SDECB AI Project Roadmap
Overview
This roadmap documents future improvements and work items identified during the comprehensive project audit (January 2026). Items are tracked against the Implementation Phase (Jan-Mar 2026, $80K budget).
Current Status: Portal v2 deployed, KB operational, training scheduled
Next Milestone: Training session (January 19, 2026)
What’s Already Done
Completed Infrastructure
Completed Documentation
Phase 1 Preparation (Before January Training)
1. Phase 1 Master Plan Specification - COMPLETED
Priority: HIGH
Status: DONE
Owner: Kris Krug
Completed:
Deliverable: /docs/implementation/PHASE2_MASTER_PLAN.md
2. Training Curriculum Development
Priority: HIGH
Timeline: Before January 2026
Owner: Kris Krug (Jan 19 delivery) + P.J. Muhigi (Feb–Mar curriculum & delivery)
Tasks:
- Session 1: AI Literacy Basics (90 min, Zoom, bilingual)
- What is AI? (non-technical explanations)
- SDECB’s specific use cases
- Addressing concerns (job displacement, privacy)
- Hands-on: Custom GPT exploration
- Session 2: AI Policy Co-Creation (90 min)
- Review existing AI policies (examples from similar organizations)
- Identify SDECB-specific requirements
- Draft policy sections collaboratively
- Address Mylene’s “public AI use” concerns
- Bilingual considerations
- Session 3: Hands-On Tools Training (90 min)
- Knowledge base navigation
- Internal chatbot usage
- Best practices for prompting
- Troubleshooting common issues
Materials Needed:
- PowerPoint/Keynote slides (bilingual)
- Demo videos (pre-recorded fallbacks)
- Hands-on exercises
- Quick reference guides (one-pagers)
- FAQ document (bilingual)
Deliverable: /docs/training/PHASE_1_TRAINING_CURRICULUM.md + training materials
3. AI Policy Framework
Priority: HIGH
Timeline: Before January 2026
Owner: Kris Krug (template), Mylene + team (co-creation)
Tasks:
- Research AI policy best practices
- Review Canadian AI governance frameworks
- Study Francophone AI ethics guidelines
- Review bilingual service delivery requirements
Policy Sections:
- Purpose and scope
- Acceptable use guidelines
- Data privacy and security
- Client consent requirements
- Bilingual service standards
- Human oversight requirements
- Incident reporting procedures
- Review and update process
Key Concerns to Address:
- Public AI use concerns (Mylene’s observation)
- Client data handling
- Third-party AI service usage
- Staff training requirements
- Accountability and governance
Deliverable: /docs/governance/AI_POLICY_FRAMEWORK.md
3b. External Chatbot Prototype Planning
Priority: HIGH
Timeline: February-March 2026
Owner: Lionel Ringenbach (technical spec), Kris Krug (governance & rollout)
Context:
Per the Jan 8, 2026 meeting with Mylène, ministry funding language includes an external/public chatbot. Phase 2 delivers the prototype plan; Phase 3 delivers the prototype build (if the stop/go gate is approved).
Phase 2 Deliverables:
- Technical specification and architecture for external chatbot
- Governance integration (how AI Policy applies to public-facing use)
- Staged rollout approach with risk mitigation
- Friendly-user pilot cohort identification (10-20 invited clients/partners)
- Scope boundaries and escalation paths
Phase 3 Dependency:
The external chatbot prototype build only proceeds if the stop/go gate is approved at end of March, based on:
- Internal assistant pilot metrics meeting thresholds
- Staff confidence demonstrably high
- AI Policy adopted and operational
Reference Docs:
/docs/implementation/WORKPLAN_NOTION_ALIGNED_SDECB_AI_PHASE2_JAN-MAR_2026.md (Section 2.2)
/docs/implementation/CLIENT-PLAN_SDECB_AI_PHASE1_JAN-MAR_2026.md (Workstream 2)
/docs/meetings/2026-01-08_Mylene_Letellier_Meeting_Phase2_Implementation.md
Deliverable: External Chatbot Prototype Plan (spec + governance + rollout approach)
4. Success Metrics Definition
Priority: MEDIUM
Timeline: Before Phase 1 kickoff
Owner: Kris Krug + Mylene
Quantitative Metrics:
- Hours saved per week (target: 36-38 hours in Phase 1)
- User adoption rate (target: 80% within 3 months)
- API response time (target: <2 seconds)
- Accuracy rate (target: 95%+)
- Cost savings (target ROI: 400%+)
Qualitative Metrics:
- Employee wellbeing (Felixe’s framework)
- User satisfaction scores
- Bilingual quality assessment
- Staff confidence with AI tools
Deliverable: /docs/PHASE_1_SUCCESS_METRICS.md
Code Quality Improvements
5. Error Handling & Logging Infrastructure
Priority: MEDIUM
Timeline: Phase 1 development
Effort: 2-3 days
Current State:
- Minimal error handling (only basic print statements)
- No try-catch blocks in most functions
- No centralized logging system
Tasks:
- Add try/catch blocks to all API calls
- Create fallback mechanisms for API failures
- Implement user-friendly error messages (bilingual)
- Set up Python
logging module with configuration
- Create JavaScript console logging with levels
- Define log levels: DEBUG, INFO, WARNING, ERROR, CRITICAL
- Implement log rotation policy
- Create error monitoring system
Deliverable: /docs/technical/ERROR_HANDLING.md
6. Automated Testing
Priority: MEDIUM
Timeline: Phase 1 development
Effort: 3-5 days
Current State:
- Only manual testing checklists exist
- No automated test suite
test_demos.py provides basic manual validation
Tasks:
- Set up pytest framework (already in requirements.txt)
- Create test directory structure (
/tests/unit/, /tests/integration/)
- Write unit tests for Python functions
- Create integration tests for API calls
- Build end-to-end tests for critical user flows
- Create bilingual content testing
- Set up test fixtures with anonymized data
- Mock API responses (avoid testing costs)
- Target 60%+ code coverage
Deliverable: /tests/ directory with automated test suite
7. Code Refactoring & Component Library - PARTIALLY COMPLETED
Priority: LOW (defer to post-Phase 1)
Status: Portal v2 has component library; legacy demos remain
Completed:
Remaining (Low Priority):
Deliverable: /portal/v2/components/ - DONE
8. Archive Cleanup & Documentation - COMPLETED
Priority: LOW
Status: DONE
Completed:
Deliverable: /portal/archive/README.md - DONE
Documentation Organization
9. Master Documentation Index
Priority: LOW
Timeline: Phase 1
Effort: 1-2 hours
Tasks:
- Update
/docs/README.md with comprehensive index
- Categorize 60+ markdown files by audience:
- Executive/Leadership
- Technical Team
- End Users
- Training Materials
- Add “Last Updated” dates to all docs
- Archive old/superseded documentation
Deliverable: Updated /docs/README.md
10. Getting Started Guide
Priority: LOW
Timeline: Phase 1
Effort: 2-3 hours
Tasks:
- Create onboarding guide for new team members
- “Where to find what” navigation
- Who to ask for help
- Quick wins for getting productive
Deliverable: /docs/GETTING_STARTED.md
Notion Workspace Organization
11. Notion Cleanup
Priority: MEDIUM
Timeline: Before January training
Effort: 3-4 hours
Current Issue: User reported “structure/organization is messy”
Recommended Structure:
SDECB AI Project/
├── 📋 Project Dashboard
│ ├── Budget Tracker ($80K)
│ ├── Timeline (Jan-Mar 2026)
│ ├── Key Contacts
│ └── Phase Overview
│
├── 🎯 Phase 1: Implementation (Jan-Mar 2026)
│ ├── Training Sessions (January)
│ ├── Knowledge Base Development (February)
│ ├── Chatbot Deployment (March)
│ └── Success Metrics Dashboard
│
├── 📚 Discovery Phase (Archive - Sept 2025)
│ ├── Interview Summaries (12 staff)
│ ├── Presentation Materials
│ ├── Demo Documentation
│ └── Meeting Notes
│
├── 📖 Documentation
│ ├── AI Policy (In Progress)
│ ├── Training Materials
│ ├── Technical Guides
│ └── User Documentation
│
├── 👥 Team & Stakeholders
│ ├── Staff Directory
│ ├── Champions & Skeptics
│ ├── Volunteer Pilot Group (7 people)
│ └── External Partners (Philippe)
│
└── 📊 Reporting
├── Weekly Progress Updates
├── Budget & Expenses
├── ROI Tracking
└── Board Reports
Tasks:
- Audit current Notion pages
- Move Discovery phase content to archive section
- Create templates for recurring updates
- Link to GitHub repository documentation
- Ensure bilingual content where needed
- Set up appropriate permissions
Deliverable: Organized Notion workspace
Production Deployment Readiness
12. API Proxy Layer (Server-Side Key Management)
Priority: HIGH (before production deployment)
Timeline: Phase 1 or before public launch
Effort: 3-5 days
Current Issue:
- API keys currently hardcoded in client-side JavaScript (HTML demos)
- Security risk - keys visible to anyone viewing source
Solution:
- Create backend API proxy service
- Client calls proxy, proxy calls AI services with keys
- Keys stored server-side only (environment variables)
- Implement request authentication
- Add rate limiting per user/IP
Technologies:
- Node.js/Express or Python/Flask backend
- Environment variable management
- CORS configuration
Deliverable: API proxy service + deployment documentation
13. Rate Limiting & Monitoring
Priority: MEDIUM
Timeline: Phase 1
Effort: 2-3 days
Tasks:
- Implement rate limiting (prevent API abuse)
- Set up usage monitoring
- Create alerting for unusual patterns
- Track costs per API (ElevenLabs, Mistral, Claude)
- Log all requests for audit purposes
Deliverable: Monitoring dashboard + alerting system
14. Data Privacy Audit
Priority: HIGH (before public deployment)
Timeline: Phase 1
Effort: 3-5 days
Tasks:
- Conduct comprehensive data privacy audit
- Review GDPR compliance (if applicable)
- Review PIPEDA compliance (Canadian privacy law)
- Document what data is collected and stored
- Create data retention policy
- Develop data deletion procedures
- Create bilingual consent forms
- Draft privacy policy
Deliverable: Privacy audit report + compliance documentation
Priority Summary
Completed
Critical (Do Before Production)
- API Proxy Layer (server-side key management)
- Data Privacy Audit
- AI Policy Framework
High Priority (Phase 1)
- Training Curriculum
- Success Metrics Definition
- Error Handling & Logging
Medium Priority (Phase 1 or Phase 2)
- Automated Testing
- Notion Workspace Organization
- Rate Limiting & Monitoring
Low Priority (Nice to Have)
- Master Documentation Index
- Getting Started Guide
Budget Considerations
Phase 1 Budget: Ministry-funded (secured December 2025)
Estimated Costs:
- Training delivery: Internal (KK) + external French trainer
- Knowledge Base development: $3K-5K
- Internal Chatbot: $5K-7K
- AI Policy co-creation: Internal (training time)
- API proxy development: $2K-3K
- Testing infrastructure: $1K-2K
- Privacy audit: $1K-2K (or external consultant $3K-5K)
Total Phase 1 Technical: ~$12K-21K (leaves budget for ongoing operations, API costs, and contingency)
Timeline Overview
January 2026:
├── Training Sessions (3 x 90 min)
├── AI Policy Co-Creation
└── Begin Knowledge Base Development
February 2026:
├── Knowledge Base Completion
├── Begin Chatbot Development
└── Error Handling & Logging Implementation
March 2026:
├── Internal Chatbot Deployment
├── API Proxy Implementation (if deploying publicly)
├── Testing & Quality Assurance
└── Success Metrics Measurement
Post-Phase 1:
├── Code Refactoring
├── Documentation Organization
└── Continuous Improvement
Risk Mitigation
Key Risks:
- Training Delay - Mitigate: Schedule sessions in early January, prepare materials in advance
- French Language Quality - Mitigate: External French trainer for delivery, native speaker review
- Technical Complexity - Mitigate: Start with MVP, iterate based on feedback
- Scope Creep - Mitigate: Clear Phase 1 specifications, prioritized roadmap
Questions for Future Decision-Making
- Archive Deletion: Keep or delete 18 archived demos?
- Current Recommendation: Keep with documentation (historical value)
- Testing Coverage: What’s the target coverage percentage?
- Current Recommendation: 60% for Phase 1 (pragmatic, achievable)
- Refactoring Timing: Refactor now or after Phase 1?
- Current Recommendation: Plan now, execute after Phase 1 (risk management)
- Source of Truth: GitHub or Notion for documentation?
- Current Recommendation: GitHub for technical docs, Notion for project management
Last Updated: January 11, 2026
Next Review: After Phase 1 completion (March 2026)
Owner: Kris Krug (The Upgrade AI)