A method and system for independent residence of intelligent agents based on multi-layer screen regions

Abstract

This invention discloses a method and system for independent agent residency based on multiple screen regions. Multiple residency regions or layers are set on the current screen page, and at least two agents are assigned to corresponding regions. A residency status record is established for each agent, including an identity identifier, residency region identifier, context space identifier, task identifier, and feedback region identifier. The system maintains the context space, task state, and feedback region of each agent based on these records, preventing them from automatically merging due to shared residency. While maintaining context space isolation, a limited task progress status summary of the other agent is provided to at least one residency agent, with the full context content not being disclosed by default. For agents or other mobile intelligent processing entry points that transition from a detached state to an embedded residency state, the continuous binding relationship is not terminated by non-termination operations such as ordinary clicks, page scrolling, or taps unless a preset release operation is detected.

Description

Technical Field

[0001] This invention relates to the fields of human-computer interaction, artificial intelligence interaction, graphical user interface, screen content processing, agent dwelling, visual field interaction, and multi-layer interface management, and particularly to a method and system for independent agent dwelling based on multi-layer screen regions.

[0002] More specifically, the present invention relates to setting multiple agent dwelling areas or dwelling layers in the current screen page or the current visible field, so that multiple agents can independently dwell in different screen layers, different berth areas or different interaction areas, and establish dwelling status records for each dwelling agent, so that agent identity, dwelling area, context space, task status and feedback area form a continuous association relationship; the system maintains the state and data boundaries of each dwelling agent when the continuous association relationship is not broken, thereby forming a multi-layer agent independent dwelling architecture in the current screen. Background Technology

[0003] With the development of artificial intelligence technology, users are now able to call artificial intelligence models, agents, or functional modules through chat boxes, voice assistants, floating windows, translation windows, search boxes, or other intelligent interaction entry points to obtain translation, explanation, summarization, question answering, content generation, image recognition, task suggestions, or other intelligent processing results.

[0004] In existing technologies, AI interaction entry points typically exist in the form of a single chat box, a single floating button, a single pop-up window, or a single assistant interface. If users need to use multiple intelligent agents or multiple intelligent processing functions simultaneously on the same screen, they usually need to repeatedly switch windows, applications, sessions, or tool entry points.

[0005] The above approach has at least the following shortcomings.

[0006] First, existing intelligent agents typically exist as a single entry point, lacking a structure that allows them to independently reside on the current screen according to different levels, regions, or task types.

[0007] Second, when multiple intelligent agents or multiple intelligent processing functions exist simultaneously, they are easily mixed up in the same chat window, the same overlay, or the same interactive context space, resulting in context pollution, chaotic task status, overlapping feedback results, or difficulty for users to distinguish the responsibilities of different intelligent agents.

[0008] Third, existing floating windows or smart assistants mostly emphasize "calling" or "displaying," lacking a mechanism that allows the intelligent agent to stay, remain, and maintain its task status for a long time, forming a stable correspondence with the current screen area.

[0009] Fourth, existing interfaces typically cannot configure the top area, bottom area, middle free area, edge area or other screen layers as independent residence spaces for different intelligent agents, making it difficult for different intelligent agents to form a hierarchical collaborative relationship on the same current page.

[0010] Fifth, in existing AI interaction methods, multiple agents are either completely isolated from each other, unable to perceive each other's task states, or they share the same context space, which easily leads to context pollution. There is a lack of an in-screen dwelling mechanism that both maintains the independent dwelling of each agent and the isolation of the context space, while allowing them to perceive the limited progress of tasks.

[0011] Therefore, a new method for independent agent dwelling based on multiple screen regions is needed, enabling one or more agents to maintain independent existence in different dwelling layers, different berth areas, or different interaction areas of the current screen, and to perform tasks, maintain states, display feedback, and share limited progress information as needed without polluting each other's context space.

[0012] Furthermore, existing screen-based dwell or hovering interactions typically fail to clearly distinguish between detached, docked, and embedded dwell states. Even after an agent or semantic entry point is dragged into a fixed area, it may still exit due to ordinary clicks, page scrolling, or slight touches, making it difficult to establish a stable in-screen dwell relationship.

[0013] In addition, existing standby areas or berth areas are mostly regarded as single entrances, lacking a mechanism for structured maintenance of the configuration status within areas where stationing relationships have been established. This makes it difficult to maintain a recognizable correspondence between different functional sub-areas within the same stationing area and task types, feedback depths, permission ranges, or result display methods.

[0014] Furthermore, existing multi-window, multi-tab, or floating component management primarily addresses the display and switching of visual windows, application sessions, or ordinary interface objects. These objects are typically associated only with their display position, foreground / background order, or visibility state. For resident agents, movable observation units, semantic entry points, or semantic object carriers awaiting delivery, the system also needs to continuously maintain the binding relationships between object identity, resident area, context space, task state, feedback area, permission scope, and limited progress state, and maintain these binding relationships when the object is hidden, collapsed, displayed with low interference, or redisplayed. Simple window keep-alive or foreground / background switching mechanisms are insufficient to solve the aforementioned technical problems of continuous resident status of multiple agents within the screen, context space isolation, and limited collaboration.

[0015] Therefore, it is necessary to further provide embedded dwelling states, optional functional sub-region configurations, and low-interference visual differentiation mechanisms in the multi-layer agent dwelling architecture, so that different dwelling regions can not only accommodate agents, but also stably maintain the dwelling objects and their optional region configurations. Summary of the Invention

[0016] (a) Purpose of the invention

[0017] The purpose of this invention is to provide a method and system for independent residence of intelligent agents based on multiple layers of screen regions.

[0018] This invention sets up multiple agent dwelling areas or dwelling layers in the current screen page or the current visible field, assigns multiple agents to corresponding dwelling areas, and establishes a dwelling state record for each dwelling agent, so that the agent's identity, dwelling area, context space, task state, and feedback area form a continuous association. The system maintains the context space, task state, and feedback area of ​​different dwelling agents respectively, thereby avoiding context mixing, task state pollution, or overlapping feedback results when multiple agents are on the same interface.

[0019] This invention aims to transform the current screen from merely a display interface for a single intelligent assistant into a layered interactive space where multiple intelligent agents can reside independently. Different intelligent agents can reside in the upper area, lower area, middle free area, side area, edge parking space, or other screen areas, and maintain an independent operating state in their respective areas.

[0020] Furthermore, the present invention can configure at least one dwelling area as an embeddable berth, a strip-shaped waiting area, or an edge waiting area. The system acquires the docking determination result for a free-floating intelligent agent or other movable intelligent processing entry point; when the docking determination result indicates that a preset docking condition is met, the corresponding object is converted to an embedded dwelling state. Objects in the embedded dwelling state can maintain low-interference display and will not exit due to ordinary clicks, page scrolling, or non-release operations when no release condition is detected.

[0021] The present invention further maintains the structure of different resident intelligent agents and their tasks through intelligent agent identity identifiers, resident area identifiers, context space identifiers, task identifiers, feedback area identifiers, permission identifiers, state version information, or limited progress state identifiers; when the resident release condition is not met, the resident state record and its binding relationship can continue to be maintained during the process of hiding, folding, low interference display, or redisplay, so that the current screen can simultaneously carry multiple intelligent agents without context crosstalk, and different intelligent agents can perceive each other's processing progress through limited state summaries when necessary.

[0022] (II) Core Technology Concept

[0023] The core technical concept of this invention is to establish multiple agent dwelling layers or dwelling areas within the current screen, and to establish a continuously maintained dwelling state record for each dwelling agent, so that the agent identity, dwelling area, context space, task state and feedback area form a stable association; when the association is not broken, different agents maintain independent context space, task state and feedback space respectively.

[0024] In other words, this invention focuses on how agents form persistent relationships across different levels of the current screen, how these relationships are bound to context space, task state, and feedback area, how different resident agents maintain data and task boundaries, and how multiple resident agents perceive only limited task progress states without sharing complete context content. This invention does not rely on activity response switching, semantic message transmission protocols, or continuous orchestration of multi-node tasks as essential technical features.

[0025] like Figure 1 As shown, the basic logic of this invention is as follows: Current screen page → Set multiple agent dwelling areas or dwelling layers → Assign multiple agents to corresponding dwelling areas → Establish dwelling state records and continuous binding relationships for each dwelling agent → Maintain their respective context space, task state and feedback area → Provide limited task progress status awareness when necessary → Maintain the independent dwelling relationship of multiple agents in the current screen when the release condition is not met.

[0026] Under this logic, independent residency does not merely refer to an interface element remaining in a certain screen position, but rather to the residing agent forming a continuously maintained association with its identity, residency area, context space, task state, feedback area, permission scope, or state version information. An agent in a residency state can be visually hidden, folded, or displayed in a low-interference manner, but as long as the residency relationship is not terminated, its residency state record, along with the associated context space and task state, continues to be maintained by the system, and the corresponding residency display and feedback association can be restored upon re-enabling or re-displaying.

[0027] Therefore, the intelligent agents on the current screen are no longer just temporary pop-up tools, but can become independent processing entities residing in different screen layers.

[0028] In one extended implementation, the same dwell area, berth area, or standby bar can be further configured with multiple functional sub-areas. The dwell status record can be further associated with functional sub-area identifiers and corresponding configuration identifiers, ensuring that objects with established dwell relationships maintain at least one configuration in different sub-areas: task type, feedback strategy or feedback depth, processing subject, permission scope, or result display method. Different functional sub-areas can be distinguished by differences in color, brightness, transparency, border, halo, texture, or dynamic prompts. This extended implementation is an optional configuration within the dwell relationship and does not constitute a necessary step in the independent dwell mechanism of this invention.

[0029] This invention focuses on establishing a hierarchical architecture for multiple agents on the current screen, characterized by layered residency, persistent state binding, context space isolation, task state maintenance, feedback area isolation, and limited progress state awareness. This invention does not limit the generation method of docking determination results; spatial relationship determination, dwell time determination, release operation recognition, or other entry-level determination methods do not constitute essential technical features of this invention. The technical focus of this invention begins with the establishment, maintenance, restoration, and release of residency state records, persistent binding relationships, and multi-dimensional states after the docking determination result is established. Real-time switching of activity response rights, single-transmission protocols for semantic objects, and continuous automatic task flow between multiple nodes are also not necessary steps in the independent residency mechanism of this invention.

[0030] (III) Technical Solution

[0031] To achieve the above objectives, this invention first provides a method for independent residence of intelligent agents based on multiple screen regions. This section describes the method steps, including the following:

[0032] S1. Display the current screen page.

[0033] The system displays the current screen page.

[0034] The current screen page may include at least one of the following: text, images, videos, web pages, documents, chat content, code, maps, product information, icons, buttons, subtitles, multimedia objects, or other visual content.

[0035] The current screen page can be a mobile terminal page, tablet computer page, personal computer page, browser page, application page, e-reading page, video playback page, office page, learning page, or other pages with display and interactive capabilities.

[0036] S2. Set up multiple agent residence areas or residence layers.

[0037] like Figure 2 As shown, the system sets up multiple agent dwelling areas or dwelling layers on the current screen page.

[0038] The intelligent agent's residence area or residence layer may include at least one of the following: upper residence area, lower residence area, middle free area, side residence area, edge berth area, top berth area, bottom berth area, floating residence area, or other screen areas that can accommodate intelligent agents.

[0039] In one implementation, the upper-layer residency region is used to accommodate agents that perform executive interpretation, deep analysis, global observation, or complex task processing.

[0040] In one implementation, the lower-level dwell area is used to accommodate an agent that performs short explanations, provides immediate feedback, assists learning, provides lightweight question answering, or provides low-interference prompts.

[0041] In one implementation, the intermediate free zone is used to accommodate intelligent agents that are movable, summonable, temporarily exist, or can be dragged to other dwelling areas.

[0042] The above-mentioned upper, lower and middle free regions are merely examples, and the present invention does not limit the specific number, location or visual form of the intelligent agent's residence area.

[0043] In one embodiment, the agent's dwelling area or dwelling layer may further include a strip-shaped standby area, a light strip-shaped standby area, a top standby strip, a bottom standby strip, an edge standby strip, or other low-interference standby areas. The standby area may be displayed with low brightness, low transparency, or a simplified outline when not triggered, and enter a response state when a semantic object, user operation, or task trigger is received.

[0044] In one extended implementation, at least one dwelling area, berth area, or standby area may be configured with multiple functional sub-areas. The dwelling status record can be further associated with functional sub-area identifiers and corresponding configuration identifiers to distinguish at least one of different task types, feedback strategies, feedback depths, processing subjects, permission scopes, or result display methods.

[0045] In one implementation, the plurality of functional sub-regions are distinguished by at least one of color differences, brightness differences, transparency differences, border differences, halo differences, texture differences, icon differences, or dynamic prompt differences, enabling users to identify the configuration of different functional sub-regions; the visual distinction is an optional configuration within the dwell area.

[0046] S3, Generate or invoke multiple intelligent agents

[0047] The system generates or invokes multiple intelligent agents.

[0048] The intelligent agent can be at least one of the following: an artificial intelligence model, an intelligent assistant, a semantic processing subject, a translation subject, an explanation subject, a summarization subject, an image recognition subject, a learning assistance subject, a task execution subject, a retrieval subject, or other subjects with intelligent processing capabilities.

[0049] The intelligent agent can be implemented by a local model, a cloud model, a hybrid model, a rule engine, a plugin module, an in-application functional module, or a combination thereof.

[0050] S4. Assign agents to the corresponding residence areas or residence layers.

[0051] like Figure 3 As shown, the system assigns at least two agents to their respective agent residence regions or residence layers.

[0052] In one implementation, the system can assign an agent to an upper dwelling area, a lower dwelling area, a middle free area, or another screen dwelling area based on the agent type, task type, user instruction, interface state, current page content, or user drag-and-drop operation.

[0053] For example, users can dock agents performing deep analysis tasks to the upper-level dwelling area, dock agents performing brief explanation tasks to the lower-level dwelling area, and keep temporarily summoned agents in the middle free-roaming area.

[0054] In another implementation, the system can obtain the docking determination result for the free agent; when the docking determination result shows that the free agent meets the preset docking conditions, it can change the free state to the stationary state.

[0055] In one embodiment, the free-floating intelligent agent may further include a movable observation unit, a mirror-like unit, a semantic entry point, a semantic object carrier unit to be delivered, an intelligent agent card, or other movable intelligent processing entry points. The system acquires docking determination results for the object and the target dwelling area; when the docking determination result is valid, it converts it into an embedded dwelling state and establishes or updates the corresponding dwelling state record. The docking determination result may be generated by spatial relationship determination, dwell time determination, release operation recognition, or other entry layer determination methods.

[0056] In the embedded dwell state, the intelligent agent or semantic entry point forms a stable binding relationship with the corresponding berth area. Unless dragging out, long press to release, close command, detach operation, or other release conditions are detected, the system will not cause it to exit the dwell state due to the user's ordinary click, scrolling, tap, or non-release operation on the current page.

[0057] In another implementation, the same residence area can accommodate one agent or multiple agents.

[0058] S5. Establish residency status records and persistent binding relationships.

[0059] like Figure 4 As shown, the system establishes a residence status record for agents assigned to different residence areas or residence layers, and forms a continuous binding relationship between the agent and the corresponding residence area, context space, task state and feedback area based on the residence status record.

[0060] The dwelling state record includes at least the agent identity identifier, dwelling area identifier, context space identifier, task identifier, and feedback area identifier. Through these identifiers, the system can determine the dwelling location, context boundary, task status, and feedback attribution of a certain agent when the dwelling relationship is not terminated, rather than simply saving its window coordinates or visibility status.

[0061] The agent identity identifier is used to distinguish different resident agents; the resident area identifier is used to determine the screen resident layer or berth area where it is located; the context space identifier is used to point to the corresponding independent context space; the task identifier is used to bind the current processing task; and the feedback area identifier is used to determine the display or feedback position corresponding to the processing result. The resident status record may also include one or more of the following: permission identifier, resident relationship status field, docking mode status field, display status field, status version information, limited progress status identifier, or release condition identifier.

[0062] In one extended implementation, the dwell state record may also be associated with one or more of the following: a functional sub-region identifier, a corresponding configuration identifier, a visual cue status, a feedback strategy identifier, or a feedback depth identifier. The dwell state record remains valid when the agent is in a hidden, folded, or low-interference display state; when the agent resumes display, the system restores at least one of its dwell region association, context space association, task state association, or feedback region association based on the record.

[0063] The independent residency states of different resident agents are distinguished from each other.

[0064] In one implementation, a first agent resides in an upper-level residency region and maintains a first context space, a first task state, and a first feedback region. A second agent resides in a lower-level residency region and maintains a second context space, a second task state, and a second feedback region. A third agent resides in an intermediate free-roaming region and maintains a third context space, a third task state, and a third feedback region.

[0065] The first context space, the second context space, and the third context space are independent of each other and are not automatically merged or contaminated by the processing of any agent.

[0066] S6. Maintain the context space, task state, and feedback region for each resident agent.

[0067] like Figures 5 to 7 As shown, the system maintains the context space, task state, and feedback region for each resident agent.

[0068] The context space may include the target content received by the agent, the current processing object, historical interaction content, user instructions, in-screen semantic messages, current field of view information, or other information related to the agent's task.

[0069] In one implementation, the system isolates the context spaces of different resident intelligent agents using at least one of the following: agent identity identifier, residency area identifier, context space identifier, task identifier, or permission identifier. Unless user authorization, task continuation, explicit delivery, or permission granting conditions are met, the system does not automatically merge the complete historical interaction content, target content, processing results, or context records of one resident intelligent agent into the context space of another resident intelligent agent; when only the condition of limited task progress status being perceptible is met, the system only opens a limited task progress status summary or status identifier, and does not automatically share the complete target content, complete historical interaction records, complete reasoning process, or complete processing results.

[0070] like Figure 7 As shown, the task status may include at least one of the following: standby status, processing status, paused status, waiting for user confirmation status, waiting for external results status, completed status, or completed and awaiting feedback status.

[0071] To avoid confusion between different state dimensions, the system maintains at least two separate states: dwell relationship state and display state. It can further maintain one or more of the following: docking state and task state. The dwell relationship state indicates whether an object maintains a continuous dwell relationship; the docking state indicates whether the object is in a detached, normal, or embedded dwelling state; the display state indicates whether the object is in a shown, hidden, collapsed, or low-interference display state; and the task state indicates the task stage the object is in. Each state dimension is updated separately; when the display state, docking state, or task state changes, the system does not automatically change the dwell relationship state.

[0072] The feedback area may include the agent's own display area, the interior of the dwell area, the adjacent area of ​​the dwell area, the current screen edge area, the overlay area, or other display areas corresponding to the agent.

[0073] In one extended implementation, the feedback area may also be associated with a functional sub-area identifier or a corresponding configuration identifier in the residency status record. The system determines at least one of the following based on the identifier or configuration relationship: feedback area, feedback strategy, feedback depth, or feedback display method.

[0074] The context space, task state, and feedback region of different resident agents are maintained separately to avoid context pollution, task state confusion, or overlapping feedback results between different agents.

[0075] S7. Provide limited task progress state awareness while maintaining context space isolation.

[0076] like Figure 8 As shown, in one implementation, the system allows different resident agents to perceive the limited task progress status of other resident agents while maintaining isolation of their respective context spaces.

[0077] The limited task progress status may include whether the task has started, whether the task has been completed, processing progress, whether the result has been generated, whether user confirmation is pending, whether there is an anomaly, whether assistance from other intelligent agents is required, or other status information that does not contain complete context.

[0078] In one implementation, the limited task progress state is represented by at least one of a status identifier, status label, progress identifier, completion identifier, exception identifier, assistance request identifier, reflow available identifier, or read-only status summary. The limited task progress state is used to provide task stage awareness to other resident agents, but does not include complete target content, complete historical interaction records, complete reasoning processes, or complete processing results; unless authorized by the user or the system detects an explicit task continuation condition, other resident agents can only read the corresponding status summary or status identifier, and cannot read the complete context content that generated the state. This limited task progress state awareness mechanism is used to express lightweight state-aware relationships between resident agents, and does not require the definition of semantic object encapsulation, target addressing, transmission acknowledgment, or multi-node continuous routing rules.

[0079] For example, if the upper-level resident agent completes a deep analysis of the current article, the lower-level resident agent can know the progress status that "the upper-level agent has completed the analysis", but does not need to read the complete analysis context of the upper-level agent.

[0080] For example, if a lower-level agent is processing a short sentence explanation task submitted by a user, the upper-level agent can know that "the lower-level agent is processing a short explanation task," but does not need to read the complete interaction content of the lower-level agent.

[0081] Thus, different resident agents can maintain their independence while avoiding complete disconnection, forming an in-screen multi-agent resident mechanism that is "isolated by context space and aware of progress status".

[0082] S8. Maintain persistent binding relationships and prevent accidental exits while in embedded persistent state.

[0083] like Figure 9 and Figure 11 As shown, when there is an intelligent agent or other mobile intelligent processing entry that changes from a free state to an embedded stationary state based on the docking determination result, the system establishes or updates the corresponding stationary state record, so that its identity, target stationary area, context space, task status and feedback area form a continuous binding relationship.

[0084] In the embedded dwell state, the system distinguishes between preset release operations and normal page operations; when no drag-out, long press release, close command, detach operation, explicit removal, drag-away operation exceeding the preset distance or other preset release operations are detected, the system will not release the corresponding continuous binding relationship due to normal clicks, page scrolling, light touch, short stay, non-target area touch or other non-release operations.

[0085] S9. Maintain the residency status field and the display status field separately.

[0086] The system maintains a dwell relationship status field and a display status field. The dwell relationship status field indicates whether the system is not dwelling, is dwelling, or the dwell relationship has been terminated. The display status field indicates whether the system is showing, hiding, collapsing, or displaying with low interference. The system may further maintain at least one of a docking mode status field and a task status field to distinguish between dwell relationship, display status, docking mode, and task stage.

[0087] When a display status change event is detected, the system updates the display status field and keeps the dwell relationship status field and the persistent binding relationship unchanged; when the display is restored, the system restores at least one of the corresponding dwell area, context space, task status or feedback area association based on the dwell status record.

[0088] S10. Terminate the corresponding residency relationship when a residency termination condition is detected.

[0089] After a residency relationship is established, the system maintains the residency state record and its associated relationships unless a residency release condition is detected. When an agent or other mobile intelligent processing entry point enters a hidden, collapsed, low-interference display, or temporarily invisible state, the system only updates the corresponding display state and maintains the residency relationship state, as well as the associations with the context space, task state, and feedback area. When the display is restored, the system restores the corresponding associations based on the residency state record.

[0090] When the system detects dragging out of the dwell area, long press to release, close command, detach operation, explicit removal, meeting preset release conditions, or other dwell release events, the system releases the corresponding dwell relationship and releases, retains, archives, or transfers the context space, task status, feedback result, or limited progress status associated with the dwell status record according to preset strategies.

[0091] (iv) Beneficial effects

[0092] 1. The present invention sets up multiple intelligent agent dwelling areas or dwelling layers in the current screen, and establishes a continuous association between intelligent agent identity, dwelling area, context space, task status and feedback area through dwelling status recording, so that the dwelling of intelligent agents in the screen is different from ordinary multi-window management that only saves window position or foreground and background status.

[0093] 2. This invention maintains independent context spaces, task states, and feedback regions for different resident intelligent agents, which can reduce context pollution, task state confusion, and overlapping feedback results.

[0094] 3. This invention supports various dwelling modes, such as upper dwelling area, lower dwelling area, middle free area, side area or edge berthing area, to enhance the current screen's ability to support multiple intelligent agents.

[0095] 4. This invention allows free-floating agents to be dragged into a fixed residence area, enabling the agent to switch between a free-floating state and a residence state, thereby improving the user's control over the agent's existence.

[0096] 5. This invention supports accommodating one or more agents in the same dwelling area, enabling the current screen to form a dynamic multi-agent dwelling structure based on task complexity.

[0097] 6. This invention provides limited task progress status awareness to different resident intelligent agents while maintaining the isolation of their context spaces, enabling multiple intelligent agents to perceive each other's processing progress without having to share the complete context content.

[0098] 7. The present invention establishes a stable foundation for the existence relationship and state binding of in-screen intelligent agents, which can be used for subsequent layer-by-layer interaction, access relationship establishment, in-screen semantic communication or task orchestration mechanism calls, without requiring the switching of activity response rights, message transmission protocols or continuous task routing as necessary steps of the present invention.

[0099] 8. The present invention supports the system to convert a free intelligent agent or other movable intelligent processing entry into an embedded dwell state based on the docking determination result, and maintain stable dwelling when the release condition is not met, so as to avoid accidental exit due to ordinary clicks, scrolling or light touch.

[0100] 9. This invention supports configuring optional functional sub-regions within a residency area, and records the associated functional sub-region identifiers and corresponding configurations through residency status records, so that tasks, feedback, or permission configurations within the same residency area remain distinguishable.

[0101] 10. The present invention can distinguish different functional sub-regions through low-interference visual differences such as color, brightness, transparency, border, halo or dynamic prompts, without having to make the spatial hit determination algorithm an essential technical feature of the present invention.

[0102] 11. This invention uses agent identity identifiers, residence area identifiers, context space identifiers, task identifiers, feedback area identifiers, permission identifiers, and state version information to structurally maintain the existence and task states of multiple residence agents, so that the residence relationship can be continuously maintained during the processes of hiding, collapsing, low-interference display, or re-display.

[0103] 12. This invention establishes a lightweight state summary layer between complete context space isolation and complete disconnection through a limited task progress state awareness mechanism, enabling different resident agents to perceive task stages without reading each other's complete context.

[0104] 13. By embedding a dwell state, a continuous binding relationship, and a release condition determination mechanism, this invention enables intelligent agents or other mobile intelligent processing entry points to form a stable binding relationship after the docking determination result is established, thereby reducing the risk of accidental exit due to ordinary clicks, scrolling, or light touches.

[0105] (v) Comparison of technical effects

[0106] In existing technologies, when a user simultaneously opens a translation assistant and a deep analysis assistant on the same current screen page, the two assistants typically share the same chat context or distinguish tasks solely based on window position. If the user first asks the translation assistant about the meaning of a sentence and then asks the deep analysis assistant about the article structure, the deep analysis assistant may mistakenly interpret the aforementioned translation request as part of its context, causing the output to deviate from the current analysis task.

[0107] In this invention, the translation assistant can reside in the lower-level residency area and maintain an independent context space identifier, while the deep analysis assistant can reside in the upper-level residency area and maintain another independent context space identifier. Their context spaces are isolated from each other and do not automatically merge due to both residing on the same screen page. Simultaneously, the translation assistant in the lower-level residency area can sense the progress of the limited task of analysis completed by the deep analysis assistant in the upper-level residency area, thereby deciding whether to prompt the user to interpret the keywords in the analysis results without needing to read the complete analysis content, historical interaction records, or reasoning process of the upper-level assistant. Thus, this invention forms a restricted state summary layer between complete isolation and complete sharing, reducing contextual crosstalk, task state confusion, and feedback overlap in scenarios where multiple agents operate on the same screen. Attached Figure Description

[0108] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0109] Figure 1 This is a schematic diagram of the intelligent agent independent residence system based on a multi-layer screen region according to the present invention;

[0110] Figure 2 This is a schematic diagram of the current multi-layer dwell area layout of the screen according to the present invention;

[0111] Figure 3 A schematic diagram illustrating the allocation of the intelligent agent of the present invention to the corresponding residence area or residence layer;

[0112] Figure 4 This is a schematic diagram illustrating the establishment of the residency status record and continuous binding relationship in this invention;

[0113] Figure 5 This is a schematic diagram illustrating the context space isolation of different intelligent agents in this invention;

[0114] Figure 6 This is a schematic diagram illustrating the isolation of feedback regions for different intelligent agents in this invention;

[0115] Figure 7 This is a schematic diagram illustrating the separate maintenance of task states for different intelligent agents in this invention;

[0116] Figure 8 This is a schematic diagram illustrating the limited progress of the present invention.

[0117] Figure 9 This is a schematic diagram illustrating the conversion of a free intelligent agent into an embedded and residing state according to the present invention;

[0118] Figure 10 This is a schematic diagram showing the configuration of optional functional sub-regions within the area where the present invention resides;

[0119] Figure 11 This is a schematic diagram illustrating the embedded dwell state holding and releasing mechanism of the present invention;

[0120] Figure 12 This is a schematic diagram illustrating the maintenance and transformation of the multidimensional states of the intelligent agent according to the present invention. Detailed Implementation

[0121] The present invention will be described below with reference to embodiments. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. Modifications, equivalent substitutions, and variations made by those skilled in the art without departing from the core idea of ​​the present invention should all fall within the scope of protection of the present invention.

[0122] The following embodiments one through thirteen are mainly described according to the method implementation. Except for the system modules described separately in "VI. System Structure", the expressions "system execution", "system maintenance", "system acquisition" etc. mentioned in this section are used to describe the execution process or execution subject in the method steps, and are not listed as system modules; the corresponding system modules and their functions are described in "VI. System Structure".

[0123] Example 1: Independent residency of agents in the upper and lower regions of the screen.

[0124] In one implementation, the system sets an upper dwell area and a lower dwell area in the current screen page.

[0125] The upper-level dwell area can be located in the upper area of ​​the current screen page, the top edge area, the top floating layer area, the top viewing area, or other display areas associated with the upper space.

[0126] The lower-level dwell area can be located in the lower area of ​​the current screen page, the bottom edge area, the bottom floating layer area, the bottom docking area, or other display areas associated with the space below.

[0127] The system assigns the first agent to the upper-layer residency area and the second agent to the lower-layer residency area. The first and second agents each have independent agent identities, residency locations, task types, context spaces, task states, and feedback areas.

[0128] For example, the first agent can be used to perform long explanations, deep analysis, global observation, structural decomposition, or complex task processing. The second agent can be used to perform short explanations, immediate feedback, learning hints, word meaning explanations, or lightweight question answering.

[0129] In this embodiment, the content processed by the first intelligent agent, historical interactions, task status, and feedback results are not automatically incorporated into the context of the second intelligent agent; similarly, the content processed by the second intelligent agent, historical interactions, task status, and feedback results are not automatically incorporated into the context of the first intelligent agent.

[0130] Thus, the upper and lower dwelling areas form independent agent dwelling spaces, enabling multiple agents to coexist on the same current screen page without context pollution.

[0131] Example 2: Residency of a mobile intelligent agent in the intermediate free region

[0132] In one implementation, the system generates a third agent on the current screen page. The third agent may reside in an intermediate, free-floating region.

[0133] The intermediate free zone refers to an area on the current screen page that is not fixed to the upper or lower dwell area, but can be moved, hovered, stayed, dragged, or temporarily existing on the current screen page according to user operations.

[0134] The third agent in the middle free area can be presented as a floating icon, a movable window, an observation unit, a mirror unit, an agent avatar, a miniature interactive entry point, a semi-transparent overlay, or other movable visual forms.

[0135] Users can control the position of the third-party agent on the current screen page by dragging, pressing, clicking, long-pressing, or other operations.

[0136] The system maintains an independent context, task state, and feedback area for the third agent. Even if the third agent moves within the current screen page, its corresponding context space and task state remain isolated from the upper-layer and lower-layer resident agents.

[0137] In one implementation, the third agent can be used to temporarily observe local content, perform immediate interpretation, collect target objects, perform short-term processing, or act as a temporary semantic entry point.

[0138] In this way, the intermediate free agent can exist as a temporary agent entry point on the current screen, and together with the fixed resident agent, form a multi-layer agent residence structure.

[0139] Example 3: Transition of a free agent to a fixed residence region

[0140] In one implementation, a user can drag an agent in the middle free zone to the upper dwelling zone, lower dwelling zone, side dwelling zone, edge berth zone, or other fixed dwelling zone.

[0141] The system obtains the docking determination results for the free agent and the target dwelling area; when the docking determination results show that the preset docking conditions are met, the system changes the free agent from the free state to the dwelling state.

[0142] This embodiment does not limit the method of generating the docking determination result. The result can be generated by spatial relationship determination, operation confirmation, or other entry identification mechanisms; the above determination methods are only used to provide the result of whether docking is successful and do not constitute the technical focus of this invention.

[0143] After the docking determination result is valid, the system generates or updates the dwelling status record for the free agent, so that the agent's identity is continuously associated with the target dwelling area, context space, task status and feedback area; the dwelling status record can also be associated with dwelling level, display status and permission control information.

[0144] For example, after a user drags an intermediate detached agent into the lower-level dwelling area, the agent changes from a temporary detached state to a lower-level dwelling state and maintains its subsequent task state and feedback state in the lower-level dwelling area.

[0145] Thus, users can determine the existence level of an agent through spatial dragging operations, transforming the agent from a temporary tool into a resident processing entity on the current screen.

[0146] Example 4: Multiple agents can be accommodated in the same residence area

[0147] In one implementation, the same agent residence area can accommodate one agent or multiple agents.

[0148] For example, the upper-level residency area can simultaneously accommodate a first agent and a fourth agent. The first agent is used to perform long text analysis, while the fourth agent is used to perform background retrieval or structured organization.

[0149] For example, the lower-level residence area can simultaneously accommodate a second agent and a fifth agent. The second agent is used to perform short explanations, while the fifth agent is used to perform pronunciation, word meaning prompts, or example sentence generation.

[0150] When multiple agents reside in the same area, the system maintains an independent identity, context space, task status, and feedback status for each agent.

[0151] In one implementation, multiple agents can be displayed in the same dwelling area in a parallel, stacked, folded, labeled, avatar-like, card-like, or other visual forms.

[0152] In another implementation, the system can automatically adjust the display order or display density of multiple agents in the same dwell area based on task type, priority, user selection, processing progress, or display space.

[0153] In this embodiment, although multiple agents in the same dwelling area share the same dwelling area, their context space and task state are still maintained separately and are not automatically merged because they are located in the same area.

[0154] In this way, the current screen can form a dynamically expandable multi-agent dwelling structure according to user needs and task complexity, rather than being limited to a single upper-level agent, a single lower-level agent, or a single free agent.

[0155] Example 5: Context space isolation between different resident agents

[0156] like Figure 5 As shown, in one implementation, the system maintains mutually isolated context spaces for different resident agents.

[0157] The context space may include user instructions, target content, current visible field information, in-screen semantic messages, historical feedback content, processing results, task parameters, permission range, or other information related to the agent's task processing.

[0158] The context space of the first resident agent is stored in the first context space. The context space of the second resident agent is stored in the second context space. The context space of the third resident agent is stored in the third context space. By default, different context spaces do not share complete content.

[0159] In one implementation, the system can distinguish the context space of different residing intelligent agents by intelligent agent identity identifier, residing area identifier, task identifier, context space identifier, or permission identifier.

[0160] For example, the upper-level resident agent is analyzing the entire article. The lower-level resident agent is interpreting the current sentence. The middle detached agent is observing a specific image region. The system maintains context spaces for each of the three agents separately, ensuring that the tasks of analyzing the entire article, interpreting the current sentence, and observing the image region do not interfere with each other.

[0161] In this way, users can keep the working state of multiple agents on the same current screen page at the same time without worrying about semantic crosstalk between different tasks.

[0162] Example 6: Feedback Region Isolation Between Different Resident Agents

[0163] like Figure 6As shown, in one implementation, the system sets or maintains corresponding feedback regions for different resident intelligent agents.

[0164] The feedback area may include the area inside the dwell area, the area adjacent to the dwell area, the screen edge area, the floating display area, the agent card area, the voice feedback channel, or other feedback spaces associated with the corresponding agent.

[0165] For example, the processing results of upper-layer resident agents are displayed in the upper-layer feedback region. The processing results of lower-layer resident agents are displayed in the lower-layer feedback region. The processing results of intermediate free agents are displayed in their vicinity or corresponding floating region.

[0166] Different feedback areas are isolated from each other to avoid the processing results of multiple agents being superimposed in the same location, causing occlusion, confusion, or reading interference.

[0167] In one implementation, when screen space is insufficient, the system can fold, shrink, hide, queue, or display multiple feedback areas in layers based on agent priority, task type, user attention status, or display area size.

[0168] In this way, different resident intelligent agents can output feedback on the same current screen page without disrupting the reading flow, browsing flow, or operation flow of the current page.

[0169] Example 7: Constrained Task Progress Status Awareness

[0170] like Figure 8 As shown, in one implementation, different resident agents can perceive the limited task progress status of other resident agents while maintaining context space isolation.

[0171] The finite task progress state differs from the complete context content sharing. The finite task progress state is used to represent the task progress or task state of a resident agent, without including the agent's complete context content.

[0172] The limited task progress status can include task started, task in progress, task completed, task paused, waiting for user confirmation, processing result generated, processing failed, user authorization required, assistance from other intelligent agents required, and anomaly or conflict exists.

[0173] For example, after the upper-level resident agent completes the global analysis of the current page, the system can provide the lower-level resident agent with the progress status of "upper-level analysis completed", but will not expose the complete analysis process and all context content to the lower-level resident agent.

[0174] For example, when a lower-level resident agent is processing a short explanation task submitted by a user, the system can provide the upper-level resident agent with the status "lower-level short explanation task in progress," enabling the upper-level resident agent to perceive the progress of the lower-level task without reading its complete interaction content.

[0175] In one implementation, the system can establish a limited progress state awareness layer and open the task state, task stage, or result availability to other resident intelligent agents through state identifiers, state tags, progress identifiers, completion identifiers, anomaly identifiers, assistance request identifiers, reflow availability identifiers, read-only state summaries, or other lightweight state data, without opening the complete target content, complete historical interaction records, complete context space, or complete generated results.

[0176] In this way, the system can achieve a balance between "contextual space isolation" and "collaborative awareness", so that different resident intelligent agents neither contaminate each other nor completely lose contact.

[0177] Example 8: Fractal Maintenance of Resident Relationships, Display Status, and Task Status

[0178] like Figure 12 As shown, in one implementation, the system maintains at least a dwelling relationship state and a display state for each agent, and may further maintain one or more of a docking form state and a task state; the system maintains the association between each state dimension and the agent's identity, dwelling area, context space, and feedback area through dwelling state records.

[0179] The system includes several key parameters: Residency status (indicating no residency, residency, or residency relationship terminated); Dock status (indicating detached, normal, or embedded residency); Display status (indicating displayed, hidden, collapsed, or low-interference display); and Task status (indicating standby, processing, paused, awaiting confirmation, awaiting external results, completed, or completed and awaiting feedback). Each parameter is maintained separately. When the Display status, Dock status, or Task status changes, the system maintains the residency status unchanged until a clear residency termination condition is detected.

[0180] For example, an agent can switch from normal display to hidden, collapsed, or low-interference display while maintaining the same dwell relationship; it can also switch from standby task state to processing, paused, or completed and awaiting feedback state while maintaining the same display state.

[0181] When an agent is in a free state, it can move, drag, or temporarily stay on the current screen page. When the docking determination result obtained by the system shows that the preset docking conditions are met, the system updates its dwelling relationship status and docking state, and establishes or updates the corresponding dwelling state record.

[0182] When an agent has established a residency relationship, it becomes bound to the corresponding residency region or residency layer and maintains the association with the corresponding context space, task state, and feedback region. When the task state changes, the system keeps the residency relationship state unchanged until a clear residency release condition is detected.

[0183] When the display state of the agent is hidden or collapsed, the system updates the display state field and hides or collapses the feedback content, while keeping the dwell relationship state field and dwell state record valid. When the display state is restored, the system restores at least one of its original dwell area, dwell level, context space association, task status, or feedback display according to the dwell state record.

[0184] When the system detects a clear condition for termination of residency, it updates the residency relationship status and removes the corresponding binding relationship. Then, it releases, archives, retains, or transfers the residency area, context space, task status, or feedback area association according to a preset strategy.

[0185] Through the aforementioned dwelling state recording and management mechanism, the system can support the continuous existence, low-interference dwelling, temporary hiding, folded display and subsequent restoration of the intelligent agent within the current screen, while not automatically canceling the dwelling relationship due to changes in ordinary visual state.

[0186] Example 9: Berth Embedding and Dwelling State

[0187] like Figure 9 As shown, in one implementation, the current screen page is configured with a top berth area, a bottom berth area, or an edge berth area. The system determines the target dwell area for a free-floating intelligent agent or other movable intelligent processing entry point and obtains the corresponding berthing determination result.

[0188] When the docking determination result indicates that the preset docking conditions are met, the system establishes or updates the corresponding dwelling state record, converts the object from a free state to an embedded dwelling state, and establishes a continuous binding relationship between its identity, target dwelling area, context space, task status, and feedback area.

[0189] like Figure 11 As shown, in the embedded dwell state, the object forms a stable binding relationship with the corresponding berth area and maintains its agent identity, task state, context space state, feedback state, dwell area identifier, or state version information. The system distinguishes release conditions from ordinary page operations. Unless the user performs a drag-out, long-press release, close command, detach operation, explicit removal, drag-away operation exceeding a preset distance, or other release operation, the system will not cause it to exit the dwell state due to the user's ordinary clicks, scrolling, taps, brief pauses, non-target area touches, or non-release operations on the current page.

[0190] In one implementation, the system can determine whether a user operation is a release operation or a normal operation based on the touch start point, touch duration, movement distance, movement direction, release position, whether it crosses the berth boundary, whether it hits the release area, or whether it triggers the close control. When determined to be a normal operation, the system maintains the embedded dwell state; when determined to be a release operation, the system switches the object from the embedded dwell state to a detached state, a hidden state, or an exit state. This method can reduce the risk of the dwelling object accidentally exiting due to page scrolling, reading clicks, or minor accidental touches. The system can also retain or release the agent's identity, context space, task state, and feedback area association based on the result of maintaining or releasing the embedded dwell state.

[0191] As a result, users can determine whether an agent or semantic entry point exists temporarily, remains in standby with low interference, or resides for a long time through spatial dragging operations, thus forming a more stable in-screen agent residency relationship than a regular floating window.

[0192] Example 10: Optional Functional Sub-region Configuration within the Residency Area

[0193] like Figure 10 As shown, in one extended implementation, the system can configure multiple functional sub-regions within a top dwelling area, a bottom dwelling area, an edge dwelling area, a berth area, or a strip-shaped standby area. These functional sub-regions are optional configurations within a multi-layer dwelling architecture and may include a first sub-region, a second sub-region, a third sub-region, or more sub-regions.

[0194] Different functional sub-areas can correspond to different task types, feedback strategies or feedback depths, processing entities, permission scopes, or result display methods. For example, the first sub-area can correspond to in-depth explanation or structural analysis configurations, the second sub-area can correspond to brief descriptions or immediate feedback configurations, and the third sub-area can correspond to save, archive, or other subsequent processing configurations.

[0195] In one implementation, different functional sub-regions are distinguished by differences in color, brightness, transparency, borders, halo, texture, icons, or dynamic prompts. These visual differences serve to indicate to the user the configuration relationships between the different sub-regions without requiring the user to open menus or perform complex settings.

[0196] When a resident object has established a resident relationship with a target resident area, the system can write a functional sub-area identifier or a corresponding configuration identifier into the corresponding resident status record, and maintain the association between the identifier and the resident object, the target resident area, and related tasks or feedback configurations.

[0197] In one implementation, the system reads the functional sub-region identifier or corresponding configuration identifier to determine at least one of the following: task type, processing subject, feedback area, feedback strategy, feedback depth, permission scope, or result display method. This invention does not rely on geometric hit determination or spatial delivery algorithms for functional sub-regions as essential technical features; its technical focus lies in the configuration association and continuous maintenance of sub-regions within the residency relationship.

[0198] Example 11: Feedback Depth Division of Labor between Upper and Lower Layers

[0199] In one implementation, the upper and lower dwell areas not only represent different screen positions but can also correspond to different feedback depths. The upper dwell area can correspond to long explanations, in-depth analysis, global observation, structural decomposition, background retrieval, or complex task processing; the lower dwell area can correspond to short explanations, instant feedback, word meaning prompts, learning assistance, low-interference question answering, or brief result display.

[0200] When a user submits the same target content or semantic object to the upper-level storage area, the system can call the corresponding agent to perform in-depth interpretation or complex analysis; when the user submits the target content or semantic object to the lower-level storage area, the system can call the corresponding agent to perform short interpretation or immediate feedback.

[0201] The aforementioned deep division of feedback does not require the upper and lower layers to be fixed to a specific function, but can be dynamically configured according to the application scenario, user settings, agent type, or current task status.

[0202] Example 12: Semantic State Maintenance of Resident Objects

[0203] In one implementation, the resident objects maintained by the system include not only ordinary visual windows, but also resident agents, movable observation units, semantic entry points, semantic object carrier units to be delivered, or other objects with intelligent processing capabilities or semantic processing entry capabilities. The system maintains not only the display position or visibility state for the resident objects, but also the agent's identity, resident area, context space, task status, feedback area, permission scope, limited progress status, or result feedback relationship.

[0204] Therefore, when a user drags a resident object from a free area into a berthed area, restores it from a hidden state, or switches between different resident layers, the system not only adjusts its visual hierarchy but also maintains its contextual spatial isolation, task state continuation, feedback area binding, and limited progress state perception relationships. In this way, the multi-layered areas of the screen are not merely an arrangement of interface windows, but a state maintenance structure in which multiple intelligent processing entities exist stably, work independently, and collaborate in a controlled manner within the same current visual field.

[0205] Example 13: Resident Status Recording and Continuous Binding Relationship

[0206] In one implementation, a first agent resides in an upper-level residency region, and a second agent resides in a lower-level residency region. The system establishes a first residency state record and a second residency state record, respectively. The first residency state record associates the identity of the first agent with the upper-level residency region, the first context space, the first task identifier, and the first feedback region; the second residency state record associates the identity of the second agent with the lower-level residency region, the second context space, the second task identifier, and the second feedback region.

[0207] When the first agent is collapsed into a low-interference icon or temporarily hidden, the system only changes its visible state field, without deleting the first dwelling state record or merging the first context space into the second context space. When the first agent is displayed again, the system reads the first dwelling state record and redetermines the association between its dwelling area, context space, task state, and feedback area.

[0208] When a user performs an explicit removal, dragging out of the residency area, closing command, or other cancellation operation, the system marks the corresponding residency status record as cancelled or invalid, and releases, archives, or retains the associated status according to a preset strategy. Thus, the establishment, maintenance, restoration, and cancellation of residency relationships form a closed loop, independent of the switching of current activity response rights.

[0209] VI. System Structure

[0210] like Figure 1 As shown, the present invention also provides an intelligent agent independent residing system based on multiple layers of screen area. This section describes the system structure in a modular manner, where each module can be implemented by software, hardware, firmware, or a combination thereof, and can execute the aforementioned method steps individually or collaboratively.

[0211] 1. Current screen display module

[0212] Used to display the current screen page. The current screen page includes at least one of the following: text, image, video, webpage, document, chat content, code, map, product information, icon, button, subtitle, multimedia object, or other visual content.

[0213] 2. Residency Area Setting Module

[0214] This is used to set multiple agent dwelling areas or dwelling layers in the current screen page. The agent dwelling areas or dwelling layers include at least one of the following: upper dwelling area, lower dwelling area, middle free area, side dwelling area, edge berth area, top berth area, bottom berth area, floating dwelling area, or other screen areas that can accommodate agent dwelling.

[0215] 3. Agent generation or invocation module

[0216] This is used to generate or invoke multiple intelligent agents. The intelligent agents include at least one of the following: artificial intelligence models, intelligent assistants, semantic processing agents, translation agents, explanation agents, summarization agents, image recognition agents, learning assistance agents, task execution agents, retrieval agents, or other agents with intelligent processing capabilities.

[0217] 4. Agent allocation module

[0218] This module is used to assign at least two agents to corresponding agent dwelling areas or dwelling layers. The agent assignment module can determine the dwelling area or dwelling layer corresponding to each agent based on agent type, task type, user command, interface state, current page content, or user drag-and-drop operation.

[0219] 5. Resident Status Recording and Binding Module

[0220] This module is used to establish residence status records for agents assigned to different residence areas or residence layers; the residence status record includes at least agent identity identifier, residence area identifier, context space identifier, task identifier, and feedback area identifier, and the above identifiers are continuously bound together; the module points to the corresponding context space, task status, and feedback area according to the identifiers in each residence status record.

[0221] 6. Context Space Isolation Module

[0222] This is used to maintain separate context spaces for different resident agents. The context spaces for different resident agents are distinct from each other, and by default, they are not merged or shared.

[0223] 7. Task Status Management Module

[0224] This is used to maintain the task states corresponding to different resident agents. The task states include at least one of the following: standby state, processing state, paused state, waiting for user confirmation state, waiting for external result state, completed state, or completed and awaiting feedback state.

[0225] 8. Feedback Area Management Module

[0226] This is used to set or maintain feedback regions for different resident agents. The feedback regions of different resident agents are isolated from each other to avoid overlapping, obscuring, or confusing feedback results.

[0227] 9. Limited Progress State Awareness Module

[0228] This is used to provide limited task progress status awareness to different resident agents while maintaining the isolation of their context spaces. The limited task progress status includes at least one of the following: whether the task has started, whether the task has been completed, processing progress, whether a result has been generated, whether user confirmation is pending, whether an anomaly exists, whether assistance from other agents is needed, or other status information that does not contain complete context content. The limited task progress status is represented by at least one of the following: status identifier, status label, progress identifier, completion identifier, anomaly identifier, assistance request identifier, or read-only status summary, and by default, the complete target content, complete historical interaction records, or complete context space are not exposed.

[0229] 10. Multidimensional State Management Module

[0230] This is used to maintain at least the dwell relationship state and the display state, and can further maintain one or more of the docking state and the task state. The dwell relationship state is used to indicate that the user is not dwelling, has dwelled, or the dwell relationship has been terminated. The docking state is used to indicate that the user is in a free state, a normal dwelling state, or an embedded dwelling state. The display state is used to indicate that the user is showing, hiding, collapsing, or displaying with low interference. The task state is used to indicate that the user is waiting, processing, paused, waiting for confirmation, waiting for external results, completed, or completed and awaiting feedback. Each state dimension is updated separately. When the display state, docking state, or task state changes, the dwell relationship state remains unchanged until a dwell termination condition is detected.

[0231] 11. Embedded resident control module

[0232] Used to obtain the docking determination result for intelligent agents or other mobile intelligent processing entry points; when the docking determination result shows that the preset docking conditions are met, it is converted from a free state to an embedded dwelling state, a corresponding dwelling state record is established or updated, and a stable binding relationship between it and the corresponding dwelling area is maintained if the release conditions are not met.

[0233] 12. Optional Function Sub-region Configuration Module

[0234] This is used to configure multiple functional sub-regions within at least one loitering area, berth area, or standby area, and to further associate loitering status records with functional sub-region identifiers and corresponding configuration identifiers.

[0235] 13. Functional Sub-region Association Module

[0236] Used to maintain at least one configuration relationship among the intelligent agent or other mobile intelligent processing entry point, target dwell area and task type, feedback strategy, feedback depth, processing subject, permission scope or result display method based on the functional sub-region identifier or corresponding configuration identifier in the dwell state record; different functional sub-regions can be distinguished by low-interference visual differences.

[0237] 14. Module for Maintaining and Terminating Resident Relationship

[0238] Used to maintain the residency status record and its binding relationship after the residency relationship is established; when the agent is hidden, folded, displayed with low interference, or temporarily invisible, only the display status is updated and the residency relationship status and binding relationship are maintained. When the display is restored, the corresponding association is restored according to the residency status record; when drag-out, long press to release, close command, detach operation, explicit removal, or other release conditions are detected, the corresponding residency relationship is released and the association status is processed according to the preset strategy.

Claims

1. A method for independent residence of intelligent agents based on multi-layered screen regions, characterized in that, Includes the following steps: S1 displays the current screen page; S2, set multiple agent dwelling areas or dwelling layers in the current screen page; S3, generates or invokes multiple intelligent agents; S4, assign at least two agents to their respective agent residence regions or residence layers; S5, establish a dwelling state record for each of the at least two agents, wherein the dwelling state record includes at least an agent identity identifier, a dwelling area identifier, a context space identifier, a task identifier, and a feedback area identifier, and make the above identifiers form a persistent binding relationship; S6, based on the identifier in each dwelling state record, read and write the corresponding context space, update the corresponding task state, and associate the processing result with the corresponding feedback area, so that the context space, task state and feedback area of ​​different dwelling agents can be distinguished from each other and will not be automatically merged because they are all dwelling on the current screen page; S7, while maintaining the isolation of the context space of different resident agents, provides at least one resident agent with a limited summary of the task progress status of another resident agent, and by default does not expose the complete target content, complete historical interaction records, complete reasoning process or complete processing results; S8. When there is an intelligent agent or other movable intelligent processing entry that changes from a free state to an embedded dwell state based on the docking determination result, the corresponding continuous binding relationship will not be terminated due to ordinary clicks, page scrolling, light touches or other non-termination operations if no preset release operation is detected. S9, maintain the dwell relationship status field and the display status field respectively, update the display status field and maintain the dwell relationship status field and the persistent binding relationship when a display status change event is detected, and restore at least one of the corresponding dwell area, context space, task status or feedback area association according to the dwell status record when the display is restored; S10, when a residency release condition is detected, the corresponding residency relationship is released, and the status associated with the residency status record is processed according to a preset strategy.

2. The method according to claim 1, characterized in that, Step S2 includes setting the plurality of intelligent agent residence areas or residence layers as at least two of the following: upper residence area, lower residence area, middle free area, side residence area, edge berth area, top berth area, bottom berth area, and suspended residence area, or setting them as multiple mutually distinguishable areas of the same type.

3. The method according to claim 1, characterized in that, In step S5, the dwell status record further includes one or more of the following: permission identifier, docking mode status field, task status field, status version information, limited progress status identifier, release condition identifier, functional sub-area identifier, or corresponding configuration identifier.

4. The method according to claim 1, characterized in that, Step S4 includes: assigning the agent to the corresponding agent dwelling area or dwelling layer according to at least one of agent type, task type, user instruction, interface state, current page content, user drag operation, target area hit relationship or dwelling strategy.

5. The method according to claim 1, characterized in that, Step S6 includes: maintaining different context spaces for different resident agents; the context space includes one or more of user instructions, target content, current visible field information, in-screen semantic messages, historical feedback content, processing results, task parameters or permission range; unless user authorization, explicit delivery, task continuation or permission release conditions are detected, the system will not automatically merge the complete context content of one resident agent into the context space of another resident agent.

6. The method according to claim 1, characterized in that, Step S6 includes: maintaining different feedback areas for different resident agents; the feedback area includes at least one of the following: the interior of the resident area, the area adjacent to the resident area, the screen edge area, the floating display area, the agent card area, the voice feedback channel, or other feedback spaces associated with the corresponding agent.

7. The method according to claim 1, characterized in that, In step S2, the same agent residence area or residence layer accommodates one or more agents; when the same agent residence area or residence layer accommodates multiple agents, steps S5 and S6 include: maintaining the residence state record, context space, task state and feedback area of ​​each of the multiple agents respectively.

8. The method according to claim 1, characterized in that, The intelligent agent includes a free intelligent agent; step S4 includes: obtaining a docking determination result for the free intelligent agent and the target dwelling area; when the docking determination result shows that the preset docking conditions are met, the free intelligent agent is converted from a free state to a dwelling state; step S5 includes: establishing or updating the corresponding dwelling state record, and making the intelligent agent identity of the free intelligent agent continuously bound to the target dwelling area, context space, task state, and feedback area; the docking determination result is generated based on at least one of spatial relationship determination, operation confirmation, location matching, dwell time, or entrance identification.

9. The method according to claim 1, characterized in that, In step S8, the preset release operation includes at least one of dragging out the dwell area, long press to release, close command, separation operation, explicit removal, dragging away beyond the preset distance, hitting the release area or triggering the close control; the non-release operation includes at least one of normal click, page scrolling, light touch, short pause, non-target area touch or touch operation that does not meet the preset movement distance.

10. The method according to claim 1, characterized in that, Step S9 further includes: maintaining at least one of the docking mode status field and the task status field; the docking mode status field is used to represent a free mode, a normal resident mode, or an embedded resident mode, and the task status field is used to represent a standby, processing, paused, waiting for confirmation, waiting for external results, completed, or completed and awaiting feedback; each status field is updated separately, and when the display status field, docking mode status field, or task status field changes, the resident relationship status field remains unchanged until a resident release condition is detected.

11. The method according to claim 1, characterized in that, In step S7, the limited task progress status summary includes at least one of the following: whether the task of another resident agent has started, whether it has been completed, processing progress, whether the result has been generated, whether user confirmation is pending, whether there is an anomaly, whether assistance from other agents is required, or whether the result can be retried.

12. The method according to claim 1, characterized in that, The limited task progress status summary is represented by at least one of status identifier, status label, progress identifier, completion identifier, exception identifier, assistance request identifier, reflow available identifier, or read-only status summary, and is used to express the task stage-aware relationship between different resident agents, without relying on full context sharing or semantic message transmission protocols.

13. The method according to claim 1, characterized in that, Step S2 includes: configuring multiple functional sub-regions within at least one agent dwelling area, berth area, or standby area; Step S5 includes: further including functional sub-region identifiers and corresponding configuration identifiers in the dwelling status record; after the agent or the mobile intelligent processing entry has established a dwelling relationship with the agent dwelling area, berth area, or standby area, the system determines at least one of the following based on the functional sub-region identifiers or corresponding configuration identifiers: processing task, feedback depth, processing subject, permission scope, or result display method.

14. A system for independent residence of intelligent agents based on multiple screen regions, characterized in that, include: The current screen display module is used to display the current screen page; The dwelling area setting module is used to set multiple intelligent agent dwelling areas or dwelling layers in the current screen page; The agent generation or invocation module is used to generate or invoke multiple agents. The agent allocation module is used to assign at least two agents to corresponding residence areas or residence layers; The dwelling state recording and binding module is used to establish dwelling state records for the at least two agents, including at least agent identity identifier, dwelling area identifier, context space identifier, task identifier and feedback area identifier, and to make the above identifiers form a continuous binding relationship; The context space isolation module, task state management module, and feedback area management module are used to maintain the context space, task state, and feedback area corresponding to different resident agents according to the identifiers in each resident state record, so that they are not automatically merged because they are all resident on the current screen page. The finite progress state-aware module is used to provide at least one resident agent with a finite task progress state summary of another resident agent while maintaining the isolation of the context space of different resident agents, and by default does not expose the complete target content, complete historical interaction records, complete reasoning process or complete processing results. An embedded dwell control module is used to prevent the corresponding continuous binding relationship from being terminated due to ordinary clicks, page scrolling, light touches or other non-termination operations when there is an intelligent agent or other mobile intelligent processing entry that changes from a free state to an embedded dwell state based on the docking determination result, and no preset release operation is detected. The multi-dimensional state management module is used to maintain a dwell relationship state field and a display state field respectively. When the display state changes, it updates the display state field while maintaining the dwell relationship state field and the persistent binding relationship. When restoring the display, it restores at least one of the following associations based on the dwell state record: the corresponding dwell area, context space, task state, or feedback area. The residency relationship maintenance and termination module is used to terminate the corresponding residency relationship and process the associated status according to a preset strategy when a residency termination condition is detected.

15. The system according to claim 14, characterized in that, The multi-dimensional state management module is also used to maintain at least one of the docking mode state field and the task state field; the docking mode state field is used to represent the free mode, the normal dwell mode, or the embedded dwell mode, and the task state field is used to represent the standby, processing, paused, waiting for confirmation, waiting for external results, completed, or completed and awaiting feedback; when the display state field, the docking mode state field, or the task state field changes, the multi-dimensional state management module keeps the dwell relationship state field unchanged until a dwell release condition is detected.

16. The system according to claim 14, characterized in that, The limited progress state awareness module is further used to provide different resident agents with limited state awareness of whether the task has started, whether it has been completed, the processing progress, whether the result has been generated, whether user confirmation is pending, whether there is an anomaly, whether other agents' assistance is needed, or whether the result can be retried, through one or more of the following: state identifier, state label, progress identifier, completion identifier, anomaly identifier, assistance request identifier, retrievable identifier, or read-only state summary. By default, the complete target content, complete historical interaction records, complete reasoning process, or complete processing result are not exposed.

17. The system according to claim 14, characterized in that, The system further includes an optional functional sub-region configuration module and / or a functional sub-region association module; the optional functional sub-region configuration module is used to configure multiple functional sub-regions within at least one dwelling area, berth area, or standby area, so that the dwelling status record further includes a functional sub-region identifier and a corresponding configuration identifier; the functional sub-region association module is used to maintain at least one configuration relationship between the intelligent agent or the mobile intelligent processing entry and the processing task, feedback depth, processing subject, permission scope, or result display method based on the functional sub-region identifier or the corresponding configuration identifier.

18. The system according to claim 14, characterized in that, The embedded dwelling control module is further used to obtain the docking determination result for the intelligent agent or other mobile intelligent processing entry; when the docking determination result shows that the preset docking conditions are met, the corresponding object is converted from the free state to the embedded dwelling state, the corresponding dwelling state record is established or updated, and the stable binding relationship between the object and the target dwelling area is maintained when no preset release operation or release condition is detected. The preset release operation or release condition includes at least one of the following: drag out, long press to release, close command, detach operation, explicit removal, or drag away operation exceeding a preset distance.

19. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 13.

20. An electronic device comprising a processor and a memory, characterized in that, The memory stores a computer program, and when the processor executes the computer program, it implements the method as described in any one of claims 1 to 13.

Images