Flui turns regional cloud providers and servers you own into one platform — simple for teams, operable by agents, and owned by you. Run it yourself for free, or have us build it with you.
Need a platform built, migrated or run for you? Talk to us →
Runs on your own servers or on the supported providers.
| Node | CPU | Status |
|---|---|---|
| control-cluster-master | 12% | Running |
| control-cluster-worker-1 | 7% | Running |
No variable sets
Manage variablesLive values every 5 seconds · history over the last 3 hours
| Node | CPU | Memory |
|---|---|---|
| control-cluster-master | 12.3% 4 vCPU | 48.8% 3.7 GB / 7.6 GB |
| control-cluster-worker-1 | 7.3% 2 vCPU | 36.8% 1.3 GB / 3.7 GB |
| Application | CPU | Memory |
|---|---|---|
metrics flui-control | 0.03 cores3% of limit | 540 MB53% of limit |
postgres-main production | 0.04 cores6% of limit | 312 MB41% of limit |
demo-app production | 0.01 cores2% of limit | 96 MB37% of limit |
logs flui-control | 0.02 cores4% of limit | 180 MB35% of limit |
One disk on the master node, shared with every other node. This is where most of your apps' data lives.
Flui picks the right one automatically based on the app. You almost never need to choose by hand.
When an app has nowhere to run, Flui provisions the machine that fits, within the monthly budget you set, and removes it when it is no longer needed.
| When | What Flui saw | What happened |
|---|---|---|
| 17:51 | A worker joined and took the waiting app. | node joined |
| 17:49 | One app had nowhere to run: 512 MiB more than the fleet holds. | node ordered |
| 04:10 | A third node had nothing left to run after the night drop. | node removed |
A preview of the dashboard. Look around the real one, live →
Pick the European provider that fits each project, or use servers you already own. Flui turns them into one platform — clusters, networking, applications, databases and observability under a single operating model — so your team spends its time shipping, not assembling.
The parts a production platform needs, running on your own servers or on the supported providers — the same way on each, where each can support it.
What ships, how each part works, and what is not finished yet. See the platform →
Providers are not identical, and Flui does not pretend they are. Each one declares what it supports, and where one falls short Flui says so and adapts.
A capability, not a topology you are pushed into. One provider is a complete setup — and when requirements change, the platform does not have to be rebuilt.
Intent and capabilities are concepts a person can review and an agent can act on, which is the difference between an API wrapper and a platform.
Flui models infrastructure so that both a person and a coding agent can understand and operate it. Agents work with applications, databases and capabilities over MCP — not with raw provider APIs — and the result is state you can still read afterwards.
Flui is open-source software, not a trial of a commercial product. Install it, connect your providers and run it with your own team — you never have to talk to us.
Pick where it runs. You get the commands to paste in your terminal: your keys stay on your machine. Read the guide
npm install -g @flui-cloud/cliflui vault initflui config set hetznerflui server-types list --provider hetzner --region nbg1 --fitsflui env create --provider hetzner --region nbg1 --node-size cx23gh issue create --repo flui-cloud/flui-core --title "Install: hetzner nbg1 cx23" --body "$(flui version)"We build Flui, and we design, migrate and operate platforms with it. It starts with a free first call, and every engagement is quoted for your case.
Choose how much help you want — including none.