Security
BLOM is product management software built for building with AI. It keeps your people and AI agents working from the same plan. How BLOM keeps each organization's records apart, who can act in it, what it records, how it recovers, and where your data lives.
An agent key. Its secret was shown once, when it was made.
Every request in BLOM runs as one person or agent, in one organization, and the database holds it there.
Each organization's records sit behind row-level rules in the database itself. A request in one organization can't read or change another's, whatever the screen asks for.
Each member is a Viewer, an Editor, or an Administrator, and BLOM checks the role on the server for every change, not only on the screen.
Inviting someone, granting a role, making an agent key, and connecting a repository are a person's acts. No agent can do them, and an agent can never do more than the person it answers to.
People sign in with an account they already protect. Agents get keys of their own. Every change is written down and never rewritten.
BLOM's records are held in the United States, by the services that run BLOM.
"We don't sell your personal information, share it for advertising, or use it to train AI models."BLOM's Privacy Policy
BLOM runs no AI models of its own, and it uses no analytics, advertising, or tracking tools. Your agents reach BLOM with their own keys, and read only what their keys reach.
What BLOM is designed to do when something goes wrong, and how we check that it works.
BLOM's operations standard keeps backups for at least 35 days, with copies in a separate account that has its own credentials, so losing or compromising BLOM's main account can't reach them.
BLOM is designed so a restore loses no more than the last 15 minutes of changes, and so it's usable again within 4 hours of deciding to restore.
The standard calls for a whole restore rehearsal every quarter, in an isolated environment with Production's data controls, destroyed afterwards, with each result recorded.
After a restore, every sign-in and agent key ends, since a backup can't tell which ones you'd already turned off. You sign in again and give your agents new keys.
Failed sign-ins, denied access, and failing backups raise alerts, which the standard requires to reach us within 5 minutes through a channel that works even when BLOM is down, and to be tested every month.
A written incident plan covers containing the problem, keeping the evidence, and telling you, and the standard calls for rehearsing it every year.
Write to service@boundarylayerlabs.com. Tell us what you found and how to see it, and we'll answer. Ask us there for more detail on anything on this page.