Deployment
How you run Cracken, from fully managed to fully air-gapped. Every model keeps your operations and sensitive data under your control; they differ in who runs the infrastructure and how the platform reaches its models.
Deployment models
- Cracken Cloud (SaaS) — the managed service at app.cracken.ai. Fastest to start; Cracken runs the platform, you run Tentacles on hosts you control so execution still happens next to your targets. SaaS data is stored on servers in the US and the UK.
- Private cloud — Cracken deployed into your own cloud account, so the platform, its database, and its Tentacles all live inside your tenancy.
- On-premises — Cracken deployed entirely within your own infrastructure, keeping your data and security assessments inside your environment.
- Air-gapped — an on-premises deployment with no real-time internet access. It operates fully offline, provided the network can reach either a commercial LLM endpoint or enough compute to host self-hosted models. Output quality is somewhat reduced without live model access.
What stays in your environment
Across every self-managed model, Tentacles and their tooling run on hosts you control, tool credentials and synced findings stay inside your boundary, and only selected, redacted results leave your network. See BYOK for using your own secret manager, model keys, and self-hosted LLM endpoints.
Getting a deployment
Private-cloud, on-premises, and air-gapped deployments are arranged with the Cracken team, who provide the detailed deployment guide, infrastructure requirements, and hardware sizing for your environment.
Contact sales@cracken.ai to scope the model that fits your constraints.