The Zero-Knowledge
AI Infrastructure Platform.
Enterprise-grade AI chat with Theeban, 1-click Hugging Face deployments, and serverless API endpoints. Secure, scalable, and proudly engineered by Saudi talent for the unique needs of the Saudi market.
Meet Theeban (ذيبان)
The Zero-Knowledge AI Assistant built exclusively for Saudi enterprise and government. Cryptographically isolated, multi-model capable, and hosted 100% locally.
Absolute Data Sovereignty
If we can leak it, it’s not secure. Your data never leaves KSA and cannot be accessed by our infrastructure team.
Model Agnostic
Switch seamlessly between fast, pro, deep-research depending on your specific task.
Ready for Compliance
Built from the ground up to meet strict Saudi healthcare, financial, and government data regulations.
Bring Your Own Model. Deploy in Seconds.
Skip the bare-metal setup. Paste any Hugging Face URL (even a private repo) and instantly spin up a dedicated VRAM allocation powered by vLLM. No Linux, no drivers, no waiting.
Dedicated VRAM Allocation
At just 0.20 SAR/GB/hr, that memory is 100% yours. Guarantee zero queue times and maximum throughput for your specific model.
True Active Billing
You only pay for the exact seconds your model is in the "Active" state. You never pay for weight downloads, cold starts, or teardown. Hit pause, and billing stops instantly.
A drop-in replacement for your current AI stack.
Our API is fully compatible with the OpenAI SDK. Access Llama 3, Mistral, and top open-weights models instantly without rewriting your application logic.
Pay Per Token
No hourly rates. No idle costs. Pay strictly for the input and output tokens you process at ultra-competitive market rates.
Blazing Fast Inference
Powered by our enterprise vLLM backend, delivering massive concurrency and minimal Time-to-First-Token (TTFT).
from openai import OpenAI
# 1. Point the client to DCP
client = OpenAI(
api_key="dcp_your_api_key_here",
base_url="https://api.dcp.sa/v1"
)
# 2. Query any available model
response = client.chat.completions.create(
model="dcp/fast",
messages=[
{"role": "system", "content": "You are an AI."},
{"role": "user", "content": "What's the meaning of life?"}
]
)
print(response.choices[0].message.content)Zero-Knowledge by Design.
Built for Saudi Arabia’s strictest compliance standards. If we can leak your data with root access, it isn’t secure—so we engineered the platform so we can’t.
100% Data Sovereignty
Every query, cache, and fine-tuned model weight is hosted exclusively on bare-metal servers located physically within the Kingdom of Saudi Arabia.
Zero-Knowledge Storage
Frictionless, transparent encryption. Your chat logs and API payloads are cryptographically secured at rest. Our infrastructure team cannot decrypt your data.
Strict Model Isolation
Our Fast, Pro, and Deep-Research tuned models run in isolated execution environments. Your private enterprise data is never used for continuous training.
Extensive Tooling Safety
Agentic workflows and extensive external tool calls are sandboxed. API keys for third-party integrations never leave your encrypted vault.
# The actual payload sent from the browser:
{
"sessionId": "2de17f49-8aca-484e-86de-67a1c34644bf",
"model": "fast",
// 1. AES Encrypted Message Body
"encMessage": "M619z-9_TraUKTpElBdQpVQ9ZGT8KRvpf3Nb8Ou4-YEb...",
// 2. ECIES Wrapped Ephemeral Key
"key": "BJpx2YaWvYcxfwQ0zipOc3_mchl9wjWszGlu5b4iDXn...",
"files": []
}
SYSTEM SECURITY ENFORCEMENT:
- Triple Encryption: Payload is secured via SSL, AES, and ECIES.
- Ephemeral Keys: The `key` is held strictly in volatile memory during inference.
- Zero Logging: The key dies with the request and is NEVER written to the database.