Skip to Content

Implementation Roadmap

Last modified by Robert Schaub on 2025/12/24 21:53

Implementation Roadmap

FactHarbor development follows a phased approach from Proof of Concept to production launch.

Development Phases

Information

Detailed Requirement Mapping: See Requirements Roadmap Matrix for complete phase-to-requirement mapping, implementation levels, and User Needs fulfillment by phase.

Philosophy

Validate AI automation quality BEFORE building production features

Each phase builds on proven capabilities from the previous phase. We never skip ahead - if AI quality is insufficient, we improve prompts and algorithms rather than adding manual workarounds.

Development Phases Overview

POC1: Core Workflow with Quality Gates

Build:

  • Automated claim analysis
  • Confidence scoring
  • Source evaluation
  • Browse/search interface
  • User flagging system
  • LLM Provider Abstraction (NFR-14): Provider-agnostic AI access
  • POC1: Anthropic Claude API primary (Haiku Stage 1, Sonnet Stages 2 & 3)
  • Architecture: Abstract LLMProvider interface, environment config, basic failover
  • Future: OpenAI/Google secondary providers, admin API, cost comparison (POC2/Beta)

Goal: Prove AI quality before adding user editing

User Needs fulfilled in POC1: UN-1, UN-2, UN-3, UN-4, UN-5, UN-6, UN-7, UN-8, UN-9, UN-12

POC2/Beta 0: User Contributions

Add only if needed:

  • Simple editing (Wikipedia-style)
  • Reputation system
  • Basic moderation
  • In-article claim highlighting (FR13)

Additional User Needs fulfilled: UN-13, UN-17

V1.0+: Continuous Improvement

  • Continuous quality improvement
  • Feature additions based on real usage
  • Scale infrastructure

Additional User Needs fulfilled: UN-14 (API access)

Deferred Beyond V1.0:

  • Federation (until multiple successful instances exist)
  • Complex contribution workflows (focus on automation)
  • Extensive role hierarchy (keep simple)

Success Metrics Framework

Note: Each phase defines specific success criteria in detail. This section provides the overall metrics framework.

System Quality:

  • Error rate by category (target: continuous improvement)
  • Average confidence score (target: increase)
  • Source quality distribution (target: more high-quality)
  • Contradiction detection rate (target: increase)

Efficiency:

  • Claims processed per hour (target: increase)
  • Human hours per claim (target: decrease)
  • Automation coverage (target: >90%)
  • Re-work rate (target: <5%)

User Satisfaction:

  • User flag rate (issues found)
  • Correction acceptance rate (flags valid)
  • Return user rate
  • Trust indicators (surveys)

User Needs Metrics:

  • UN-1: % users who understand trust scores
  • UN-4: Time to verify social media claim
  • UN-7: % users who access evidence details
  • UN-8: % users who view multiple scenarios
  • UN-15: % users who check evolution timeline
  • UN-17: % users who enable in-article highlighting; avg. time spent on highlighted vs. non-highlighted articles

Status

Current Phase: POC1 Specification Complete
Next Action: Begin POC1 Implementation
Version: V0.9.70