Kingly IntelligenceSystems Start a project

Security & Trust

What we do, and what we do not claim

For owners deciding whether to trust us with their systems. Here are the eight practices we run on our own products, and what we do not claim.

In our own stack

Private networking

Tailscale zero-trust networking for remote access, instead of opening ports.

In our own stack

Sign-in in front of everything

Workspaces sit behind sign-in, and the app looks up your machine at runtime instead of baking in its address.

In our own stack

Least access

Scoped device tokens and explicit consent, so a device or agent gets only what it was granted.

In our own stack

Humans approve sensitive actions

Gated actions wait for the workspace owner. Agents cannot approve themselves.

In our own stack

Isolation by default

InfiniBot Cloud gives each workspace its own processes, storage and gateway. No shared runtime or pooled memory between customers.

In our own stack

Database-level rules

Postgres row-level security enforces access in the database itself, not only in the app.

In our own stack

Secrets stay secret

Stored keys are encrypted with AES-256-GCM and shown only masked. We keep secrets out of source code.

In our own stack

Audit trails

A record of what ran, who approved it and what changed.

Pick a sample request to see where it stops.

The eight practices in plain words

What each one does, and what it stops

These run in our own stack. Which of them apply to your system is something we agree with you during discovery and design.

  1. Isolation by default

    InfiniBot Cloud gives each workspace its own processes, storage and gateway, the front door its messages pass through. Nothing runs in a shared pool with another customer.

    Who or what does it
    The platform sets this up for every workspace.
    What goes wrong without it
    In a shared setup, one customer’s fault or leak can spread to the next.
  2. Secrets stay secret

    A secret is a password or key that lets software into another service. Stored keys are encrypted with AES-256-GCM, a strong standard that makes them unreadable without the right key. Screens only ever show them masked, and we keep them out of source code.

    Who or what does it
    The platform encrypts them. Our engineers keep them out of the code.
    What goes wrong without it
    Anyone who sees the code or the screen could copy a key and use your accounts.
  3. Sign-in in front of everything

    Nobody reaches a workspace without signing in. The app looks up your machine when it needs it, instead of having the address written into the app.

    Who or what does it
    The sign-in service checks every visitor first.
    What goes wrong without it
    A workspace address in the app itself is easier for outsiders to find.
  4. Least access

    A device or agent gets a token, a digital pass limited to what you agreed to, and only after you give consent.

    Who or what does it
    You grant it. The system enforces the limit.
    What goes wrong without it
    One lost device or misbehaving agent could reach far more than it needs.
  5. Humans approve sensitive actions

    Gated actions stop and wait for the workspace owner. An agent cannot approve its own request.

    Who or what does it
    The workspace owner decides.
    What goes wrong without it
    An AI mistake goes straight through, with no person to catch it.
  6. Database-level rules

    Postgres is the database. Row-level security means the database itself checks who may see each record, so the rule still applies even if the app gets something wrong.

    Who or what does it
    The database enforces it on every request.
    What goes wrong without it
    One bug in the app could show one customer another customer’s records.
  7. Audit trails

    A record of what ran, who approved it and what changed.

    Who or what does it
    The system writes it as work happens.
    What goes wrong without it
    When something goes wrong, nobody can tell what happened or why.
  8. Private networking

    Tailscale links approved devices over a private connection. Zero trust means no device is trusted just for being on the network. We use this for remote access instead of opening ports, which are doors a machine leaves open to the internet.

    Who or what does it
    We set it up. You decide which devices belong.
    What goes wrong without it
    An open door on the internet is one more place for outsiders to try their luck.

Where your data lives

Follow each kind of data through InfiniBot Cloud

Everything starts the same way: you, then sign-in, then your workspace’s gateway. After that, each kind of data goes to its own place. Pick a kind to light up its path, or read all four below the map. Another workspace sits outside yours, with no shared runtime.

Workspace dataYour files and running work stay on your workspace’s own processes and storage. So a problem in another workspace does not reach yours.

Keys and secretsKeys go to an encrypted store and only ever appear masked. So a glance at a screen or a stolen file does not give them away.

Database recordsRecords go to the Postgres database, which checks who may see each row itself. So the rule holds even if the app slips.

ActivityEach action adds a line to the audit trail: what ran, who approved it and what changed. So you can always look back.

Plain speaking

What we do not claim

  • We do not hold SOC 2, ISO 27001 or similar certifications, and we do not claim HIPAA compliance.
  • We do not publish client names on security work without permission.
  • No system is perfectly secure. We reduce risk, make failures visible and keep recovery paths ready.

If you have a compliance requirement, raise it on the first call. We will tell you plainly what we can support.

Want a look at your exposure?

Start with a Systems Assessment. We review your access, secrets and data, and give you a prioritized list.