GeneralHow it worksPricingFor Startups
GitHubGet started
Real-time AI control plane

Control your AI app
in real time.
One line change.

One real-time control plane for every LLM call —
protect, observe, govern, and block prompt injection, jailbreaks, and PII before they reach users.

Get started free See the docs5/5on G2

One line change · Free to 20,000 calls/mo · No credit card

protect_chatbot.py
# Before — direct call to OpenAI
from openai import OpenAI
client = OpenAI(api_key="sk-...")
# After — change base_url, full protection
from openai import OpenAI
client = OpenAI(
    base_url="https://api.collieai.com/v1",
    api_key="clai_your_project_key",
)

# Same code, now protected
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{ "role": "user", "content": user_msg }]
)

One engine, three ways to run it — drop-in proxy · streaming · async jobs

SDKs:pip install collieainpm install @collieai/sdkdotnet add package CollieAi.Client
Covers OWASP Top 10 for LLM
GDPR Compliant
AICPA SOC 2 — In Progress
ISO 27001 — In Progress
Live Demo

See CollieAi catch
a real injection — right now

Type any message and watch how CollieAi classifies it in milliseconds. Try a benign request or a prompt injection attempt.

Runtime detection
Every prompt scanned before it reaches your model
Full observability
Every request logged with triggered rules, tokens, and latency
Input / output
Filter both user prompts and model responses with configurable policies
CollieAi Proxy — Live Inspector● LIVE
User prompt
Waiting for input
Click "Analyze prompt" to inspect
Liked what you saw?
20,000 requests/month · no credit card · live in 5 minutes
Policy Engine

Configure rules visually

Set up input and output filtering policies with an intuitive rule builder — no code required.

Quick start

How CollieAi keeps your AI safe

3 simple steps — integrate in minutes, see blocked attacks immediately.

1
Point your client at CollieAi
Sign up for a free API key and change your base_url. No SDK required, no architecture changes.
# Change two config values
base_url = "https://api.collieai.com/v1"
api_key  = "clai_your_project_key"

# Free: 20K tokens/month
→
2
Your code stays the same
CollieAi is a transparent proxy. Use the standard OpenAI SDK — every request is filtered by your policy rules automatically.
from openai import OpenAI

client = OpenAI(
    base_url="https://api.collieai.com/v1",
    api_key="clai_...",
)
client.chat.completions.create(...)
→
3
Threats blocked, full audit trail
Only safe, compliant responses reach your users. Dashboard shows every blocked request with triggered rules and context.
// Blocked response (OpenAI-compatible)
{
  "error": {
    "message": "Content blocked by policy",
    "type":    "content_blocked",
    "code":    400
  }
}

Works with OpenAI, Anthropic Claude, Google Gemini, DeepSeek, Azure OpenAI, AWS Bedrock and any OpenAI-compatible or self-hosted endpoint (vLLM, Ollama).

Get started free
OWASP LLM Top 10 — covered

Ship AI to production — securely.

AI guardrails and AI firewall for your LLM applications. Block prompt injection, jailbreaks and PII leaks in real time, with one line of code. Start free on our cloud, scale when you're ready.

No surprise bills. You only pay for traffic we actually process — blocked and rejected requests are free.
Free
$0
forever

Protect a real app for free — 20,000 requests a month with the full guardrail stack. Launch a pilot and watch attacks get blocked on day one.

≤ 50 ms latency
AI Guardrails
  • 20,000 API calls / month — with grace buffer
  • Prompt injection & jailbreak detection
  • PII & financial data masking
  • National ID detection (8 countries)
  • Secrets & API keys detection
  • AI content moderation (23 languages)
  • Malicious URL & hidden payload scanning
  • Bidirectional filtering (input + output)
Providers
  • OpenAI, Claude, DeepSeek, Gemini
Platform
  • OpenAI-compatible proxy
  • 1 project · Monitor & Protect modes — test rules in Monitor, then switch to Protect
  • Playground, dashboard & logs
  • Custom policies & rules
  • Community support (GitHub)
Start free — no card
No credit card. Live in 5 minutes.
★ Most popular
Growth
$49
per month

For teams shipping AI to real users. Deeper detection that catches what pattern-matching misses, plus the scale, support and audit trail production needs.

≤ 20 ms latency
250,000 API calls included · then just $0.001 per request
AI Guardrails
  • Everything in Free
  • 3-layer detectionpatterns → ML → LLM
  • Catches paraphrased & novel injections
  • Context-aware detection — understands intent, not keywords
  • Indirect prompt injection analysis
  • Multi-turn drift detection
  • Custom dictionaries
  • Explainability & full audit trail
Providers
  • Supports virtually every major LLM provider
Platform
  • Unlimited projects & API keys
  • Async jobs & webhooks (bring your own model)
  • 30-day log retention
  • 5 team members
  • Ticket support, 24h SLA
Get started
7-day money-back guarantee · No surprise bills
Enterprise
Custom
contact us

For regulated teams and high volume. Full data control, your own deployment, and the compliance controls auditors expect.

≤ 20 ms latency
AI Guardrails
  • Everything in Growth
  • Unlimited API calls
  • Real-time streaming (SSE) — output guardrails enforced token-by-token on the live stream
  • Dedicated security model tuning
  • Custom detection models
Providers
  • All providers + custom / self-hosted
Platform
  • On-premise / dedicated deployment
  • SSO (OIDC / SAML) & RBAC
  • SIEM integration & audit logs
  • GDPR data export
  • Custom log retention
  • Unlimited team members
  • Dedicated support + CSM
Talk to us
Custom volume, compliance & onboarding.
Building a startup? Get Growth free for 6 months →
FAQ
Full documentation available at docs.collieai.io
What is CollieAi?
CollieAi is a generative AI security platform — an AI firewall and LLM guardrails layer that sits between your app and any large language model. It blocks prompt injection, jailbreaks, PII leaks and unsafe output in real time through a drop-in, provider-agnostic secure proxy that works with OpenAI, Anthropic Claude, Google Gemini, DeepSeek and any other model — with no architectural change.
What is an AI firewall and how does CollieAi work?
An AI firewall inspects every prompt and every model response before it passes. CollieAi works as a drop-in proxy and LLM security gateway: it authenticates the request, applies input guardrails, forwards to your provider, filters the output, and returns an OpenAI-compatible response — bidirectional protection at under 20 ms latency.
How does CollieAi protect against prompt injection and jailbreaks?
CollieAi stops prompt injection and jailbreaks with a 3-layer detection pipeline: deterministic pattern rules, ML classifiers, and a generative LLM that reasons about intent. Because the layers overlap, paraphrased, encoded and adversarial attacks that slip past one are caught by another — with fewer false positives than a single classifier.
How does CollieAi handle indirect prompt injection in RAG?
CollieAi defends against indirect prompt injection by inspecting retrieved documents and tool output, not just the user message. Its output-direction guardrails and semantic analysis catch malicious instructions smuggled through RAG content or web data before they reach your model or your users.
Does CollieAi secure agentic AI, AI agents and MCP workflows?
Yes. CollieAi provides agentic AI security and AI agent guardrails for autonomous agents, tool calls and Model Context Protocol (MCP) pipelines. It enforces runtime policy and governance on every agent action and inspects tool output for indirect prompt injection — acting as a security gateway for production agent workflows.
Does CollieAi provide PII redaction and DLP for LLM applications?
Yes. CollieAi offers automated PII redaction and data loss prevention (DLP) for LLM applications. It detects and masks emails, phone numbers, credit cards and IBANs (checksum-validated), national IDs for 8 countries, and secrets/API keys — in both prompts and responses.
Does CollieAi offer GenAI content moderation?
Yes. CollieAi includes GenAI content moderation across 23 languages, filtering toxicity, profanity and banned or off-brand content on both input and output. Moderation runs in real time as part of the same guardrails policy, with custom dictionaries for brand-safety and competitor/codename blocking.
Is CollieAi compliant with the OWASP LLM Top 10?
Yes, on every plan. CollieAi maps directly to the OWASP LLM Top 10 — including LLM01 prompt injection, LLM02 sensitive information disclosure, LLM06 excessive agency and LLM08 insecure output handling — and produces a full audit trail for OWASP compliance reporting.
How quickly can I integrate CollieAi?
In about 5 minutes. CollieAi is OpenAI-compatible, so you change one base_url and your existing OpenAI or Claude SDK keeps working — no SDK, no rewrite. This makes it fast to integrate for small dev teams and startups, with rapid deployment to production.
Can I self-host CollieAi, and is it open-source?
Yes. CollieAi runs fully self-hosted for complete data control — your traffic never leaves your perimeter — and has an open-source core. Self-hosted deployments run with billing disabled: unlimited projects, calls and prompt sizes.
What compliance and governance does CollieAi support?
CollieAi gives you the controls and audit trail that support SOC 2, GDPR, HIPAA and PCI DSS programs: policy-based governance, SSO and RBAC, configurable data retention, GDPR data export, and SIEM integration for AI security posture management.
Will CollieAi slow down my app?
No noticeable impact. Pattern and PII rules run in under 5 ms; ML and LLM layers add 10–50 ms. On Growth, end-to-end guardrail latency stays at or below 20 ms — real-time protection your users never feel.
Which LLM providers and models does CollieAi support?
OpenAI (GPT-4o, GPT-4), Anthropic Claude, Google Gemini and DeepSeek, plus Mistral, Cohere, Llama and any OpenAI-compatible endpoint — including self-hosted models via vLLM or Ollama.
How much does CollieAi cost — is there an AI firewall for under $500/month?
Yes — CollieAi is one of the most budget-friendly options. Free covers 20,000 requests/month; Growth is $49/month with 250,000 requests, putting a full AI firewall and LLM guardrails well under the $500/month many teams budget.
What happens if I exceed my plan limit?
Nothing breaks. Free plans get a grace buffer; on Growth, extra requests are billed at $0.001 each — no throttling, no shutdown. Blocked and rejected requests never count toward your quota.

Add AI guardrails to your stack today

Start with 20,000 free requests. No credit card. Live in 5 minutes.

Start free