Fiddler AI

1. Deploy LLM application.

2. Integrate Fiddler SDK to capture input prompts and model outputs.

3. Configure guardrails to detect toxicity using Fiddler Trust Models.

4. Receive alerts when toxic content is detected.

5. Review flagged content and fine-tune LLM or guardrails as needed.

1. Deploy agentic workflow.

2. Instrument agents with Fiddler SDK to track data flow and transformations.

3. Configure guardrails to detect PII/PHI using pattern matching and contextual analysis.

4. Receive alerts when potential PII/PHI leakage is detected.

5. Investigate incidents and implement data masking or redaction as needed.

1. Deploy RAG system.

2. Integrate Fiddler SDK to capture queries and responses.

3. Define custom evaluators to measure metrics like accuracy, relevance, and recall.

4. Visualize performance metrics in Fiddler dashboards.

5. Identify areas for improvement and fine-tune RAG system components.

1. Deploy LLM application.

2. Integrate Fiddler SDK to analyze input prompts.

3. Configure guardrails to detect prompt injection attempts using Fiddler Trust Models.

4. Receive alerts when suspicious prompts are detected.

5. Block or sanitize malicious prompts to protect LLM integrity.

1. Define AI governance policies in Fiddler.

2. Map policies to specific AI systems and components.

3. Configure guardrails to enforce policies automatically.

4. Monitor policy compliance using Fiddler dashboards.

5. Generate audit reports for regulatory compliance.