Buy it. Host it. Own it.

Some organizations can't put their data in anyone else's cloud - banks, hospitals, defense contractors, public institutions. For them, the cloud version of an AI platform isn't a smaller risk; it's a non-starter. So ToRun ships a second way: the entire platform, running inside your own network, against your own local models, with nothing ever leaving your walls.

This isn't a stripped-down "self-hosted edition" with the good parts removed. It's the same platform - chat, generation, workflows, personas, the marketplace, billing, moderation, the audit trail - deployed on your hardware and under your control.

Talk to us about on-prem · See deployment & licensing options

Run it on your own local LLMs

The point of on-prem is that your data and your models stay yours. ToRun's capability-first router doesn't care whether a model lives at a cloud provider or on a GPU in your own rack - it routes to whatever you've configured. Point the platform at your private model endpoints - Ollama, vLLM, LM Studio, or any OpenAI-compatible server - and every chat, every workflow step, every generation runs against models you host. No prompt, no document, no customer record ever leaves your perimeter.

You can mix and match, too: run sensitive workloads on local models while still reaching a cloud provider for the occasional task that needs it - with a clear, per-call record of exactly which calls went where.

One command. Zero setup. No consultant.

On-prem AI platforms are notorious for six-figure "implementation" engagements. ToRun's installer makes that headline cost zero. The whole stack - the database, the cache, the event bus, the .NET API, and the web app - is built from source inside Docker and brought up with a single command.

• The only prerequisite is Docker. No .NET SDK, no Node, no special tooling on the host - the build happens in containers.
• Secrets generate themselves. On first run the installer creates strong random credentials, a signing certificate, and an at-rest encryption key. You don't hand-craft a thing.
• It's idempotent. Re-run it any time to rebuild and reconcile - migrations and seeds are safe to apply again.
• One origin, one port. A single web edge serves the app and proxies the API, so there's no CORS tangle and only one thing to expose.

When it finishes, you get a URL and an admin login. That's the install.

A tier for every kind of buyer

Self-hosting isn't one product - it's a spectrum, and ToRun meets you where your compliance team needs you to be.

• On-prem deployment - the full platform inside your private network, licensed per seat per year with maintenance bundled in. Your data never leaves; your auditors can inspect the running system directly.
• Source-code license - for banks and the most regulated industries, a perpetual snapshot of the codebase for unlimited internal seats. You hold the source. Pair it with an annual Platform Evolution & Expert Knowledge subscription that keeps new models, providers, and security hardening flowing - plus architect advisory hours and a priority SLA.
• Signed auto-updates - a verified update channel delivers new releases to your on-prem install without a manual upgrade project, so staying current doesn't mean a migration every quarter.
• Code escrow - an optional escrow arrangement gives you continuity guarantees no matter what happens upstream. For a regulator, that's the difference between "trust the vendor" and "the operation is protected in writing."

The same trust guarantees, now under your roof

Everything that makes ToRun auditable in the cloud comes with it on-prem - and now you control the keys to all of it.

• Hash-chained audit trail - every sensitive action is cryptographically linked to the one before it, so tampering is mathematically detectable. A nightly integrity check re-walks the chain. On-prem, that record lives entirely on your infrastructure.
• Atomic, itemized billing - even when the models are yours, every AI call still writes one transparent record, so internal chargeback and cost attribution work exactly as they do in the cloud.
• Encrypted secret vault - any provider keys you do configure are encrypted at rest with a key that never leaves your install.
• GDPR & KVKK flows - data-subject access, export, and erasure run end-to-end inside your environment, producing the paperwork your auditors expect.

Why it matters

Most vendors treat on-prem as an afterthought - a tarball, a PDF, and a steep services bill. ToRun treats it as a first-class way to run the product: the same code, the same features, installed in minutes, pointed at your own models, owned outright if you need it. You get the entire AI platform without the trade-off of handing your data to someone else's cloud.

What you get

• Run on your own local LLMs bi-cpu - point the platform at your private Ollama, vLLM, or LM Studio endpoints. Every call runs against models you host - nothing leaves your network.
• One-command install bi-terminal - the whole stack builds from source inside Docker and comes up with a single command. The only prerequisite is Docker; secrets generate themselves.
• Per-seat on-prem licensing bi-hdd-network - the complete platform in your VPC or data center, licensed per seat per year with maintenance included. Zero data leakage by design.
• Source-code license for banks bi-file-earmark-code - a perpetual codebase snapshot for unlimited internal seats, with an ongoing evolution subscription and architect advisory.
• Signed auto-updates & escrow bi-shield-check - a verified update channel keeps you current without a migration project, and optional source escrow guarantees continuity.
• The full trust stack, your keys bi-lock - hash-chained audit, atomic billing, encrypted vault, and GDPR/KVKK flows all run inside your perimeter, under your control.