Project Overview
As part of IBM Client Engineering, I helped design and deliver an AI-powered impact assessment solution that dramatically reduced the effort required to evaluate software upgrades. The engagement focused on automating requirement discovery and impact analysis by connecting release note changes to enterprise requirements documentation, enabling analysts to complete assessments faster, with greater confidence, and with less dependency on subject matter experts.
The resulting solution demonstrated how generative AI and semantic search could transform a traditionally manual process into a scalable, intelligent workflow, helping position additional modernization and AI adoption opportunities.
Business Challenge
The organization relied on analysts to manually review thousands of requirements and map software release changes to the appropriate functional specifications during upgrade assessments. This process:
- Required extensive manual effort.
- Relied heavily on SME knowledge.
- Introduced risk of missed requirement mappings.
- Extended upgrade planning and assessment timelines.
- Created downstream impacts for testing and validation activities.
Solution
Our team developed an AI-powered impact assessment assistant that automates requirement discovery and relevance matching. The solution:
- Analyzes software release notes and technical changes.
- Generates requirement-focused summaries.
- Performs semantic search across requirements repositories.
- Returns relevancy-ranked requirement recommendations.
- Consolidates associated risk, testing, and implementation information into a single report for analyst review.
Rather than manually reviewing thousands of records, analysts receive a curated and prioritized set of recommendations to accelerate decision-making and validation activities.
Design Process
1. Problem Framing & Workflow Analysis
- Collaborated with stakeholders to understand the current upgrade assessment process, identify bottlenecks, and define opportunities for AI-driven automation.
2. User-Centered Solution Design
- Designed an analyst-centered experience focused on reducing cognitive load while increasing traceability and confidence in assessment outcomes.
- User workflows informed recommendation ranking, reporting outputs, and information hierarchy.
3. AI Solution Strategy
- Worked with technical teams to align business needs, data sources, and AI capabilities, ensuring the solution could generate meaningful recommendations while fitting into existing assessment workflows.
4. Rapid Prototyping & Validation
- Translated requirements into a working proof of concept that demonstrated end-to-end impact analysis, requirement discovery, and report generation capabilities.
5. Executive Storytelling & Value Realization
- Developed the narrative and business value story used to demonstrate measurable productivity gains and justify future investment in AI-enabled operations.
My Role
- Led discovery and workflow analysis activities
- Mapped analyst journeys and pain points
- Designed the user experience and reporting workflow
- Bridged business requirements with AI capabilities
- Collaborated with engineers on solution strategy and validation
- Developed business value and ROI storytelling
- Created executive-facing demonstrations and solution narratives
Impact
- 88% reduction in manual review effort, narrowing review scope from ~4,000 requirements to ~500 prioritized candidates, with downstream reductions in QA, regression testing, and operational qualification effort.
- 85% Top-10 match rate, surfacing the correct requirement within the top ten recommendations for most evaluated records.
- Potential 75% reduction in assessment timelines:
- Minor upgrades: from 3–4 weeks to ~1 week
- Major upgrades: from 4–6 weeks to ~1–2 weeks