Frequently Asked Questions

What is TalonAI?

TalonAI is an AI security gateway that protects your AI applications from prompt injection attacks, data leakage, and other security threats. It acts as a security layer between your users and AI models, analyzing both inputs and outputs in real-time.

How does TalonAI detect threats?

TalonAI uses a combination of pattern matching, machine learning models, and semantic analysis to detect various threat types including prompt injection, jailbreak attempts, data exfiltration, and policy violations. Our detection models are continuously updated based on emerging threats.

Does TalonAI add latency to my requests?

TalonAI is optimized for low-latency processing. Typical analysis adds 50-100ms to requests. For streaming responses, we process content in parallel with minimal impact on time-to-first-token. You can also configure async mode for non-blocking analysis.

Which AI providers does TalonAI support?

TalonAI supports all major AI providers including OpenAI, Anthropic, Google Vertex AI, AWS Bedrock, Azure OpenAI, and any custom LLM deployments. Our SDKs provide native integrations for seamless setup.

Is my data stored or logged?

By default, TalonAI only stores metadata and threat signals, not your actual content. You can configure data retention policies, enable full audit logging for compliance, or disable all storage for maximum privacy. Enterprise customers can use dedicated deployments.

Can I customize detection rules?

Yes! TalonAI supports custom policies with keyword rules, regex patterns, topic filters, and sentiment analysis. You can create multiple policies for different use cases and apply them based on user groups, applications, or content types.

What happens when a threat is detected?

You control the response. Options include blocking the request, returning a safe fallback message, redacting sensitive content, or logging for review while allowing the request through. Webhooks can notify your systems in real-time.

Does TalonAI support PII detection?

Yes, TalonAI detects and can automatically redact 10+ types of PII including emails, phone numbers, SSNs, credit cards, addresses, API keys, and more. You can configure which PII types to detect and how to handle each.

Is TalonAI SOC 2 compliant?

Yes, TalonAI is SOC 2 Type II certified. We also support HIPAA compliance for healthcare customers and GDPR compliance for European data. Contact us for compliance documentation and BAA agreements.

How do I get started?

Getting started takes just a few minutes:

1. Sign up at portal.talonai.io
2. Get your API key from the dashboard
3. Install our SDK: npm install @talonai/sdk
4. Add a few lines of code to your application