Agentic AI is a probabilistic intelligence system designed to adapt dynamically to changing data and operational environments. Unlike deterministic automation tools like RPA (Robotic Process Automation), Agentic AI makes decisions based on pattern recognition, context awareness, and statistical likelihoods, enabling it to automate complex workflows where traditional systems fall short.
This work is part of a broader cybersecurity initiative titled "Agentic AI using Secure AgenticAI", which is a four-part solution for real-time cybersecurity compliance and threat response. Developed in partnership with Momentum and SecureGPT, the project leverages large language models and intelligent agents to ensure organizations maintain cybersecurity compliance and mitigate risks effectively.
Our team, Agentic Ticket Masters, is responsible for Goal 2 of the project — building an AI-powered Service Ticket Management Agent that automates the lifecycle of service tickets, from categorization and team assignment to SLA tracking and breach notifications.
Our solution improves IT service responsiveness, ensures SLA compliance, and enhances overall support operations using agentic reasoning.
This solution leverages SecureGPT for intelligent classification and decision-making. It is deployed as a modular, serverless pipeline on AWS, triggered through S3 events and manual invocations, allowing real-time processing and alerting without human intervention.
! [Deployment] (deployment.svg)
We have provided the prompts to generate the specific code modules:
- Ticket Categoriser — Ticket_categoriser_prompt.txt
- SLA Monitor — Sla_monitor_prompt.txt
- Email Sender — Email_sender_prompt.txt
The following diagram illustrates the flow of data and interactions between the components of the system:
-
Jira Ticket Ingestion
Tickets are retrieved from Jira Cloud using the Jira REST API. -
Ticket Categorization (
Ticket_Categoriser.py)- Runs as an AWS Lambda function.
- Sends ticket data to SecureGPT to classify priority, severity, urgency, component, and team.
- Updates Jira and uploads
categorized_tickets.csvto S3.
- SLA Monitoring (
Sla_Monitor.py)- Triggered by the upload of
categorized_tickets.csv. - Evaluates whether each ticket meets SLA conditions using SecureGPT.
- Outputs results in
sla_tickets_alerts.csv.
- Triggered by the upload of
- Email Alerts (
Email_sender.py)- Triggered by the upload of
sla_tickets_alerts.csv. - Sends automated SLA breach notifications via SMTP to ticket assignees and CC'd stakeholders.
- Triggered by the upload of
- All modules are deployed as AWS Lambda functions with event-driven automation via S3 triggers.
- All credentials and tokens are securely handled via environment variables.
- GitHub Push Protection ensures secrets never reach version control.
- Logs and outputs are securely stored in Amazon S3 for transparency and traceability.
The overall Architecture is as shown below
| Script | Description |
|---|---|
Ticket_Categoriser.py |
Fetches and classifies Jira tickets via SecureGPT |
Sla_Monitor.py |
Assesses SLA compliance and generates alert CSVs |
Email_sender.py |
Sends SLA breach notifications via SMTP email |
Ensure environment variables are set or .env file is loaded, then run:
python Ticket_Categoriser.py
python Sla_Monitor.py
python Email_sender.py