: see Playground, referred to a series of User and Assistant)
and
execute, via the calendar application, the additional computer code to allocate the amount
of time to the series of calendar events.
(See in the Figure in p. 1: the path Execute “add event” to Final answer, and the Answer is based on the user’s prompt).
Toffoli-Calendar does not explicitly address [large language model] in the claim, but rather AI and with generation of calendar event with user Prompt.
Toffoli-LLM discloses “large language model”
See the same rationales addressing claim 9 above)
As per claim 20: Incorporated with limitation of claim 8, Toffoli-Calendar and combining Toffoli-LLM, where
Toffoli-Calendar further discloses the limitations in bold as below:
20. The non-transitory computer readable medium of claim 15, further comprising
instructions that, when executed by the at least one processor, cause the at least one processor to:
determine, utilizing [ the large language model], a priority for the set of executable tasks
based on a relationship between the set of executable tasks and scheduled calendar events;
(Toffoli-Calendar: See in p. 1, the complete response to the prompt with “Morning run scheduled for tomorrow at 9am”: In the context of the prompt: “30 mine slot” and “tomorrow morning” )
and
generate, the computer code executable by the calendar application to generate the series
of calendar events based on the priority of the set of executable tasks.
(See in the Figure in p. 1: the path Execute “add event” to Final answer, and the Answer is based on the user’s prompt).
Toffoli-Calendar does not explicitly address [large language model] in the claim, but rather AI and with generation of calendar event with user Prompt.
Toffoli-LLM discloses “large language model”
(See the same rationales addressing claim 9 above)
Claims 3, 4, 7 are rejected under 35 U.S.C. 103 as being unpatentable over
Toffoli, “Basic Calendar Manager AI Agent”, 2023, Medium, https://Medium.com, 6 pages (hereinafter: Toffoli-Calendar), in view of Toffoli, “Building LLM-powered products - Part 1”, Medium, https://Medium.com, 6 pages 13 pages (hereinafter: Toffoli-LLM), and in further view of Yao et al., “Exploring Large Language Models for Knowledge Graph Completion”, 2023, arXiv, 7 pages.
As per Claim 3: : Incorporated to the recite “computer code” taught by Toffoli-Calendar in combining Toffoli-LLM; Toffoli-Calendar in combining Toffoli-LLM do not explicitly address the limitation below.
Yao discloses below limitations,
3. The computer-implemented method of claim 1, wherein generating the computer code comprises:
generating, from a knowledge graph defining relationships among the set of executable tasks, a prompt modifier for generating the event generation prompt according to data indicated by the knowledge graph; and
providing the prompt modifier to the large language model with the event generation prompt.
(Yao: p. 2, right column, “3.1 Knowledge Graph Completion Tasks In this chapter, we describe the three tasks in knowledge graph completion: triple classification, relation prediction, and entity (link) prediction, and how to transform them into simple prompt questions for LLM to complete the tasks.”: read on ‘generating from a knowledge graph’, and See in “Relation Prediction”: reads on ‘defining relationships among the set of executable tasks’
and
, and see in “Triple Classification. Given a triple (h, r, t), the task is to classify it as correct or incorrect.”: read on ‘prompt modifier’).
Since knowledge graph is part of AI technology; it is well-known in AI-driven application for helping the AI to analyzing prompt context and make decision.
Therefore, it would be obvious to an ordinary of skills in the art before the effective filing of the application to include the teaching of the Knowledge Graph of Yao with the teaching executable tasks for the User Prompt of Toffoli-Calendar and combining Toffoli-LLM. The combination would yield predictable results because knowledge graph is well-known in AI technology which is participated in-model or external model of a large language model (LLM) for assisting in analyzing prompt structure and thus improving accuracy response from the LLM.
As per Claim 4: Incorporated to claim 3 and the recite “computer code” taught by Toffoli-Calendar in combining Toffoli-LLM, in view of Yao,
Toffoli-Calendar further discloses the limitations of:
4. The computer-implemented method of claim 3, further comprising generating the computer code to order the set of executable tasks based on the relationships among the set of executable tasks. (Toffoli-Calendar: The Figure in p. 1, the executions of A, B,C,D, E, F, G)
As per Claim 7: Incorporated to the recite “executable tasks” in claim 1, taught by Toffoli-Calendar in combining Toffoli-LLM,
Toffoli-Calendar discloses
7. The computer-implemented method of claim 1, further comprising automatically executing an executable task of the set of executable tasks (Toffoli-Calendar: The Figure in p. 1, the executions of A, B,C,D, E, F, G)
But Toffoli-Calendar and in combining Toffoli-LLM does not explicitly disclose the limitations below:
Yao discloses the limitations below:
by communicating with one or more external models.
(Yao See in abstract: “We also find that fine-tuning relatively smaller models (e.g., LLaMA-7B, ChatGLM-6B) outperforms recent ChatGPT and GPT-4.4”.
Therefore, it would be obvious to an ordinary of skills in the art before the effective filing of the application to include the commutations with one or more external models of Yao’s teaching with executable tasks of Toffoli-Calendar and in view of Toffoli-LLM.
The combination would yield predictable results for conforming with the network communication standard providing links to all models connected to a network.
Allowable Subject Matter
Claims 13, 18 objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Ted T Vo whose telephone number is (571)272-3706. The examiner can normally be reached 8am-4:30pm ET.
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, Wei Y Mui can be reached at (571) 272-3708. 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.
TTV
May 13, 2026
/Ted T. Vo/
Primary Examiner, Art Unit 2191
Prosecution Insights