AI-BA SYSTEM

AI-Assisted Feature Request & Prioritization Engine

Solution Description Document v1.7 | Bilingual | Multi-App Governance

The Requirements Crisis

Modern software development faces a critical bottleneck: the "Garbage In, Garbage Out" problem. Development teams are overwhelmed by vague requests, while users feel their feedback disappears into a "Black Hole." The AI-BA (AI Business Analyst) solves this by using Microsoft Copilot to interview users, refine requirements, and leverage a decision board for structured prioritization before any code is written.

🗣

The Noise

Unfiltered, vague requests cluttering backlogs.

🕵

The Black Hole

Users have zero visibility into request status.

📊

Feasibility Gap

Approvals made without cost/effort data.

🌐

Language Barrier

Duplicate requests across English & French.

The "Smart Intake" Funnel

The AI-BA acts as a rigorous filter. Instead of allowing every thought to become a ticket, the system uses Tier-0 Deflection (Smart Search) and AI Interviews to aggregate demand.

The chart illustrates the dramatic reduction in "Noise." From thousands of raw user interactions, only high-value, pre-approved, and feasible stories reach the Development Team in Azure DevOps.

Request Volume Filtration Pipeline

Data-Driven Decision Making (R.I.C.E.)

The Board and Apps Manager do not guess; they calculate. Using the RICE Score (Reach × Impact × Confidence ÷ Effort), the system objectively ranks features. Architects adjust "Effort" and "Confidence" during the feasibility review, dynamically altering the priority.

X-Axis: Effort (Lower is better) | Y-Axis: Impact (Higher is better) | Bubble Size: Final Priority Score

1. Reach & Impact

Defined by the Business/Board based on user value.

2. Effort & Confidence

Refined by Architects during Feasibility Review.

3. The Result

High Impact + Low Effort = Quick Wins (Top Right).

Microsoft Tech Stack Integration

Robust Technical Architecture

The solution leverages the existing Microsoft ecosystem to ensure security, scalability, and compliance.

  • Copilot Studio: The "Brain" handling intake, search, and translation.
  • Power Apps: The "Face" for the Board, Apps Manager, and Users.
  • Dataverse: The "Memory" storing Requests, Votes, and App Registry.
  • Azure DevOps: The "Engine" where actual development happens.

Process Workflow: From Idea to Release

This "Happy Path" demonstrates the governance model, including the new Multi-App selection and Feasibility Checks.

1. Intake & Deflection

User Search

Tier-0 Deflection checks Knowledge Base & Existing Requests.

App Selection

User selects "HR Portal" or "Finance App".

AI Interview

GenAI refines "Wants" into "User Stories". Auto-translation (EN/FR).

2. Governance Gates

Owner Review

Business App Owner endorses or rejects.

Board Pre-Approval

Apps Manager checks RICE score desirability.

Tech Feasibility

Architect estimates Cost & Effort. Security check.

3. Execution

Final Approval

Board gives the "Go" signal.

DevOps Sync

Automated creation of User Story in Azure DevOps.

Release & Notify

Status syncs back to Power App. User notified in Teams.

Operational Efficiency Impact

Transforming the Process

Moving from a manual email/ticket based system to the AI-BA dramatically improves key operational metrics.

Requirement Clarity

AI interviews ensure developers get complete User Stories, not vague one-liners.

Cost Control

Feasibility checks prevents expensive features from being approved blindly.

Bilingual Compliance

Automated translation ensures 100% adherence to OLA standards.