AI Product / Methodology
Intelligent Requirements Engineering Methodology
Turning ambiguous requirements into structured product decisions.
A methodology combining Agile Product Management, Requirements Engineering and Generative AI to improve backlog quality, decision support and business alignment.
- Role
- Product Management / Methodology Design
- Focus
- Product Discovery · Requirements · AI
- Duration
- 3-month controlled evaluation
+18%
Backlog items delivered
Controlled evaluation · 3 months · 8 Agile teams: 4 IREM + C3P0 vs. 4 traditional
100%
Stories with explicit business success metrics
IREM-generated User Stories vs. 76% with the traditional approach, in the controlled evaluation
The challenge
Better requirements are not just about better writing.
In the context of the initiative, product teams often moved from business requests directly into backlog items. The result could be requirements that were technically understandable but weakly connected to business outcomes, difficult to validate and inconsistent in quality.
The challenge was to create a repeatable way to move from an initial problem or request to a development-ready requirement while preserving business context, measurable outcomes and decision quality.
The problem
From requests to requirements, too much context was being lost.
Ambiguity
Requirements could reach delivery teams without enough contextual clarity.
Business alignment
User Stories were not always explicitly connected to measurable business outcomes.
Consistency
The quality and completeness of requirements could vary between teams.
Validation
Acceptance criteria and test scenarios were not always derived from a clear understanding of the expected outcome.
The product question
How might we improve decision quality before development begins?
- Business problem
- Context
- Outcome
- Requirement
- Acceptance criteria
- Validation
Discovery
The real opportunity was not generating more text. It was improving the decision process.
The methodology was designed around structured discovery and decision support: each step makes a decision explicit before the next artifact is written.
- Discovery
- Context
- Outcome
- Requirement
- Validation
- Ready for delivery
The method
IREM turns requirements engineering into a structured product decision process.
IREM — Intelligent Requirements Engineering Methodology — combines Agile Product Management, Requirements Engineering and structured decision models with Generative AI.
The methodology is the core of the work. AI is used as structured decision support and to operationalize the methodology.
Understand
- Problem identification
- Stakeholder / persona definition
Frame
- Strategic objective validation
- Business outcome definition
- Measurable KPI definition
Structure
- Contextual requirement discovery
- Adaptive guidance
- Story decomposition
- Dependency mapping
Validate
- Completeness verification
- BDD acceptance criteria
- Business-oriented test scenarios
Prepare
- Definition of Ready validation
AI enablement
C3P0 operationalizes the methodology inside the product workflow.
C3P0 — Collaborative Product Planning & Prioritization Optimizer — is the AI-enabled assistant created to operationalize the IREM methodology.
Integrated into Atlassian Jira through Rovo AI.
- Development-ready User Stories
- BDD acceptance criteria
- Business-oriented test scenarios
- KPI definition
- Dependency analysis
- Definition of Ready validation
- Epic decomposition
- Feature slicing
- Value traceability
Controlled evaluation
Does structured requirements engineering actually improve delivery?
To test the methodology, eight Agile teams were compared over three months: four worked with IREM + C3P0 and four kept working with traditional requirements engineering.
8Agile teams
3months
4
IREM + C3P0
4
Traditional requirements engineering
Results
The results showed measurable improvement.
+18%
More backlog items delivered
Teams using IREM delivered 18% more backlog items than teams using conventional requirements engineering during the controlled evaluation.
100%
Explicit business success metrics
100% of User Stories generated through IREM included explicit business success metrics, compared with 76% in the traditional approach.
| IREM vs. traditional approach in the controlled evaluation | IREM | Traditional |
|---|---|---|
| Business success metrics in stories | 100% | 76% |
| Backlog items delivered | +18% | Baseline |
IREM vs. traditional approach in the controlled evaluation
Business success metrics in stories
- IREM
- 100%
- Traditional
- 76%
Backlog items delivered
- IREM
- +18%
- Traditional
- Baseline
What changed
From requirement writing to requirement quality.
Traditional flow
- Request
- User Story
- Development
IREM-oriented flow
- Problem
- Context
- Outcome
- Requirement
- Acceptance
- Validation
- Development
Quality by design
Explicit checkpoints before development begins.
The methodology introduces explicit checkpoints so governance happens before development, not after it.
- Business outcome
- Requirement completeness
- Acceptance criteria
- Testability
- Dependencies
- KPI definition
- Definition of Ready
Key product decisions
Three decisions that shaped the approach.
Structure decisions before generating artifacts
- Problem
- Requests moved directly into backlog items, losing context, outcomes and validation criteria.
- Alternative
- Use AI to write User Stories faster from the original request.
- Decision
- Make problem, persona, objective, outcome and KPI explicit first; generate artifacts only from that structure.
- Trade-off
- More upfront structuring per requirement, in exchange for consistency and traceability to business outcomes.
AI as decision support, not decision maker
- AI responsibility
- C3P0 drafts User Stories, BDD criteria, test scenarios, KPIs and dependency analysis, and checks Definition of Ready.
- Human responsibility
- Problem framing, prioritization and the product decision itself remain with the team.
- Why it mattered
- IREM is the methodology; C3P0 is its AI-enabled operationalization. Quality comes from the method, not from automation alone.
Measure quality through observable outcomes
- Decision
- Evaluate the methodology with measurable indicators instead of subjective perception.
- Design
- Controlled evaluation over 3 months: 8 Agile teams, 4 with IREM + C3P0 and 4 with traditional requirements engineering.
- Indicators
- Backlog items delivered (+18%) and User Stories with explicit business success metrics (100% vs 76%).
Trade-offs
What the approach gains, and what it asks for.
Structure vs. speed
Each requirement passes through explicit stages before delivery, instead of going straight from request to backlog.
Automation vs. judgment
AI accelerates artifact generation; framing and decisions stay with the product team.
Consistency vs. flexibility
A shared structure reduces variation between teams, while adaptive guidance adjusts discovery to each context.
My contribution
What I owned, and what the teams made possible.
My contribution
- Methodology design (IREM)
- Product discovery
- Requirements structuring
- Definition of the decision-support approach
- KPI definition
- Evaluation design
- Productization of the methodology through C3P0
Teams & organization
- Agile teams applying IREM + C3P0 in practice
- Teams in the traditional group, enabling comparison
- Delivery of backlog items measured in the evaluation
- Atlassian Jira and Rovo AI as the workflow platform
What I learned
Three things I took from this work.
AI creates more value when it is embedded in a decision framework.
The important part was not asking AI to generate better text. It was structuring the decisions that should happen before the text is generated.
Requirements are a product problem.
A requirement is not just a development artifact. It is a representation of a business problem, expected outcome and validation logic.
Measurement changes the conversation.
Connecting requirements to explicit business success metrics creates a stronger bridge between product intent and delivery.
The takeaway
Better delivery starts with better decisions.
IREM + C3P0 explores how Product Management, Requirements Engineering and Generative AI can work together to improve the quality of decisions before development begins.