1. Introduction to the AI Analyst Agent
The AI Analyst Agent is an agentic AI assistant integrated into the ITSM Application, engineered to optimise ticket management for IT Analysts. Operating via a unified chat-based interface, the agent automates routine monitoring, data collection, and record updates. This enables analysts to oversee their complete workload, adhere to critical deadlines, and maintain service level expectations without the inefficiency of context switching or manual record tracking.
Key Objectives
- Efficiency: Eliminate the manual review burden by providing instantaneous access to ticket intelligence and administrative actions through natural language processing.
- Workload Management: Support the proactive prioritisation of urgent incidents and service requests via automated, data-driven summaries.
- Compliance: Maintain the integrity of service records through schema-defined mandatory field validation and comprehensive audit trails.
2. Automatic Ticket Query and Summarisation
Initiate the retrieval of concise, AI-generated overviews of assigned work to circumvent manual searches within the ITSM Application UI.
List vs Summarisation Workflow
The agent distinguishes between high-level retrieval and deep-dive summarisation:
- List Command: Requesting a list (e.g., “Please list all my Incidents”) prompts the agent to return the first 10 incidents assigned to the analyst, including Ticket ID, Symptom, Priority, and Status.
- Summary Command: Requesting a summary (e.g., “show me a summary of tickets assigned to me” or “What’s the status of my tickets?”) triggers a detailed AI-driven synthesis of all open and in-progress records.
Summary Components
| Data Point | Description |
|---|---|
| Ticket Status | The current lifecycle state (e.g., New, In-Progress, Pending). |
| Priority | The urgency level (e.g., P1 – Critical, P2 – High). |
| Workgroup | The specific functional team assigned to the record (e.g., Application Support). |
| Last Updated Time | Precise timestamp of the most recent modification. |
| Symptom | The brief statement of the observed issue. |
| AI-Generated Summary | A narrative paragraph synthesising current progress, missing details, and impact. |
| Recommended Next Steps | AI-generated advice for follow-up actions (e.g., “Continue monitoring and resolution”). |
Prioritisation Logic
The agent does not present summaries in random order. All outputs are structured according to ticket metadata, prioritising records based on urgency, criticality, or impending SLA breaches.
Error Handling Protocols
- No Active Assignments: If no open or in-progress tickets are detected, the agent responds: There are no open incidents assigned to you at the moment.
- System Failure: If a query or summarisation process fails, the agent communicates: Sorry, I was unable to retrieve your ticket summary. Please try again later.
3. Creating Incidents on Behalf of End Users
Execute the logging of incidents for other users through a conversational workflow that ensures data accuracy and formalised documentation.
User Identification and Ambiguity
Initiate the process with a conversational intent, such as: “I need to create an Incident for John.”
- Ambiguity Resolution: If multiple users match the criteria, the agent presents a numbered list of matches.
- Selection Syntax: You must follow the agent’s prompt: “Please select the correct user (enter number) or type ‘cancel’ or ‘retry’ to search again.”
Data Extraction and Formalisation
Upon user selection, provide the issue description. The AI autonomously extracts and formalises the “Symptom” and “Description” fields.
Example: Before vs After Formalisation
- Analyst Prompt: Unable to capture screenshot from Snagit tool
- AI Formalised Symptom: Snagit tool cannot capture screenshot
- AI Formalised Description: The user is unable to capture a screenshot using the Snagit tool. Details about the error message displayed, system environment, or edition of the tool being used are needed to proceed.
Information Validation
The AI evaluates input against clarity requirements. If the input is insufficient (e.g., “Snagit Tool”), the agent issues a targeted follow-up: “Could you please provide more details about the issue you’re experiencing…?”
- Strict Data Restriction: To ensure efficiency, the agent is restricted to capturing only the symptom and description. It will never request or collect non-essential fields, such as device ID or location, during this phase.
Finalisation Commands
Review the formalised details and issue one of the following commands:
'Yes': Implementation and creation of the incident in the ITSM Application.'Edit': Modification of the description.'Cancel': Immediate cessation of the creation action.
4. Ticket Identification and Status Updates via Chat
Modify existing service records directly through the chat interface to ensure real-time accuracy and reduce context switching.
Ownership and Authorisation Validation
For security and data integrity, the AI restricts updates to tickets currently assigned to the requesting analyst. If you attempt to modify an unassigned record, the agent responds: Not Authorised. Incident #[ID] is not assigned to you. You can only update incidents assigned to you.
Structured Status Transitions
The agent adheres to the logical lifecycle of the ticket schema. When an update is requested, the agent presents a numbered list of valid next statuses:
- Example: If the current status is
In-Progress, the agent suggests:1. Pending,2. Resolved,3. Cancelled. - Action: You must select the corresponding number or type the status name to proceed.
Invalid Input Handling
If you suggest an unsupported status (e.g., “Set to Finished”), the agent provides the following corrective format: Invalid status: ‘[Input]’. Only the following statuses are allowed: [Numbered List of valid statuses].
5. Advanced Update Commands and Requirements
Utilise complex interaction capabilities to accelerate administrative throughput.
Combined Commands
Execute status changes and the addition of work notes in a single conversational string.
- Syntax Example: Update [Ticket ID] to In Progress and add note: [Note Text].
- Outcome: The agent processes both updates simultaneously and returns a summarised confirmation of the changes.
Solution Before Resolution Requirement
The ITSM Application enforces strict data quality for ticket closure. Before a record can transition to a Resolved state, the AI intercepts the request and prompts: Please provide the solution to resolve [Ticket ID]. The status update is implemented only after the solution or work note is provided.
6. Mandatory Field Validation and Data Integrity
The compliance layer ensures that all schema-defined mandatory fields are populated before the database is updated.
Status Change Logic
- Verify Mandatory Fields: The AI fetches the requirements for the target status from the backend Incident Form.
- Prompt for Missing Data: If required fields are null, the AI requests the data directly in the chat.
- Input Data: Provide the information via selection or text.
- Implementation: The status change is applied only upon successful validation of all fields.
Case Study: Transition to Pending
When moving a ticket to a Pending status, the agent requires a Pending Reason. You must select one of the seven validated options:
- Under Observation
- Scheduled Ticket
- Vendor Dependency
- Awaiting user response
- User Confirmation Pending
- Third-Party Service Restoration
- Business Approval Pending
7. Live Agent Transfer (LAT) and Escalation Logic
The agent defines clear boundaries for automation, transitioning to human-to-human support for complex, high-priority, or aged issues.
Aged Service Request (SR) Triggers
The agent monitors the duration for which Service Requests remain open. If a query is made regarding an SR that exceeds a configured threshold (default is 7 days, but can be configured up to 5 weeks), the AI identifies the record as “Aged.”
Workflow and VIP Routing
- Detection: The AI identifies a query referencing an open SR number or context.
- Routing Logic: The system applies special routing logic for VIP users and aged tickets to ensure prioritised human handling.
- Notification: The agent informs the user: This service request has been open for [X] weeks, which requires special attention. I’m connecting you with a human agent…
- Confirmation: A transfer is only initiated upon a clear ‘Yes’ or ‘No’ response from the analyst or user.
8. The Unified IT Analyst Experience
The “Unified Analyst” architecture allows IT staff to manage professional duties and personal self-service needs within a single session, utilising a Single Login and One Chatbot Client.
Comparison of Capabilities
| Self-Service Features (Personal) | Analyst Tools (Professional) |
|---|---|
| Password resets | Ticket management and status updates |
| Tracking personal IT requests | Escalating and routing tickets for others |
| Knowledge Base (KB) searches | Resolving incidents for end users |
| Submitting own service requests | User administration and metadata management |
This unified environment ensures that no account switching is required. Security and permissions remain active and are applied contextually based on the specific action performed.
9. Audit, Logging, and Traceability
Every action performed by the AI Analyst Agent is recorded with 100% accuracy to ensure a complete audit trail within the ITSM Application.
Change History Metadata
The system records the following data points for every update:
- Source: Logged as “ChatBot”.
- Medium: Logged as “ChatBot”.
- Changed By: The full name of the analyst (e.g., Emily Johnson).
- Timestamp (Change Date): Precise date and time of the modification.
- Fields Updated: Including Status, Resolution Deadline, and Pending Reason.
- Value Tracking: Complete record of the “Old Value” vs the “New Value” for every modified field.
Reviewing Traceability Logs
To audit AI-driven updates within the primary application:
- Navigate to Incident > Manage Incident > Incident List.
- Open the specific Incident ID.
- From the Action panel, select Change History to view the comprehensive log of new values and automated updates.