IBM AI ticket reconciliation

Overview

I led design strategy and execution for Guardium Insights' AI Ticket Reconciliation feature, transforming a heavily manual SOX compliance workflow into an automated, AI-powered experience.

The solution supports IBM's broader initiative to automate compliance, auditing, and reporting—helping compliance analysts reduce manual effort, improve accuracy, and focus on higher-value work.

The problem

Data security analysts spend an average of 3 hours per day performing manual tasks across multiple tools to validate unauthorized data changes for SOX compliance.

The opportunity

With 4,300 publicly traded U.S. companies subject to SOX compliance, automating ticket reconciliation represents a $3B platform opportunity—reducing systemic compliance risk, driving operational leverage at scale, and positioning Guardium as a differentiated, AI-first enterprise solution.

4,300

Publicly traded U.S. companies require SOX compliance

$3B

Market opportunity for IBM

Manual

Competitor solutions still rely heavily on manual workflows

Research and discovery

I partnered with stakeholders, SMEs, and customers to deeply understand the reconciliation workflow.

Key activities

Stakeholder design thinking workshops
SME and customer interviews
As-is journey mapping and competitive analysis

1. Stakeholder design thinking workshops

I led cross-functional design workshops to align on goals, explore concepts, and prioritize solutions, using IBM's Big Idea and Crazy 8s exercises mapped against an impact-vs-investment framework—ultimately converging on an AI-powered Watsonx solution to automate SOX ticket reconciliation.

Design thinking workshop with Big Idea and Crazy 8s exercises

2. SME and customer interviews

I interviewed IBM SME's and customers to understand real-world SOX ticket reconciliation workflows and pain points.

Key pain points

1.

Manual and time-consuming process

Data security analysts spend 3 hours daily reconciling SOX change tickets, manually cross-referencing multiple tools.

2.

Error-prone process

Outdated methods increase the risk of mistakes, potentially leading to compliance violations.

3.

Approval delays and risks

Slow approvals and inefficient workflows create security gaps and regulatory risks.

4.

Loss productivity

Analysts spend excessive time on this process instead of focusing on higher-value security and compliance tasks.

3. As-Is journey mapping and competitive analysis

I mapped the end-to-end reconciliation workflow to identify high-friction automation opportunities, uncovering a 15-step manual process and a competitive landscape still heavily dependent on human intervention with minimal AI support.

Manual reconciliation workflow

15-step reconciliation process

Step 1

Review ticket and task details

Step 2

Verify guardium event logs

Step 3

Check event classification

Step 4

Cross reference ticket and task with relevant controls

Step 5

Verify user access records

Step 6

Audit and validate data integrity

Step 7

Check the timeline of events

Step 8

Confirm match of task and ticket timelines

Step 9

Review action taken on identified issues

Step 10

Validate evidence of remediation

Step 11

Check for impact on reports

Step 12

Reconcile audit logs

Step 13

Ensure documentation of findings

Step 14

Check for additional notifications

Step 15

Task closure (approval or rejection)

The challenge

How might we automate SOX ticket reconciliation to reduce manual effort, minimize errors, and improve analyst confidence—without removing human oversight?

The solution

I designed an AI-powered reconciliation experience using IBM’s Watsonx, embedded directly within existing compliance workflows to automate reconciliation without disrupting established audit processes.

Key features

AI ticket matching

Automatically correlates database change logs with the correct ticket, eliminating manual cross-referencing.

Context-aware decision support

Ranks and surfaces relevant matches with confidence scores to enable faster, high-confidence decisions.

Seamless integration

Direct integration with ServiceNow and other ticketing systems removed copy-paste workflows and tool switching, enabling reconciliation within a single audit-ready interface.

Workflow automation

Eliminates redundant steps while preserving compliance traceability and documentation integrity.

Design process

Rapid iteration with task flows, wireframes, and low-fi designs
Usability testing with customers, IBM Business Partners, and tech sellers
Feedback validated confidence, speed, and decision clarity

1. Design exploration and validation

Because ticket comparison was the core of the experience, I explored four distinct design approaches, each presenting ticket match results and task metadata in different formats—from tabular layouts with metadata surfaced above the table to variations using a right-side panel to display ticket match insights or contextual metadata. Through SME and user feedback, the final design narrowed each ticket to high-signal data, retained a familiar tabular structure, and introduced sub-rows to enable direct comparison between ticket details and internal system logs for faster, more confident evaluation.

Key insights from user testing

Feature was highly valued
AI similarity scores increased confidence
Participants expected significant time savings

"This is the first I've seen that actually correlates logic to activity."
— U.S. Banking Customer

Final ticket comparison screen showcasing AI explainability tooltips and results sidepanel.

Exploration of four distinct design ideas for the ticket matching screen and flows.

IBM AI Ticket Reconciliation hero mockup

Various screens of the ticket reconciliation flow including configuration, workflow setup, integrations and various detail screens.

2. UI design - Carbon Design System components

I designed the solution as a scalable, reusable component set aligned with the IBM Carbon Design System, providing detailed development guidance and a production-ready Figma handoff that enabled engineers to implement efficiently with minimal design follow-up.

The final experience focused on:

Clear prioritization
Fast decision-making
Low cognitive load

The images below show the dev-ready screen specifications including component logic, functionality, and design tokens.

Impact

75%

Efficiency improvement

90 sec/task

Ticket reconciliation time reduced to minutes

8

Manual steps eliminated

Challenges and outcomes

Key Challenges

Complex workflows
Simplified a fragmented 15-step process into a streamlined journey.
Limited user pool
Validation relied on a small number of qualified compliance users.
Post-validation changes
Designs were refined to align with backend updates and MCP server integration.

Outcome

Delivered an AI-powered reconciliation feature that:

Reduced manual effort
Improved compliance accuracy
Supported IBM's long-term automation strategy

What I learned

AI features must build trust, not just automate
Confidence signals (like similarity scores) are as important as speed
Designing for compliance requires balancing automation with human control

IBM AI-driven compliance automation