# Citadel Cloud Management

> Enterprise AI consulting that moves beyond proof of concept. Citadel designs, secures, governs, deploys, and operates production AI systems and governed AI agents for enterprises — plus free, production-grade cloud education. Fixed-fee engagements, senior architect-led.

Citadel Cloud Management helps enterprises take AI from pilot to production without the governance, security, and reliability gaps that stall most AI projects. Its flagship offering, SIAS (Software In A Service), is a catalog of 506 governed AI agents across 8 domains — each a first-class principal with its own identity, scoped credentials, and full action logging, run under three autonomy tiers (T1 read-and-report, T2 supervised, T3 human-gated). Citadel also publishes free cloud courses and technical guides, open to read without an account.

## Services
- [Enterprise AI Agents (SIAS)](https://www.citadelcloudmanagement.com/ai-agents): 506 governed, production AI agents across 8 domains with per-agent identity, tiered autonomy, and full audit trails.
- [Enterprise consulting & pricing](https://www.citadelcloudmanagement.com/enterprise): Fixed-fee AI strategy, platform architecture, AI security, LLMOps, RAG, and governance; engagements typically $10K–$500K+, scoped after a discovery call.
- [Security & governance posture](https://www.citadelcloudmanagement.com/security): How Citadel isolates tenancy, issues ephemeral credentials (SPIFFE/SPIRE, RFC 8693), and enforces four-eyes controls for AI agents.
- [Book a discovery call](https://www.citadelcloudmanagement.com/book-a-call): A 30-minute, no-pitch working session to scope the highest-value, lowest-risk first step.

## Definitions
- Governed AI agent: an AI agent operated as a first-class principal — with its own workload identity, scoped and ephemeral credentials, an assigned autonomy tier, and a full, attributable audit trail of every action — so a security team can approve it for production.
- Autonomy tiers (T1/T2/T3): T1 (autonomous) is read-and-report only; T2 (supervised) writes to systems of record within scoped credentials and change windows; T3 (human-gated) holds anything touching money, people, or legal behind named approvers.
- SIAS (Software In A Service): Citadel's model of delivering managed, governed AI agents that ship real work — designed, secured, hosted, and monitored by Citadel — rather than a demo chatbot.

## Knowledge hub
- [Resource hub](https://www.citadelcloudmanagement.com/resources): Interconnected reference nodes for enterprise AI, cloud, security, and development — the entry point to Citadel's knowledge graph.
- [What are governed AI agents](https://www.citadelcloudmanagement.com/what-are-governed-ai-agents): Definition and the engineering required to run agents in production.
- [RAG vs fine-tuning](https://www.citadelcloudmanagement.com/rag-vs-fine-tuning): When retrieval beats fine-tuning for enterprise AI, and when it does not.
- [LLMOps vs MLOps](https://www.citadelcloudmanagement.com/llmops-vs-mlops): What changes operationally when the model is an LLM.
- [AI agent frameworks compared](https://www.citadelcloudmanagement.com/langchain-vs-llamaindex-vs-autogen): LangChain, LlamaIndex, and AutoGen for enterprise use.
- [Enterprise AI security](https://www.citadelcloudmanagement.com/enterprise-ai-security): Protecting LLMs, data pipelines, and model endpoints.
- [What is agentic AI](https://www.citadelcloudmanagement.com/what-is-agentic-ai): Planning, tool use, memory, and autonomy vs a chatbot — and the governance it demands.
- [What is the Model Context Protocol (MCP)](https://www.citadelcloudmanagement.com/what-is-mcp): The open standard connecting LLMs and agents to tools and data.
- [AI governance](https://www.citadelcloudmanagement.com/ai-governance): The engineering framework — identity, tiered autonomy, audit — that gets AI approved for production.
- [Zero Trust for AI](https://www.citadelcloudmanagement.com/zero-trust-for-ai): Never-trust, least-privilege applied to agents, model endpoints, and data pipelines.
- [AI agent security](https://www.citadelcloudmanagement.com/ai-agent-security): The agent threat model (prompt injection, excessive agency) and its controls.
- [Vector databases for RAG](https://www.citadelcloudmanagement.com/vector-databases-for-rag): pgvector, Pinecone, Weaviate, and Milvus compared for retrieval.

## Guides
- [The Enterprise AI Agent Blueprint](https://www.citadelcloudmanagement.com/enterprise-ai-blueprint): The governance framework behind SIAS — identity per agent, autonomy tiers, scoped credentials, and audit trails.
- [Blog](https://www.citadelcloudmanagement.com/blog): Practical guides on RAG pipelines, LLMOps, MLOps, AI security architecture, multi-cloud AI, and AI compliance (HIPAA/FedRAMP/GDPR).
- [Free cloud courses](https://www.citadelcloudmanagement.com/free-courses): 17 free courses across AWS, Azure, GCP, DevSecOps, and AI/ML — no signup wall.

## About
- [About Citadel](https://www.citadelcloudmanagement.com/about): Mission, vision, and the architect-led delivery model.
- Contact: info@citadelcloudmanagement.com