Jollibee Group - Brand Operating System (BOS)
Use Cases & Technical Overview
Prepared by Contino | November 25, 2025
Executive Summary
The Brand Operating System (BOS) is an AI-powered platform designed to eliminate brand inconsistency across franchisees of brands of Jollibee Group. By replacing PDF-based standards with an intelligent chat interface and asset validation system, BOS will reduce time-to-answer from hours to minutes, ensure brand compliance at the point of creation, and provide actionable insights into franchisee behavior.
Core Use Cases
Use Case 1: Brand Standards Q&A
Scenario: A franchisee in Malaysia wants to run a Valentine's Day promotion but doesn't know the approved color palette, typography rules, or promotional layout guidelines.
Current State:
- Hunt through 300+ pages of PDF guides, including brand guidelines
- Call Manila customer service and wait for response
- Hire a junior local designer ($25-50/hr) with no brand context
- High probability of non-compliant output
BOS Solution:
- Franchisee asks: "What are the guidelines for Valentine's Day promotional materials?"
- System returns: Approved colors, typography, layout grids, photography style, logo placement rules, and templates for Valentine's Day
- Response time: Under 30 seconds
- Confidence level: Verified against official brand standards
Technical Approach:
- RAG (Retrieval Augmented Generation) pipeline ingests all brand documentation
- Vector embeddings enable semantic search across standards
- Response synthesis ensures complete, contextual answers
- Source attribution shows exactly which guideline sections apply
Use Case 2: Asset Compliance Validation
Scenario: A franchisee's designer creates promotional artwork and needs verification before printing/publishing.
Current State:
- Submit to Manila for manual review (24-72 hour turnaround)
- Inconsistent feedback depending on reviewer
- Often discover issues after materials are already printed
BOS Solution:
- Designer uploads artwork to BOS portal
- System analyzes against brand guidelines:
- Logo placement and sizing ✓
- Color accuracy (hex/RGB matching) ✓
- Typography usage ✓
- Photography style compliance ✓
- Clear space and margin rules ✓
- (Other brand guideline rules checking) ✓
- Returns pass/fail with specific feedback:
- "Logo positioned 15px too close to edge. Minimum clear space: 40px. See Section 3.2 of Brand Guidelines."
Technical Approach:
- Multi-agent architecture with specialized validators:
- Logo Compliance Agent: Analyzes logo usage, sizing, placement, clear space
- Color Compliance Agent: Validates color palette adherence via color space analysis
- Typography Agent: Checks font usage, hierarchy, and styling rules
- Composition Agent: Evaluates layout against approved grid systems
- Photography Agent: Assesses image style, quality, and brand fit
- Agents run in parallel for speed, results aggregated into unified report card
- Each violation links to specific guideline section and remediation steps
Use Case 3: Franchisee Onboarding
Scenario: A new franchisee needs to learn brand requirements and how to use BOS.
Current State:
- Receive PDF documents with no verification of understanding
- No standardized onboarding across regions
BOS Solution:
- Short onboarding course covering brand basics and how to use BOS
- Simple quiz to verify understanding
- Completion required before accessing full system
Technical Approach:
- Web-based course module with localization support
- Quiz with pass/fail tracking
Technical Architecture Overview
Four Core Components
┌─────────────────────────────────────────────────────────────────┐
│ BRAND OPERATING SYSTEM │
├─────────────────────────────────────────────────────────────────┤
│ │
│ ┌──────────────┐ ┌──────────────┐ ┌──────────────┐ │
│ │ System │ │ Tools │ │ RAG │ │
│ │ Prompts │ │ │ │ Pipeline │ │
│ └──────────────┘ └──────────────┘ └──────────────┘ │
│ │ │ │ │
│ └────────────────┼─────────────────┘ │
│ │ │
│ ┌──────▼──────┐ │
│ │ Evaluation │ │
│ │ & │ │
│ │ Observability│ │
│ └─────────────┘ │
│ │
└─────────────────────────────────────────────────────────────────┘
1. System Prompts
- Define agent identity, capabilities, and guardrails
- Specialized prompts per use case (Q&A, validation, onboarding)
2. Tools
- Validation engines: Image analysis, color matching, layout checking
- Escalation: Human support handoff when needed
- LATER: Aprimo API: Digital asset retrieval
3. RAG Pipeline
- Vector database storing all brand documentation
- Semantic search for contextual answer retrieval
- Metadata filtering by brand, region, campaign
- Query alignment to maximize retrieval accuracy
4. Evaluation & Observability
- Response quality tracking (user feedback, human review)
- Cost and latency monitoring
- Self-improvement cycles based on failed queries
- Comprehensive audit trail for compliance
Multi-Agent Validation Architecture
For asset compliance checking, we employ specialized agents rather than a single general-purpose validator:
┌─────────────────┐
│ Uploaded Asset │
└────────┬────────┘
│
┌──────────────┼──────────────┐
│ │ │
▼ ▼ ▼
┌──────────┐ ┌──────────┐ ┌──────────┐
│ Logo │ │ Color │ │Typography│
│ Agent │ │ Agent │ │ Agent │
└────┬─────┘ └────┬─────┘ └────┬─────┘
│ │ │
│ ┌──────────┐ │
│ │Composition│ │
│ │ Agent │ │
│ └────┬─────┘ │
│ │ │
└──────────────┼──────────────┘
│
┌────────▼────────┐
│ Aggregator │
│ Report Card │
└─────────────────┘
Why Multi-Agent?
- Each agent specializes in one domain → higher accuracy
- Agents run in parallel → faster response times
- Easier to maintain and update individual rule sets
- Clear accountability for each compliance dimension
Self-Improvement Cycle
A key differentiator of BOS is its ability to get smarter over time:
User Question
│
▼
Agent Attempts Answer
│
├── Success → Log & Learn Patterns
│
└── Cannot Answer → Escalate to Human
│
▼
Human Provides Answer
│
▼
Extract Q&A Pair
│
▼
Add to Knowledge Base
│
▼
Future Questions Answered Automatically
Every escalation becomes training data. Every failed query identifies a documentation gap. The system continuously improves without manual intervention.
Success Metrics (Phase 1)
| Metric | Target | Measurement |
|---|---|---|
| User Adoption | 75% of pilot users active | Weekly login tracking |
| Answer Accuracy | 90% correct on first response | Human validation sampling |
| Time to Answer | Under 5 minutes (down from hours) | Query-to-resolution timing |
| Support Call Reduction | 50% decrease | Manila call center metrics |
| Compliance Rate | 80% first-submission pass | Validation pass/fail tracking |
Enterprise Requirements
Security & Compliance:
- GDPR compliant data handling
- Role-based access control
- Audit logging for all actions
- Data encryption at rest and in transit
Scalability:
- Designed for 5,000+ concurrent users
- Regional deployment options for latency optimization
- Load-tested for campaign peak periods
Integration:
- LATER: Aprimo REST API connection
- SSO/SAML for enterprise authentication
- Webhook support for external system notifications
Implementation Phases
Phase 1: Pilot (Chow King)
- Brand standards documentation ingestion
- Chat interface deployment
- Asset validation MVP
- Testing with 10-15 real franchisees
Phase 2: Refinement
- Optimize based on real usage data
- Expand validation capabilities
- A Primo deep integration
- Self-improvement system activation
Phase 3: Scale (Jollibee + Additional Brands)
- Multi-brand architecture
- Full enterprise rollout
- Advanced analytics dashboard
- Continuous improvement operations
Why This Approach Works
-
Specialized over General: Multi-agent architecture ensures each compliance dimension is handled by an expert system, not a general-purpose AI that might miss nuances.
-
Self-Improving: Unlike static rule engines, BOS learns from every interaction. New questions become new answers. Failed validations reveal documentation gaps.
-
Human-in-the-Loop: Critical edge cases escalate to humans. Their resolutions feed back into the system. Brand judgment is preserved, not replaced.
-
Observable & Measurable: Full tracing of every query, every validation, every cost. You'll know exactly how the system performs and where to improve.
-
Built for Scale: Architecture designed for enterprise deployment across 60+ countries with thousands of concurrent users.
Contino | Brand Strategy & Technology