- 1. Where it runs
- 2. Your AI
- 3. Your purpose
- 4. Your account
Does your project need cloud or on-premises?
This comes down to one question: can your documents leave your building? If they hold trade secrets, unreleased product plans, or data a contract or regulation keeps inside your network, that answer decides it.
Cloud
Your Thyme RAG Capsule runs on Cloudflare's global network. Self-serve, ready in minutes, free to start.
- Your data: stored encrypted at rest in Retenta's cloud, isolated to your organization.
- Who processes it: Retenta, on Cloudflare. The AI you pick sees only the passages found for each question, never your whole library.
- Choose a frontier lab (Claude, ChatGPT) and those passages also go to that lab. Sage Cloud keeps them on Cloudflare.
- Setup: minutes. Cost: $0 to start.
- Best for: internal knowledge, support content, product docs, research: material that's private but isn't a crown jewel.
On-premises
A custom deployment on hardware you own, behind your firewall. Retenta can help you buy and install it.
- Your data: with private retrieval and a private model (Sage Private), documents, questions and answers are processed on your hardware.
- Who processes it: your infrastructure. If you later connect an external model such as Claude, it receives the passages sent to it.
- Trade secrets: no third-party host holds your corpus, and no provider's retention policy applies to it.
- Setup: requirements intake, assessment, a quote, then validation on your workload. Weeks, not minutes. Cost: quoted per project.
- Best for: formulas, source code, M&A files, defense and health data: anything whose leak costs you a patent, a deal or a compliance finding.
| Cloud | On-premises | |
|---|---|---|
| Where documents live | Retenta's cloud (Cloudflare), encrypted at rest | Your servers only |
| Who can see a question | Retenta's runtime, plus a frontier lab if you choose one | Your infrastructure, unless you connect an external model |
| Trade-secret exposure | Low, but it is a third party holding your data | Lowest: no third-party host for the corpus |
| Time to first answer | Minutes | Weeks |
| Who runs it | We do | You do, with Retenta's help to source, install and validate it |
| Price | Free tier, then from $7/mo | Custom quote |
Not sure? Start in the cloud with documents that aren't sensitive. The capsule format is the same either way, so you can move it on-premises later.
It sounds like you need a custom solution.
Private deployments are scoped for your hardware and your rules: requirements, then an assessment, a quote for hardware and deployment, validation on your workload, and only then activation. Tell us what you need. Nothing is created in the cloud for you.
Choose the AI that answers
Your Thyme RAG Capsule finds the right passages. This model writes the answer from them. You can switch at any time without rebuilding anything.
What will your solution do?
This becomes your capsule's brief. It shapes how answers are written. The more detail, the better.
Create your account
We'll email you a sign-in link. No password. When you open it, your capsule is created with the AI and brief you chose, and your dashboard opens on a chat with it.
By continuing you agree to the terms. A 45Press product.
Already have an account? Sign in