Skip to content

PLAN_AI_BACKLOG_IMPROVEMENT.md

Latest commit

 

History

History
95 lines (83 loc) · 2.81 KB

PLAN_AI_BACKLOG_IMPROVEMENT.md

File metadata and controls

95 lines (83 loc) · 2.81 KB

BacklogAI AI Content Improvement Plan

Objective

Improve AI-generated backlog content so users only provide context and objective. The system will perform market research, competitive analysis, refine the need, generate INVEST-compliant JIRA stories aligned to strategic goals, and sync to JIRA.

Scope

  • New v2 generation flow: context + objective input
  • Market research via SerpAPI
  • Multi-stage AI pipeline (understand -> research -> generate -> quality revise)
  • Expanded story output schema suitable for JIRA sync
  • UI updates for new flow
  • Tests for quality and research fallbacks

Implementation Plan

1) API Contract: Context + Objective

  • Add new v2 request model with:
    • context
    • objective
    • target_user
    • market_segment
    • constraints
    • success_metrics
    • competitors_optional
  • Keep v1 models for backward compatibility

2) Market Research Service (SerpAPI)

  • New service module to query SerpAPI using objective + market segment + competitors
  • Summarize findings:
    • trends
    • competitor_features
    • differentiators
    • risks
    • sources
  • Cache by request hash to reduce repeated calls
  • Env var: SERPAPI_API_KEY

3) Multi-Stage Generation Pipeline

  • Stage A: Need understanding (scope, non-goals, assumptions, user pain, business value)
  • Stage B: Inject research brief
  • Stage C: Draft story + AC + sub-tasks + NFRs + metrics + rollout + dependencies
  • Stage D: Quality gate; revise if warnings exceed threshold

4) Expanded Output Schema (JIRA-ready)

  • summary
  • description
  • acceptance_criteria
  • sub_tasks
  • dependencies
  • risks
  • metrics
  • rollout_plan
  • non_functional_reqs
  • validation_warnings
  • pillar_scores, priority_score, moscow_priority

5) JIRA Sync Mapping

  • Build a description template:
    • Background
    • Objective
    • Research Summary
    • User Story
    • Acceptance Criteria
    • Non-functional Requirements
    • Risks
    • Metrics
    • Rollout Plan
  • Leave Epic Link, Components, Labels blank unless provided

6) UI Updates

  • Replace input form with context/objective fields
  • Show research summary + story preview + warnings
  • Add edit-before-sync step

7) Tests and Evaluation

  • Tests for:
    • SerpAPI fallback behavior
    • v2 schema validation
    • INVEST quality thresholds
  • Small evaluation set (10-20 cases)

Files to Change

  • backend/app/schemas.py
  • backend/app/services/market_research_service.py (new)
  • backend/app/services/story_engine.py
  • backend/app/services/quality_engine.py
  • backend/app/main.py
  • backend/app/services/jira_service.py
  • composeApp/src/commonMain/kotlin/com/backlogai/ui/screens/InputScreen.kt
  • composeApp/src/commonMain/kotlin/com/backlogai/ui/screens/ResultScreen.kt
  • shared/src/commonMain/kotlin/com/backlogai/shared/network/BacklogApi.kt

Notes

  • Use existing model configuration (no model changes)
  • SerpAPI is the recommended web search API