Bring your AI in-house.
Install Eldric on hardware you control in minutes. Pre-built packages — no build from source, no cloud account, nothing to phone home.
One command, or plain dnf.
Supported on Fedora 42+, RHEL 9+, CentOS Stream 9+ and the RHEL-family derivatives (Rocky, Alma). No GPU required — Eldric runs on CPU and consumer hardware, even a microprocessor; a GPU only makes it faster.
Recommended — bootstrap & run
Or add the repo and install the package
Then browse to https://<host>/chat — the first sign-up becomes the admin.
A client, or the whole thing.
Desktop client (GUI + CLI)
The native macOS app and command-line client — talk to an Eldric you run elsewhere, or to cloud backends, from your desk.
Eldric on your server (full stack)
The complete AI on your own hardware — chat, retrieval, agents, models and cluster, all from one install.
What you get, day one.
Chat shell + agents
A ready web chat and a full set of specialized agents, out of the install.
140+ scientific APIs
A unified surface over scientific data sources, answered from authoritative catalogs.
Plugin marketplace
Extend it with tools, filters and integrations — configured from the admin UI.
Multi-tenant + RBAC
Isolated tenants and role-based access, so many teams share one install safely.
Training pipeline
Fine-tune and distil models on your own data and GPUs.
Audit ledger
A tamper-evident record of what happened — for regulated environments.
One box today, a cluster when you need it.
Start on a single machine and add nodes later — the same install scales from an edge box to a multi-tenant datacenter. The full deployment and clustering guide lives in the docs.
Your AI, running tonight.
New to running your own AI? We help you adopt it — from first install to production. Just ask.