Cloud environments in minutes. Governed from the first click.

EnvCraft is a platform for self-service cloud environments. A data scientist or developer picks a cloud, a software stack and a machine size; EnvCraft checks the request against the company's policies, quotas and budgets, builds the machine with Terraform, installs the software with Ansible, mounts the datasets and records every step. What normally takes a ticket and several weeks takes minutes, and it stays inside the rules the company has set.

Amazon Web Services, Microsoft Azure, Google Cloud and Hetzner Cloud are supported today. STACKIT, gridscale and on-premise clouds such as OpenStack are on the roadmap. Each company decides which clouds its people may use, with which credentials and which instance types. A cloud is treated as an endpoint plus credentials rather than a hard-coded vendor name, so a private cloud with a Terraform provider fits the same model.

In yours. EnvCraft provisions into your own cloud accounts with your own credentials, which are encrypted at rest with AES-256-GCM. There are two ways to run the platform itself: as a managed service that brandcompete operates inside your infrastructure, or as an on-premise licence your own platform team runs. In both cases your data and your cloud accounts stay under your control.

Thirty-eight stacks are ready to use, each built and tested on a real machine. They range from a minimal Rocky Linux base through notebooks and machine learning (JupyterLab, a Data Science notebook, R, scikit-learn, TensorFlow and PyTorch on CPU, MLflow, Spark), databases (PostgreSQL with pgAdmin, SQL Server Express and Oracle Database Free with CloudBeaver, DuckDB), BI and apps (Superset, Streamlit, Dash, Chainlit, Airflow, Node-RED, Baserow, Appsmith) to developer and operations tools (VS Code in the browser, a container host, Node, Flask, a JVM workspace, a GitHub Actions runner, Traefik, observability, BookStack and MediaWiki). Thirty-one of them publish a web interface that opens in one click, with the sign-in generated for you. The Knowledgebase in the product lists every stack's software and licences.

Four ways. Quotas cap vCPU, memory, disk and the number of machines per person, and are checked before anything reaches a cloud. Budgets set a monthly ceiling per company or per person that either warns or refuses a new machine. Power schedules start and stop a machine at set times, say weekdays from seven to seven, so nobody pays for a notebook server overnight; on AWS, Azure and Google Cloud a stopped machine is billed only for its storage and address. And the Costs screen shows the fleet run rate, an itemised estimate per appliance and a spend analysis over any period, filtered and split by vendor, instance type, person, stack or project, including machines that have already been destroyed. The figures are estimates from the providers' public rate cards rather than a reading of your bill; billing data straight from the vendor is on the roadmap.

In projects. Everyone has a personal workspace, and a project adds colleagues at Read, Write or Admin: Read sees what is in it, Write uses and changes it and mounts its datasets read/write, Admin removes things and manages the members. Every appliance, dataset, data source and repository belongs to exactly one project, the switcher in the header narrows every list to it, and the Costs screen shows what each project has spent. Where the installation has a public zone, an appliance is also published under its project's address with a real certificate. A GitHub repository registered in a project can be deployed straight onto a new web server by anyone with Write.

Sign-in is by single sign-on with Microsoft Entra ID or Google, restricted to the directories and email domains you name, or by password. Seven built-in roles from Administrator to Auditor can be extended with your own custom roles, over more than ninety fine-grained rights, and every change to a role shows a diff before it is saved. On top of the role, projects give each member Read, Write or Admin, and an action needs both. Each company sees only its own people and resources, and somebody else's environment is not merely forbidden but invisible. The audit trail records sign-ins, provisioning, sharing, terminal sessions and refused actions, and exports to CSV, Excel or PDF.

Mason is the assistant built into EnvCraft. You describe what you need, such as a Jupyter box for the churn analysis or what your team is spending on AWS this month, and it looks things up, proposes an appliance and builds it once you say so. It cannot break your rules, because it has no rights of its own: every lookup and every action carries the signed-in person's own token, so the same policies, quotas, budget and audit trail apply as in the wizard, and an appliance it creates is yours in the audit trail. Anything that spends money, grants access or changes a setting waits for your confirmation, and it is never given secrets, the terminal or the power to delete people. Mason runs on Anthropic's Claude models; the key stays on the server and never reaches a browser, using Mason is a right of its own, and an installation that must not call an external model leaves it switched off. The same tools are available to Claude Desktop, Claude Code or your own client through EnvCraft's MCP server.

No. Environments are built with plain Terraform modules and Ansible playbooks that your team can read, run and version, in your own cloud accounts. What EnvCraft installs is open source, with three exceptions that are free of charge under their vendors' terms: SQL Server Express, Oracle Database Free and the Netdata dashboard in the observability stack. Every stack's licences are listed in the product's Knowledgebase with what they permit and require, and nothing is offered to your users over the network under the AGPL, apart from one stack still under review, R with Shiny Server.

Pricing depends on how you run it — managed service or on-premise licence — and on the size of the installation. Write to sales@brandcompete.com and we will put a figure against your case. Cloud resources themselves are billed by your provider to your own account; EnvCraft estimates them for you before and after they are created.

With a pilot. About one week to set EnvCraft up in your cloud account, connected to your single sign-on and carrying your policies and budgets; four to six weeks in which one team works on real projects with real data and real guardrails; then about a week for a usage, cost and governance report and a roll-out plan. You decide on facts rather than on a demo.