llms.txt
# Hyperfluid
> Hyperfluid is an internal cloud: 20 self-service services in a single console, installed in your datacenter, your cloud account or an air-gapped network.
> It is software, published by Nudibranches Technologies (France). You install it inside
> your own perimeter and your teams provision resources themselves, from the same console.
What Hyperfluid is:
- A complete internal cloud with a 20-service self-service catalog, running inside your perimeter.
- The same software and the same console on your datacenter, your own public cloud account
(BYOCloud) or an air-gapped network, with no code rewrite between targets.
What Hyperfluid is not:
- Not a hosting provider: we do not run a datacenter for you. The contract, the bill and the
quotas of the underlying infrastructure stay in your name.
- Not a single SaaS product: it is a catalog of 20 services sharing one console, one IAM and
one monitoring stack.
- Not a set of open source building blocks to assemble: you consume the services, you do not
integrate them.
Key facts:
- Category: internal cloud platform (private cloud, self-hosted, air-gap capable).
- Catalog: containers, private registry, dev workstations, S3 buckets, managed PostgreSQL 17,
Redis-compatible key-value store, managed Kafka, a federated lakehouse (Trino SQL engine on
Apache Iceberg, PostgreSQL, MySQL, ClickHouse and the JDBC family), a big data analytics SQL IDE, Data Product APIs (SQL, GraphQL, REST, MCP, search),
connectors (open data, Google Drive, IMAP email), local inference on your own GPUs behind an
OpenAI-compatible endpoint plus a third-party model catalog, AI agents, document pipelines
that build a queryable knowledge base (OCR, classification, summaries, routing), AI dashboards,
Git forge with managed CI runners, GitOps operators and a Terraform provider, IAM with SSO and
OIDC federation and a readable audit log, secrets vault, monitoring and logs.
- Deployment targets: your datacenter or private cloud, your own public cloud account
(BYOCloud), or an air-gapped network. No outbound connection is required to run.
- Pricing: shared managed tenant with a 6-month free startup program, then from €200/month, indexed on the
resource envelope you reserve;
self-hosted €145,000/year for one region and
€60,000 per additional region, "Onde Porteuse" revenue sharing at 9.5% of net revenue.
- Roadmap, not available yet: VMware replacement (early 2027), Rainbow Container rack (mid 2027).
- The site exists in French (default, at the paths below) and English (same paths under /en/).
## Key pages
- [Home](https://hyperfluid.cloud/): What Hyperfluid is: an internal cloud of 20 self-service services, installed inside your own perimeter.
- [Pricing](https://hyperfluid.cloud/pricing): Shared managed tenant, six months free for startups then from €200/month per resource envelope, a self-hosted annual license, and Onde Porteuse revenue-sharing for software vendors building on Hyperfluid.
- [Manifesto](https://hyperfluid.cloud/manifesto): Why cloud infrastructure should depend on no one, our stance on European digital sovereignty.
## By audience
- [For CIOs / IT](https://hyperfluid.cloud/dsi): A sovereign internal cloud with a self-service catalog of data and AI services, and the governance you define, without depending on hyperscalers.
- [For software vendors](https://hyperfluid.cloud/editeur): Start fast and sell to enterprise without rewriting: build on the cloud, then deploy on-prem when clients demand it, without changing a line of code.
- [For Integrators](https://hyperfluid.cloud/integrateur): Deliver sovereign, agentic projects faster with off-the-shelf building blocks: Trino analytics hub, Kafka streaming, data APIs, agentic blocks.
## Platform features
- [Knowledge base building](https://hyperfluid.cloud/features/document-agents): Document pipelines that turn unstructured documents (PDF, Word, images, email attachments) into a structured, queryable knowledge base: OCR extraction, AI classification, progressive summaries (Hippocampe) and routing, tracked from a pipelines dashboard.
- [AI Agents](https://hyperfluid.cloud/features/ai-agents): Self-service autonomous AI workers: a prompt, a model and read-only data access, running inside your sovereign perimeter (local vLLM models or EU APIs).
- [Connectors & ingestion](https://hyperfluid.cloud/features/connectors): One-click import of 23 open data sources into Iceberg tables, Google Drive sync and inbound email capture, with no pipeline to write.
- [AI Dashboards](https://hyperfluid.cloud/features/dashboards): Copilot-driven analytics: ask questions in natural language, pin the answers as widgets, share governed dashboards in the console.
- [Containers as a Service (Caastor)](https://hyperfluid.cloud/features/caastor): Deploy your container applications in a few clicks, with automatic public URL, HTTPS and secrets injection.
- [Data APIs & MCP](https://hyperfluid.cloud/features/data-apis): Multi-protocol access to your data: SQL, GraphQL, REST, MCP and search, with enterprise security.
- [Federated lakehouse (Data Hub)](https://hyperfluid.cloud/features/data-hub): Multi-source queries with the Trino SQL engine and Apache Iceberg: PostgreSQL, Iceberg, MySQL, MariaDB, ClickHouse and the JDBC family queried in one statement, for cross-source analytics and complex KPIs.
- [GitOps & Infra as Code](https://hyperfluid.cloud/features/gitops): Infrastructure as Code with declarative operators (YAML CRDs) and a Terraform provider, for reproducible, automated deployment.
- [Managed Kafka streaming](https://hyperfluid.cloud/features/kafka): Provision a Kafka cluster in a few clicks: optional Schema Registry, high availability, presets from development to enterprise.
- [Secrets vault (Passe-Partout)](https://hyperfluid.cloud/features/passe-partout): Secrets vault: store secrets once, injected automatically into your applications, with no passwords in code.
- [S3 object storage](https://hyperfluid.cloud/features/ceph): S3-compatible object storage deployed inside your own infrastructure, managed from the console.
- [Private AI platform](https://hyperfluid.cloud/features/sapience): Model serving on your own GPUs with an OpenAI-compatible endpoint: shared generation, embedding and OCR models, vLLM and HuggingFace runtimes, a catalog of third-party models behind the same endpoint, API keys, rate ceilings and consumption tracking.
- [Big data analytics (Petite Requête)](https://hyperfluid.cloud/features/petite-requete): SQL IDE in the console on top of the Trino engine: query all your data at scale, real-time results, history, saved queries, built-in governance.
- [Monitoring & logs](https://hyperfluid.cloud/features/monitoring): Built-in monitoring of CPU, memory, request rate and latency in the console, with no external tool to set up.
- [Managed PostgreSQL](https://hyperfluid.cloud/features/managed-postgresql): Sovereign managed PostgreSQL 17: high availability, scheduled backups and point-in-time restore.
- [Managed Redis-compatible cache](https://hyperfluid.cloud/features/managed-key-value-cache): Redis-compatible in-memory cache, provisioned self-service from your sovereign console.
- [Private container registry](https://hyperfluid.cloud/features/container-registry): Private container registry in your cluster with role-based access control and vulnerability scanning.
- [Git forge & managed CI](https://hyperfluid.cloud/features/source-control): Private, sovereign Git forge hosted in your cluster, with fully managed CI runners, SSO and dev-workstation integration.
- [AI Dev Workstations](https://hyperfluid.cloud/features/cloud-dev-environments): IDE, terminal, Git and an AI coding agent in an isolated, ready-to-use workspace.
- [IAM, governance & audit](https://hyperfluid.cloud/features/security): Fine-grained authorization, office-hours and zero-trust controls, and every access decision readable in the audit log, in the console.
