The free beta for Windows 10 and 11 is open, and your key arrives the moment you ask. Get the beta

Syrebrum

Built so you stay in charge.

Sy is meant to work without nagging you, and to never do something that matters without you. These rules are enforced in the software itself, not left to the AI.

What Sy may do, and what waits for you

What Sy may do on its own

  • Read and analyze your data
  • Write reports and drafts
  • Update its own records
  • Publish social posts and reply to reviews, within daily limits you can change

You can switch any of these to "ask me" or "off".

What always waits for you

  • Payments, transfers, refunds and purchases
  • Ad spend
  • Contracts, signatures, legal or tax filings
  • Major changes: prices, accounts, hiring, deleting data, new connections
  • Changing security settings on your computers
  • Contacting people who have not contacted you first

No setting unlocks these. Sy never handles your passwords, keys or tokens.

Safety nets

On top of the hard rules, four more layers sit between Sy and anything that goes out.

Learning week

After setup, anything that goes out waits for your OK for seven days.

Pause switch

Stops everything that goes out.

Record

Every automatic action is recorded and summarized.

Platform rules

No review incentives, no duplicate posts, platform limits respected, and a recording notice on phone greetings.

Your data

Syrebrum is built to keep your business on your own hardware.

Read the privacy policy

  • Sy runs on your own computers. Your files, mail and records are not uploaded to us.
  • Sign-in tokens and keys are stored on your computer, readable only by you.
  • Updates are cryptographically signed, checked before install, and roll back automatically if the new version is unhealthy.
  • Connections to Google, Microsoft, Meta, X and Square ask for the minimum access, read-only by default.
  • With local models through Ollama, Syrebrum can run in a mode that keeps everything on your network.

Zero Trust for AI

Syrebrum and Failstop-Sentinel apply NIST SP 800-207 Zero Trust principles to AI agents and multi-machine workflows. See the full mapping: Zero Trust architecture for AI.

About the security checkup

Syrebrum's security checkup is a checkup, not a guarantee. It does not replace antivirus or endpoint protection, and a clean report does not mean a machine is secure.

How the checkup works

Report a problem

Found a security issue, or something Sy did that it shouldn't have? Email info@syrebrum.com. A person reads it.

Questions about control

Can I give Sy permission to make payments?

No. Payments, transfers, refunds and purchases always wait for you, and no setting unlocks them.

Why enforce the rules in software instead of telling the AI?

Instructions to an AI can be misunderstood or talked around. A rule built into the software doesn't depend on the AI agreeing to follow it. That's why the list of things that always wait for you isn't left to the AI.

What does the learning week do?

For seven days after setup, anything that goes out waits for your OK, so you can watch how Sy works before it does anything alone.

Can I see what Sy did?

Yes. Every automatic action is recorded and summarized.

Is Syrebrum a safety-rated system?

No. It's beta software for your own computers and accounts. Don't rely on it for anything where a failure could cause injury, loss of life or serious harm.

Tell Sy what you want done.

YouCheck every machine and tell me what needs fixing first.

Get the free beta

Free during the beta. Runs on Windows 10 or 11,
with Ubuntu, Debian and TrueNAS machines as workers.