Platform
Offerings

Production AI infrastructure for the rooms that can't use frontier APIs. Deployed on customer hardware. Audited end-to-end. Compounding on customer data.

↳ Explore the platform
Solutions
Company
Get Started

Talk to us about a CASTLE deployment in your environment.

↳ Request a demo
DOCSREFERENCE
APIOPENAI-COMPATIBLE
DEPLOYMENT4 HOURS TO PROD
6 COMPONENTSONE CASTLE STACK

DOCS

Everything you need to deploy and operate CASTLE — Echo inference, Atlas database, Shield auth, Tower access control, Lens intelligence, Crown knowledge.

SCROLL
DOCUMENTATION / TECHNICAL RESOURCES

Documentation

Everything you need to deploy and operate CASTLE

QUICK START

Get running in four hours

E
Deploy Echo
INFERENCE ENGINE
Get Canadian inference running on your hardware. Multi-model routing, OpenAI-compatible API, streaming support. Production-ready in days.
A
API Reference
INTEGRATION GUIDE
OpenAI-compatible endpoints, authentication methods, streaming protocols, error handling, and rate limiting. Full spec for developers.
C
Architecture Guide
SYSTEM DESIGN
How CASTLE components work together. Component relationships, deployment topologies, scaling patterns, and infrastructure requirements.
COMPONENT DOCUMENTATION

Six integrated products. One Canadian stack.

C
Crown
KNOWLEDGE API
Vector search, semantic indexing, RAG integration. Store institutional knowledge as embeddings. Retrieve with sub-100ms latency.
A
Atlas
DATABASE SDK
PostgreSQL with row-level security and real-time subscriptions. Your data persisted locally, replicated on your terms.
S
Shield
AUTH & IDENTITY
OAuth 2.0, MFA, push authentication. Zero-trust security model. API key management and session control.
T
Tower
ACCESS CONTROL
RBAC, ABAC, policies-as-code. Fine-grained authorization from teams to entire government agencies.
L
Lens
BROWSER INTELLIGENCE
Controlled browser automation API. DOM extraction, web intelligence, research automation. Complete audit trails.
E
Echo
INFERENCE ENGINE
Chat completions, model routing, tool calling, streaming. Run 22 models from 0.6B to 32B parameters on your hardware.

Getting Started

Quick Start — Deploy Echo
$ docker pull axe/echo:latest
$ docker run -p 8090:8090 axe/echo

Listening on localhost:8090
Health check: /health
Python — Chat Completion
import httpx

resp = httpx.post(
  "https://echo.castle.local"
  "/v1/chat/completions",
  headers={
    "X-AXE-Key": key
  },
  json={
    "model": "edge-1",
    "messages": [...]
  }
)
Health Check
$ curl -s \
  https://echo.castle.local/health

{
  "status": "ok",
  "ollama": true,
  "models": 22
}
RESOURCES

Everything you need to succeed

API Reference
Complete OpenAI-compatible API specification. Endpoints, authentication, request/response formats, error codes.
Deployment Guide
Step-by-step instructions for deploying CASTLE. Hardware requirements, network setup, configuration options.
GitHub Repository
Open source CASTLE components. Contribute, fork, deploy. Complete codebase available for audit.
CASTLE.md Operator File
Comprehensive operator manual. Administration, monitoring, troubleshooting, scaling patterns.
FIPS Gap Analysis
FIPS 140-2 and FIPS 140-3 compliance assessment. Protected B architecture readiness documentation.
Community Forum
Ask questions, share implementations, connect with other CASTLE operators. Active technical support.

/ Contact · we read every inquiry

Talk to AXE.

Demos, partnerships, government RFPs, technical questions. A person reads every form. You hear from someone — not a queue.

Inquiry type

Replies within one business day · Knox audit chain records every inquiry