Skip to content

Aegis Preflight

Control sensitive data before it leaves.

Check data before it reaches AI tools, vendors, SaaS, or APIs. Instant yes/no. No compliance tickets. No security reviews. Just build.

  • Built for LLM pipelines, RAG systems, and AI agents

  • HIPAA, GDPR, SOC2 guardrails out of the box

  • Get started in 5 minutes

Get Started View SDK Docs

Why Aegis?

  • Ship AI Features Faster

    No more waiting for compliance review. Engineers get instant yes/no before sending data to LLMs.

  • Stay Compliant

    Automated guardrails for HIPAA, GDPR, SOC2. Policies enforced at the API layer, not in meetings.

  • Self-Service for Engineers

    Engineers decide what can go to AI tools, with confidence. No tickets, no bottlenecks.

  • Full Visibility

    Every LLM call is logged. Compliance team gets audit trails without being gatekeepers.

Quick Start

from aegis_sdk import Aegis

# Initialize (no license required for basic use)
aegis = Aegis()

# Check content before sending to AI
result = aegis.process(
    text="Customer SSN: 123-45-6789",
    destination="AI_TOOL"
)

if result.is_blocked:
    print(f"Blocked: {result.summary}")
else:
    # Safe to proceed with masked content
    send_to_ai(result.masked_content)

from aegis_sdk import AegisOpenAI

# Drop-in replacement for OpenAI client
client = AegisOpenAI(api_key="sk-...")

# PII automatically masked before sending
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{
        "role": "user",
        "content": "Customer email: [email protected]"
    }]
)
# OpenAI only sees: "Customer email: j***@example.com"

# Install
pip install aegis-sdk

# Scan for PII
aegis scan "Customer email: [email protected]"

# Mask PII
aegis mask "My SSN is 123-45-6789"

# Process with policy
aegis process "Customer data" -d AI_TOOL

Common Use Cases

Scenario Before Aegis With Aegis
Send customer ticket to ChatGPT Wait 2 days for security approval Instant check, auto-mask PII
Export data to vendor analytics Manual review of every file Policy-based auto-approval
Build RAG pipeline with internal docs Block entire project for compliance Scan at ingestion, ship same day

Built for AI Engineering Teams

  • AI/ML Engineers

    Building LLM-powered features? Check data before it hits OpenAI, Anthropic, or your fine-tuned models.

  • Data Platform Teams

    Building RAG pipelines or data exports? Ensure PII doesn't leak into vector stores or external systems.

  • Backend Engineers

    Integrating with AI tools or third-party APIs? Get instant approval without waiting for security review.

How It Works

sequenceDiagram
    participant App as Your App
    participant SDK as Aegis SDK
    participant AI as AI Tool

    App->>SDK: aegis.process(text, destination)
    SDK->>SDK: Detect PII patterns
    SDK->>SDK: Apply policy rules

    alt Content is safe
        SDK->>App: ALLOWED
        App->>AI: Send original content
    else Contains PII (maskable)
        SDK->>App: ALLOWED_WITH_MASKING
        App->>AI: Send masked content
    else Contains blocked data
        SDK->>App: BLOCKED
        App->>App: Handle appropriately
    end

Next Steps

Support