Artificial intelligence (AI)-driven standardized intake tool
The disclosed techniques pertain to artificial intelligence (AI) techniques. Particularly, techniques are disclosed for providing an AI-driven standardized intake tool. An intake request can be accessed and processed using a plurality of computational models to generate a set of results. The set of results and a machine-learning model can be used to predict a set of workflow actions for the intake request and a set of tools for an entity associated with the intake request can be updated based on the set of workflow actions. Updating the set of tools for the entity can initiate a workflow for incorporating a target of the intake request within the entity.
1 . A computer-implemented method comprising:
accessing, from a configuration management database (CMDB) storing a catalogue of applications deployed within an entity, an intake request associated with a process for deploying an application within the entity, wherein the intake request identifies an intake type for the intake request;
applying a rule-based model to the intake request to determine that the intake request includes a valid intake type, a target team, and an initial set of impacted applications;
in response to determining that the intake request includes the valid intake type, the target team, and the initial set of impacted applications, processing the intake request by:
generating a feature representation of the intake request by extracting structured attributes and vectorized metadata from the intake request and related CMDB records;
executing a machine-learning model on the feature representation to predict additional impacted applications deployed by the entity, wherein the machine-learning model is executed on a computing platform co-located with the CMDB;
retrieving a set of historical impacted applications by executing rules that query the CMDB for prior intake requests with the same intake type and returning prior matched applications;
combining the additional impacted applications and the set of historical impacted applications to produce a consolidated set of results;
inputting the consolidated set of results into a workflow prediction model that is fine-tuned on prior intake-to-deployment mappings to output a set of workflow actions comprising scheduled workflow tasks, resource reservations, and message templates for participants; and
automatically updating one or more enterprise tools by programmatically:
creating or updating a project entry in a project management database with a Gantt chart generated from the scheduled workflow tasks;
reserving calendar resources via a resource scheduling tool for the scheduled workflow tasks to avoid scheduling conflicts; and
populating a messaging tool with notification messages and participant lists derived from the message templates, wherein automatically updating the one or more enterprise tools initiate the process for deploying the application.
2 . The computer-implemented method of claim 1 , further comprising:
updating the rule-based model by deriving a new rule from data and information collected while processing the intake request, and storing the new rule in the CMDB.
3 . The computer-implemented method of claim 1 , further comprising:
fine-tuning the machine-learning model with labelled training data generated from the intake request and related CMDB records collected during processing so that the workflow prediction model is improved for future intake-to-deployment mappings.
4 . The computer-implemented method of claim 1 , wherein the machine-learning model comprises a transformer-based model that accepts the feature representation and leverages learned embeddings of application metadata to predict additional impacted applications.
5 . The computer-implemented method of claim 1 , further comprising:
performing a capacity check by comparing schedule information for currently active projects retrieved from the project management database and resource scheduling tool against the scheduled workflow tasks; and, in response to determining that impacted projects are at capacity, automatically flagging the intake request and modifying the set of workflow actions to include an escalation or rescheduling action.
6 . The computer-implemented method of claim 1 , further comprising:
enforcing access controls by authenticating a user or system requesting the intake request and by verifying permissions to modify the project management database, resource scheduling tool, and messaging tool prior to automatically updating the one or more enterprise tools.
7 . The computer-implemented method of claim 1 , wherein the Gannt chart comprises creating one or more visualizations that depict the scheduled workflow tasks, task dependencies, resource assignments, and timeline, and storing the visualizations in the project management database for access by participants.
8 . A system comprising:
one or more processors; and
one or more computer-readable media storing instructions which, when executed by the one or more processors, cause the system to perform operations comprising:
accessing, from a configuration management database (CMDB) storing a catalogue of applications deployed within an entity, an intake request associated with a process for deploying an application within the entity, wherein the intake request identifies an intake type for the intake request;
applying a rule-based model to the intake request to determine that the intake request includes a valid intake type, a target team, and an initial set of impacted applications;
in response to determining that the intake request includes the valid intake type, the target team, and the initial set of impacted applications, processing the intake request by:
generating a feature representation of the intake request by extracting structured attributes and vectorized metadata from the intake request and related CMDB records;
executing a machine-learning model on the feature representation to predict additional impacted applications deployed by the entity, wherein the machine-learning model is executed on a computing platform co-located with the CMDB;
retrieving a set of historical impacted applications by executing rules that query the CMDB for prior intake requests with the same intake type and returning prior matched applications;
combining the additional impacted applications and the set of historical impacted applications to produce a consolidated set of results;
inputting the consolidated set of results into a workflow prediction model that is fine-tuned on prior intake-to-deployment mappings to output a set of workflow actions comprising scheduled workflow tasks, resource reservations, and message templates for participants; and
automatically updating one or more enterprise tools by programmatically:
creating or updating a project entry in a project management database with a Gantt chart generated from the scheduled workflow tasks;
reserving calendar resources via a resource scheduling tool for the scheduled workflow tasks to avoid scheduling conflicts; and
populating a messaging tool with notification messages and participant lists derived from the message templates, wherein automatically updating the one or more enterprise tools initiate the process for deploying the application.
9 . The system of claim 8 , the operations further comprising:
updating the rule-based model by deriving a new rule from data and information collected while processing the intake request, and storing the new rule in the CMDB.
10 . The system of claim 8 , the operations further comprising:
fine-tuning the machine-learning model with labelled training data generated from the intake request and related CMDB records collected during processing so that the workflow prediction model is improved for future intake-to-deployment mappings.
11 . The system of claim 8 , wherein the machine-learning model comprises a transformer-based model that accepts the feature representation and leverages learned embeddings of application metadata to predict additional impacted applications.
12 . The system of claim 8 , the operations further comprising:
performing a capacity check by comparing schedule information for currently active projects retrieved from the project management database and resource scheduling tool against the scheduled workflow tasks; and, in response to determining that impacted projects are at capacity, automatically flagging the intake request and modifying the set of workflow actions to include an escalation or rescheduling action.
13 . The system of claim 8 , the operations further comprising:
enforcing access controls by authenticating a user or system requesting the intake request and by verifying permissions to modify the project management database, resource scheduling tool, and messaging tool prior to automatically updating the one or more enterprise tools.
14 . The system of claim 8 , wherein the Gannt chart comprises creating one or more visualizations that depict the scheduled workflow tasks, task dependencies, resource assignments, and timeline, and storing the visualizations in the project management database for access by participants.
15 . One or more non-transitory computer-readable media storing instructions which, when executed by one or more processors, cause a system to perform operations comprising:
accessing, from a configuration management database (CMDB) storing a catalogue of applications deployed within an entity, an intake request associated with a process for deploying an application within the entity, wherein the intake request identifies an intake type for the intake request;
applying a rule-based model to the intake request to determine that the intake request includes a valid intake type, a target team, and an initial set of impacted applications;
in response to determining that the intake request includes the valid intake type, the target team, and the initial set of impacted applications, processing the intake request by:
generating a feature representation of the intake request by extracting structured attributes and vectorized metadata from the intake request and related CMDB records;
executing a machine-learning model on the feature representation to predict additional impacted applications deployed by the entity, wherein the machine-learning model is executed on a computing platform co-located with the CMDB;
retrieving a set of historical impacted applications by executing rules that query the CMDB for prior intake requests with the same intake type and returning prior matched applications;
combining the additional impacted applications and the set of historical impacted applications to produce a consolidated set of results;
inputting the consolidated set of results into a workflow prediction model that is fine-tuned on prior intake-to-deployment mappings to output a set of workflow actions comprising scheduled workflow tasks, resource reservations, and message templates for participants; and
automatically updating one or more enterprise tools by programmatically:
creating or updating a project entry in a project management database with a Gantt chart generated from the scheduled workflow tasks;
reserving calendar resources via a resource scheduling tool for the scheduled workflow tasks to avoid scheduling conflicts; and
populating a messaging tool with notification messages and participant lists derived from the message templates, wherein automatically updating the one or more enterprise tools initiate the process for deploying the application.
16 . The one or more non-transitory computer-readable media of claim 15 , further comprising:
updating the rule-based model by deriving a new rule from data and information collected while processing the intake request, and storing the new rule in the CMDB.
17 . The one or more non-transitory computer-readable media of claim 15 , further comprising:
fine-tuning the machine-learning model with labelled training data generated from the intake request and related CMDB records collected during processing so that the workflow prediction model is improved for future intake-to-deployment mappings.
18 . The one or more non-transitory computer-readable media of claim 15 , wherein the machine-learning model comprises a transformer-based model that accepts the feature representation and leverages learned embeddings of application metadata to predict additional impacted applications.
19 . The one or more non-transitory computer-readable media of claim 15 , the operations further comprising:
performing a capacity check by comparing schedule information for currently active projects retrieved from the project management database and resource scheduling tool against the scheduled workflow tasks; and, in response to determining that impacted projects are at capacity, automatically flagging the intake request and modifying the set of workflow actions to include an escalation or rescheduling action.
20 . The one or more non-transitory computer-readable media of claim 15 , the operations further comprising:
enforcing access controls by authenticating a user or system requesting the intake request and by verifying permissions to modify the project management database, resource scheduling tool, and messaging tool prior to automatically updating the one or more enterprise tools.