Your Data Centre.
Your Rules.
Run the whole Prometheux platform on your own servers, behind your firewall. Business rules execute on data where it already lives, every answer traces to its rules and source rows, and your team operates it.
rules v12 · one rule setProven in production
Their Cloud.
Or Your Walls.
Most AI platforms only run as a service in the vendor's cloud. For banks, insurers, pharma, defence and the public sector, that ends the project at the security review. Prometheux installs inside your own infrastructure, so the review is about your network, not ours.
Your Data Stays
In-House
Prometheux installs on your own Linux servers and connects to the databases, warehouses and files you already run. Rules execute next to the data, results are written where you choose, and access stays under the controls your security team already owns.
Your Own Servers
The full platform runs as containers on standard Linux servers inside your network: engine, APIs, assistant and notebooks.
Data Stays Put
Prometheux reads sources in place and applies rules at query time. No copies shipped to a vendor, no new store to secure.
Your Security Model
Run under your firewall rules, network segmentation and the access controls you already have.
Scale on Your Cluster
For heavy workloads, run the engine on your own Kubernetes cluster.
Pass security review on your own terms. Keep regulated data inside its jurisdiction. Own the logic, the data and the box it runs on.
The Rules Run.
The Model Talks.
Answers come from your business rules, computed deterministically on your data. Language models are used around them: to explain results, translate rules into plain language and power the assistant. You choose the provider, and agents, including the model you may already host, call governed logic through MCP and APIs.
Deterministic Core
Same input, same output. Rule execution doesn't depend on a model's mood.
Your LLM Choice
The LLM provider and model are platform settings you control, separate from the rules.
Agents via MCP
Your agents call business rules through MCP, the REST API or the Python SDK, inside your network.
Lineage on Every Result
Each answer links back to the rules and source data that produced it, ready for audit.
Ground in-house AI in your official rules. Show auditors how every answer was produced. Swap models without rewriting logic.
From Security Review
to Running Rules
Agree the Architecture
Review network, data flows, storage and integrations with your security and architecture teams before anything connects.
Install on Your Servers
Deploy the platform with Docker Compose on your Linux servers. For heavy workloads, run the engine on your own Kubernetes cluster.
Connect Data in Place
Connect the databases, warehouses and files you already run. Nothing is copied out.
Run and Govern
Teams and agents call the rules through apps, APIs and MCP, with lineage on every result.
Built for Data
That Can't Leave
Regulated Flows In-House
Run credit, risk and regulatory reporting rules inside the bank's own infrastructure, with lineage for the supervisor.
Restricted Networks
Reason over engineering and product data under strict classification or export controls, without sending it to a vendor cloud.
Research Data at Home
Apply rules to clinical and research sources where they are governed, instead of exporting them to a vendor cloud.
Plant Risk, Privately
Flag any plant whose supplier chain, at any tier, includes a disrupted supplier, on data that stays in your data centre.
“Prometheux enabled us to build a system that can access our databases, process all the client's historical variables, and provide an intelligent, understandable explanation autonomously.”
Massimiliano MalinHead of Credit Risk · Banca Sella
“With Prometheux, we achieved a speed-up of over 250% on critical tasks involving complex reasoning over knowledge graphs with hundreds of millions of nodes and billions of edges.”
Financial institution
Questions,
Answered
What infrastructure does Prometheux need on-premises?
Standard Linux servers (Ubuntu, Debian, Fedora or RHEL) running Docker Compose. For large workloads, the platform can also run on your own self-managed Kubernetes cluster, planned with our engineers. The install guide is public.
Do we have to send data to a vendor cloud?
No. Prometheux runs inside your infrastructure and connects to your databases, warehouses and files where they live. Rules run at query time on that data, and results stay in your environment.
Who operates the platform?
Your team does. It runs on your servers, under your firewall, network and access policies, and our engineers help plan the install and the architecture review. Data, rules and results stay in your environment.
Do on-prem results depend on an LLM?
No. Results come from your business rules, executed deterministically. LLMs are used for explanations, translating rules into plain language and assistant tasks, and the provider and model are settings you choose.
Is our logic tied to the on-prem install?
No. Rules are readable text you keep in your own version control, and Prometheux deploys in the cloud or on-premises, so the logic you write is not locked to one deployment.
Related Capabilities
Bring AI Inside
Your Walls
Tell us your network, compliance and hardware constraints. We'll walk your security team through the architecture.
