> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://conductorai.ferndocs.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://conductorai.ferndocs.com/_mcp/server.

# Security

> Built for the most secure environments with FedRAMP High accreditation and multi-classification support.

The Conduit platform is built on a microservice architecture supporting deployment into complex environments, including robust CVE mitigation, hardened images, and support for multiple LLMs.

## Accreditations

| Accreditation | Status     |
| ------------- | ---------- |
| FedRAMP High  | Authorized |
| NIPR (IL-5)   | Deployed   |
| SIPR (IL-6)   | Deployed   |
| JWICS         | Deployed   |

## Architecture

### Microservice design

Each platform capability runs as an independent, containerized service:

* **Search service** — Document indexing and retrieval
* **Workflow engine** — Orchestration and step execution
* **Review service** — Collaboration and approval management
* **Auth service** — Identity, access control, and audit logging
* **LLM gateway** — Model routing with data-loss prevention guardrails

### Deployment models

#### Cloud-hosted

Fully managed deployment on ConductorAI infrastructure. Data residency options available for US Government customers.

#### Self-hosted

Deploy via Docker containers within your own infrastructure. The platform supports air-gapped environments with no external network dependencies.

```bash
docker pull conductorai/conduit:latest
docker-compose up -d
```

#### Hybrid

Core processing runs in your environment while management and monitoring connect to ConductorAI's control plane through an encrypted tunnel.

## Data protection

* **Encryption at rest** — AES-256 for all stored data
* **Encryption in transit** — TLS 1.3 for all communications
* **Key management** — Integration with AWS KMS, Azure Key Vault, and HSM devices
* **Data isolation** — Tenant-level isolation with dedicated encryption keys

## LLM security

The platform supports multiple LLM backends with strict data controls:

* No training on customer data
* Data-loss prevention filters on all LLM inputs and outputs
* Support for on-premises model deployment
* Configurable model selection per classification level