Foundation QC Analysis
Content to Analyze
Paste your content below or upload a file to begin the analysis.
Rules Applied
Default rules from Roche Global Copy Style Guide Book are shown. Upload a JSON file with rules under "what_to_do" and "what_to_prohibit" to override.
Gemini 2.5 Pro Report (Precise Teacher Model)
Precise Gemini analysis report will appear here.
Analyzing with Gemini (Precise)...
Gemini 2.5 Flash Report (Creative Student Model)
Creative Gemini analysis report will appear here.
Analyzing with Gemini (Creative)...
Analysis Comparison
Common Issues
Precise Model Only
Creative Model Only
Human Verification
Select issues to send back to the Teacher Model for verification.
Global Compliance Codes
- IFPMA
- EFPIA
- PHRMA
- PAAB
- Roche Copy style Guide Book
Upload & Extract Local Rules
Upload rule documents (e.g., .txt, .md, .pdf, images) to automatically extract rules using AI.
No rules extracted yet. Upload a document to begin.
Claim Checker
1. Approved Claims Database
Upload PDF or JSON containing approved claims to build the rule set.
Extracted Claims Context
Claims Loaded
2. Marketing Asset
Upload an image or PDF to check against approved claims.
Extracted Text Blocks (Editable)
Text extracted from asset will appear here as editable blocks.
Analysis Results
Comparing marketing text against approved claims...
Final Verification Report
Personalization Checker
1. Target Persona (Ground Truth)
Select a persona or upload a profile definition.
2. Content to Check
Upload the email template or content JSON.
Analyzing content fit against persona with Reasoning LLM...
Fit Assessment
| Persona | Fit Score | Winning Element | Key Barrier |
|---|
Adaptation Analysis
Personalization Strategy
Subject Line Feedback
Tone Assessment
Content Completeness & ISI
CTA Effectiveness
Analyzed Content JSON
Architecture Framework
Waterfall Mechanism for Rule Filtering
A waterfall approach (IP → Country → Usecase → Topics) is applied to systematically reduce and track the number of rules that need to be checked, ensuring a more efficient and focused analysis.
Student-Teacher Model for Rule Verification
Utilizes a dual-model system. The 'Precise Teacher Model' guides the analysis based on established rule knowledge, while the 'Creative Student Model' verifies the rules, suggests conflicts, and proposes new ideas to challenge the rule set.
Human-in-the-Loop (HITL) Mechanism for Conflict Resolution
When models disagree, an expert reviews unique findings. The user selects ambiguous rules and sends them back to the 'Teacher Model' for a final, justified verification, solidifying the knowledge base have more trackable results.