Bring your own AI. Citadel provides the memory.

The problem with AI today

Every AI session starts from zero. Your LLM doesn’t know what you decided last week. It doesn’t know what’s in progress. It doesn’t know what failed or why.

You can paste context. You can write long prompts. But that’s manual memory — it doesn’t scale, it doesn’t compound, and it disappears the moment the session ends.

What drawbridge changes

Drawbridge is an MCP connection between your AI and Citadel.

When your AI connects through drawbridge, it gains access to your organization’s structured memory:

• Plans and outcomes — what you’re trying to achieve and why
• Decisions and rationale — what was decided, by whom, with what evidence
• Tasks and progress — what’s in flight, what’s blocked, what’s done
• Standards and playbooks — how your organization does things
• Relationships — how everything connects to everything else

Your AI reads this context at the start of every session. It writes back what it learns. The next session — whether it’s you, a colleague, or a different AI — picks up where the last one left off.

Bring your own LLM

Citadel doesn’t replace your AI. It makes your AI better.

• Use Claude, GPT, a local model — anything that speaks MCP
• Your data stays on your infrastructure
• Citadel provides the persistence layer, not the intelligence layer
• Switch models anytime without losing organizational context

What the setup looks like

1. A Linux machine — your server, your cloud, your homelab
2. Claude Code (or any MCP-capable AI client)
3. Citadel MCP — connects your AI to the knowledge base
4. Your first session — Citadel teaches your AI how your organization works

That’s it. No SaaS dashboard. No seat licenses. No vendor lock-in on the AI side.

Two ways to start

Self-hosted

Run Citadel on your own infrastructure. Full control, full privacy. We provide the software and the guidance to get started.

Guided

Connect to a Citadel hub through drawbridge. Your AI learns from curated knowledge — patterns, standards, and playbooks that accelerate your setup. Graduate to self-hosted when you’re ready.

What happens after setup

Your AI starts building organizational memory from day one:

• Decisions get captured as structured artifacts
• Plans accumulate instead of dissolving
• Context survives across sessions, people, and tools
• Work becomes inspectable, challengeable, and resumable

After a week, your AI knows more about your organization’s plans and decisions than most team members do. After a month, it’s the institutional memory you always wanted.

Talk to us →

What Citadel is, and what it is not

Not just: another chat window, a bigger context window, a prettier knowledge base, or a memory patch for one model.

Closer to: a shared memory for planned work, a living record of decisions and rationale, a structure humans and AI can both return to, and a continuity layer behind multiple interfaces.