Thundercloud TechnologySovereignty over AI.
Your rules and your record: you decide what your AI agents may do, and Saturn is designed to keep a record you can review of what they asked to do and what it decided. Saturn runs as a hosted service operated by Thundercloud, in private preview.
Saturn is designed to check each request it handles against that agent’s contract.
Status: private preview. See what works today, and what doesn’t yet →
Agents have real access. Who’s in control?
AI agents now read inboxes, change code and touch business records. Many borrow a person’s own login, so they can do anything that person can. A prompt can ask an agent to behave. It can’t enforce anything, and it can’t tell you afterwards what really happened.
Saturn keeps the rules and the record outside the model, in a separate layer. Each agent gets its own identity and its own contract: what it may read, what it may change, and what it must leave to you.
- ReadAllowedlisted in contract
- DraftAllowedlisted in contract
- SendRefusednot in contract
Give AI a job, not the keys.
- 01
Set limits.
Name the agent, its owner and its job. Grant each agent what its job needs.
- 02
Check what goes through it.
Saturn is designed to check each request against the agent’s contract before it goes through.
- 03
Review the record.
See what was allowed, what was refused, and what Saturn couldn’t observe. When you replace a contract, the old one stops working. Our reference test checks this.
Start with a discovery request that changes nothing and needs no credentials. Once preview access is set up, connect over the HTTP API. Developer guide →
Proof, not promises.
In a local test with simulated email, our reference engine refused an agent that reached past its contract, honoured a replacement contract, and kept one agent’s permissions from leaking to another. We show you exactly where the test stops.
Not shown by that test: live email delivery, complete interception, production security, or a completed outcome for an outside team.
Sovereign AI still needs governing
“build sovereign Canadian AI foundations in compute, data, talent, and infrastructure, ensuring Canadians can trust that the systems they adopt are built and governed on Canadian terms.”Canada’s National Artificial Intelligence Strategy: AI for All, Innovation, Science and Economic Development Canada. Source ↗
Thundercloud Technology Ltd. is not affiliated with, endorsed by or funded by the Government of Canada or any other organisation named here.
Sovereign AI is about more than where the model runs. It’s about who decides what it’s allowed to do.
Here, sovereignty means your rules and your record: you set what your agents may do, and you can review the record of what they asked to do and what Saturn decided. It is not a claim about where data is stored or who runs the infrastructure.
With Saturn, your agents’ limits are set by you and kept outside the model. Saturn is designed to check each request it handles against that agent’s contract.
We’re building Saturn to help teams review each request their agents send through it, and the permission that allowed or refused it.
Designed for hardware-key approvals.
We’re designing Saturn so that granting or changing what an agent may do can require a touch and PIN on a hardware security key. With a hardware key, the private key is designed to stay on the device, so an agent can’t approve its own access. This isn’t offered to preview participants yet.
Built in Calgary, Alberta.
Start with one agent.
Choose one agent, one job, and one action it must never take. Email samantha@thundercloud.gg. We reply to agree scope and environment before any access is granted.
Need help with your first workflow? Thundercloud also offers separately scoped reviews of AI agent access and records, agreed in writing first. You do not need Saturn for this. Talk to us →