Architecture
State-of-the-art large language models
ClaudIA uses a combination of GPT 3.5 and GPT 4 models to provide the best possible customer support resolution. These models are designed to understand and generate human-like text, and are trained on a diverse range of internet text. For enterprise customers, we offer custom models hosted through Azure.
Architecture
State-of-the-art large language models
ClaudIA uses a combination of GPT 3.5 and GPT 4 models to provide the best possible customer support resolution. These models are designed to understand and generate human-like text, and are trained on a diverse range of internet text. For enterprise customers, we offer custom models hosted through Azure.
Search across internal company customer support docs
Answering customer support questions
Generating personalized messages
Training
We collect data from users when they type information into our AI chatbot. Examples of customer support questions and data we collect and train our AI models on:
Product questions
Pricing questions
Sales questions
Personally identifiable information (PII)
Sensitive or protected data (health, political views, etc.)
Protected characteristics (race, gender, etc.)
Biometric data
Inference
During inference, Cloud humans utilizes the following data to generate recommendations and insights, ensuring privacy and compliance:
User search queries
Profile data (without PII)
Input prompts for generating personalized messages
Personally identifiable information (PII)
Sensitive personal data
AI Data Governance
Source of data for AI models
Compliance with data sharing for LLM foundation models
Exclusion of customer data from model training
PII redaction compliance
AI Model Testing & Monitoring
Assurance of model effectiveness and validity
Continuous model performance monitoring
Event logs collection for models
Enhancement of models' explainability
AI Cybersecurity
Implementation of malicious prompts protection mechanisms
Safeguard against training data poisoning
Enforcement of unique access control and authentication
Oversight of AI system operations
Third-Party AI tools and audits
Assurance of model effectiveness and validity
Continuous model performance monitoring
Event logs collection for models
Enhancement of models' explainability
