--- title: "Rock8Cloud Enterprise - On-Prem Platform With AI Agents" description: "Run the whole Rock8Cloud platform inside your network, on your hardware with your own models. AI code review, agent sandboxes, custom blueprints, SSO and SLA." --- # Your hardware. Our platform. Rock8Cloud Enterprise runs the whole software factory inside your network: deployment, AI code review, agent sandboxes and logs & health, on your hardware or your private cloud. Builds, databases and logs stay on your machines. Host your own models and AI requests stay there too. ## The same platform, on your machines. Everything rock8.cloud runs, installed inside your boundary. The Rock8Cloud MCP server is included, so the Claude app, Claude Code, ChatGPT, Codex and Cursor operate your install the way they operate ours. [Connect Claude and Claude Code](https://rock8.cloud/how-to/connect-claude) · [Connect ChatGPT and Codex](https://rock8.cloud/how-to/connect-chatgpt-codex) - **Deployment**: Push to deploy, a preview environment for every pull request and one-click rollback, on your own machines. - **Code review**: Every pull request gets an AI review before your team looks at the code, run by the models you choose. - **Agent sandbox**: Hand a task to a cloud agent. The agent works in an isolated sandbox on your hardware and returns a pull request. - **Logs & Health**: Build logs, runtime logs, metrics and uptime for every service, readable by your team and your agents. - **Databases**: PostgreSQL, Dragonfly (Redis) and S3-compatible storage, provisioned in one click and wired to your services. - **Pages**: Agents publish reports and previews to their own address on your domain, private by default. ## You decide who runs Rock8Cloud. Same platform, same hardware. You choose who handles upgrades, monitoring and on-call. - **Managed by us**: Rock8Cloud runs on your hardware and our engineers keep the install running through the access you grant. Your team ships and never maintains the platform. Who does what: Hardware: you, Upgrades: us, Monitoring: us, On-call: us, Escalation under SLA: us. - **Run by your team**: We install Rock8Cloud and hand the keys over. Your operations team runs the platform, with tested releases and our engineers behind them. Who does what: Hardware: you, Upgrades: you, Monitoring: you, On-call: you, Escalation under SLA: us. ## Your models, your keys. Code review, agent sandboxes and custom agents run on the providers you connect in your install, per workload. - **Bring your own keys**: Connect your Anthropic, OpenAI or other provider accounts. Usage bills to you directly, at your own rates. - **Models inside your network**: Point the platform at any OpenAI-compatible endpoint, including models you host yourself. Prompts and code stay in the building. - **Offline, except git**: With models hosted inside your network, everything else runs without internet access. Git goes through GitHub today. Self-hosted GitLab support is on the way. ## Every new app starts on your standards. Your install ships with custom blueprints built from your own template repositories. Developers and their agents start every new app from them, so teams stay aligned with company code standards from the first commit. - **Your stack, ready to fork**: Your runtimes, folder structure, linting, tests and Dockerfile in one template, wired to the right databases. A new app forks the blueprint and Rock8Cloud builds and deploys the app in one step. - **Agents start there too**: Claude, ChatGPT, Codex and Cursor start new apps from your blueprints through the MCP server. The agent rules in your template come along, so agents write code your way. - **Your services, one click**: The one-click flow Postgres has today, extended to your own databases and internal services. ## What security and procurement ask for. - **Single sign-on**: Your team signs in through your identity provider, set up as part of your install. - **Audit log**: Deploys, rollbacks, member changes, agent sessions and API keys, recorded with who and when. Searchable, with retention you set. - **Roles and scopes**: Owners, members and read-only readers, with narrower API scopes for each member. - **Isolated agents**: Every agent session gets a separate sandbox with no route into your private network. The agent works on a private copy of the repository and never holds your git credentials. - **Contractual SLA**: Response times, availability and limits agreed up front, in writing. - **Direct support**: A direct line to the engineers who build the platform. ## From first email to first deploy. 1. **Talk to us**: Email support@rock8.cloud with Enterprise in the subject. Tell us about your team, your infrastructure and your compliance rules. 2. **Scope the install**: We agree on the hardware, the operator and the models, and put everything in writing. 3. **Install**: Our engineers install the platform on your hardware, connect your git, your identity provider and your models, and set up your first blueprints with you. 4. **Ship**: Your team deploys the first service. From there we keep operating the platform, or your team takes over with our support. ## FAQ ### Can Rock8Cloud run on our own servers? Yes. On the Enterprise plan, Rock8Cloud runs the whole platform inside your network: deployment, AI code review, agent sandboxes, logs & health and Pages, on your own hardware or your private cloud. We operate the platform for you, or your team runs Rock8Cloud with our support. Email support@rock8.cloud with Enterprise in the subject. ### Does our code leave our network? Builds, deployments, logs and databases stay on your hardware. AI requests stay inside when the models do, or go straight to your own provider account when you use your keys. Git is the one exception today: repositories live on GitHub until self-hosted GitLab support lands. ### Can Rock8Cloud run air-gapped, without internet access? Almost. With models hosted inside your network, everything except git runs without internet access. Git goes through GitHub today, and we are working on self-hosted GitLab support so the whole platform can run offline. Claude Code, Codex and Cursor reach the MCP server from inside your network. The Claude and ChatGPT apps connect from their own cloud, so they need the MCP endpoint reachable from there. We scope every install with you, so you know exactly what runs where. ### Can every new project follow our company code standards? Yes. Your install ships with custom blueprints built from your own template repositories: your stack, structure, linting, tests and agent rules, wired to the databases they need. Developers and their coding agents start every new app from them, so code follows company standards from the first commit. We set up the first blueprints with you during the install. ### Can we try Rock8Cloud before the enterprise conversation? Yes. The cloud version at rock8.cloud is the same platform. Start a 30-day free trial with no credit card and deploy a real service. Your services are plain git repositories and Dockerfiles on PostgreSQL, Redis and S3, so they move to your own hardware with you. ### How much does Rock8Cloud Enterprise cost? Enterprise pricing is quoted per install, based on your team size, your infrastructure and who operates the platform. Email support@rock8.cloud with Enterprise in the subject and we send a quote. Self-serve plans start at €5 a month and are listed at rock8.cloud/pricing. ## Contact Email [support@rock8.cloud](mailto:support@rock8.cloud?subject=Enterprise) with "Enterprise" in the subject. Pricing quoted per install. To try the same platform first, [start the cloud version](https://app.rock8.cloud?utm_source=enterprise&utm_medium=landing&utm_campaign=enterprise), 30 days free. Self-serve plans: [rock8.cloud/pricing](https://rock8.cloud/pricing). Canonical page: [rock8.cloud/enterprise](https://rock8.cloud/enterprise).