When using Datavizor, you’ll notice the product pages not only tell you all about the functionality built into the protocol, but also where the protocol stops protecting your data, what gets written onchain, what we don’t verify, and what we haven’t built yet.
This issue is all about why you don’t have to ask us where the edges of our products are, and why we think it’s important to tell you.
Stay up to date as we build the infrastructure layer for secure AI collaboration:
OpenMatter is chairing the Decentralized AI Agent Alliance’s Agentic Privacy & Security subgroup. Our next meeting will be held on August 26, 2026 at 1pm EDT, and every other Wednesday thereafter. Please join us.
If you’re curious what DAIAA is all about, you can watch the August 12 meeting to learn more.
Ask Where the Guarantee Stops
The Vendors You Can Trust Will Name Their Own Limits
Every company in this space says the same thing. Your data is protected. Encryption everywhere. Zero trust. The words have been spread so evenly across so many products that they’ve stopped carrying information, which is a problem if you’re the one who has to choose.
So ask this instead: where does the guarantee stop?
Encryption that protects a private key, but exposes the address it points at.
Verification that checks one input, but takes your word for another.
A network that hides your traffic, but publishes the list of who is on it.
Every system has edges like these. They’re where your risk lives, and they’re what a vendor decides whether to disclose.
At OpenMatter, we mention our own limits before you have to ask.
What We Tell You
We describe our limits on our product pages:
Bucket Metadata Is Public: Attach a storage bucket to Datavizor and your access keys are encrypted in your browser before they go anywhere. Not even we can read them back. The walkthrough says so. It also says the bucket name and object path are written onchain, where anyone can read them. Your key is protected. Its address isn’t.
Readiness Is an Assertion, Not a Verification: Registering a volume ends with a button marked Ready. The guide tells you this is your assertion that the data is present, not a verification that it is. We don’t check.
Credential Rotation Requires Re-creation: Changing a volume’s credentials means retiring it and starting over. We haven’t built automated key rotation for volumes yet.
Bounded Execution: A collaboration session runs built-in model types and won’t execute arbitrary code.
Private Network, Public Roster: Private networking keeps your traffic inside an encrypted mesh but puts the membership roster in public.
Immutable Ownership: An organization’s owner can never be changed, which is either fine or disqualifying, and better to know before you create one.
None of that makes the product look better. All of it makes it possible to decide whether the product fits what you’re doing, which is the only question you need answered.
Why So Few Products Do This
Publishing where your products’ features stop provides your competitor a line for their comparison page and gives a buyer a reason to hesitate. The incentive runs toward the vague claim, which is why that kind of claim is everywhere.
For anyone who has to live with the decision, the incentive runs the other way. A limit you find during evaluation is a constraint you design around. The same limit found in production is an incident. The vague claim only serves to delay when you meet the truth.
What to Ask
Ask this of every vendor, including us.
Where does the encryption stop? What can the provider see? What gets recorded publicly? What happens when a key is compromised? What can’t this do yet?
Specific answers tell you about the product. A sentence about being secure tells you about the vendor.
Every example above can be found in Datavizor right now. Open any of those sections and you’ll see the product’s limitations before you commit to anything.
The free plan is ready to try today at datavizor.openmatter.network.
— The OpenMatter Team
If you know someone who would benefit from reading this article, please share it:
OpenMatter is building the verifiable trust layer that enables AI agents to securely collaborate on sensitive data sets. If you’re in a regulated industry and need a better way to prove that your data is secure, contact our team to learn how masked compute can help.