Services · Security & data
Where your data lives. Who can touch it. How actions are controlled.
Security begins with the scope: what data a system needs, which actions it may take, who can access it, and how the business can review what happened. Ridgeway documents those boundaries before production access is granted.
01 / The posture
Eight controls. All real.
These controls form the Ridgeway baseline. The exact configuration is adapted to the systems, data, vendors, and actions included in each engagement.
Isolation
Per-client environments
Managed clients can run on isolated infrastructure and separate provider projects so access, usage, and operational boundaries remain client-specific.
Credentials
Vaulted per client
Access lives in a per-client 1Password vault. One client’s keys never sit next to another’s, and revoking access is one vault, not a scavenger hunt.
Control
Human-in-the-loop gates
Customer-facing, high-risk, or ambiguous actions wait for a human. Which steps may run, draft, or need approval is written down per workflow.
Containment
Allowlists + sandboxing
Agents can only call the tools they’re explicitly given, inside explicit limits. Not “be careful” — can’t.
Evidence
Full audit logging
Configured workflow actions are written to a log: what ran, when, and against what. The record supports review and investigation.
Injection posture
Inbound content is untrusted
Agents that read email and documents treat them as untrusted input: read-tools stay separate from act-tools, and outbound actions sit behind approval gates. That doesn’t make injection impossible — it makes the blast radius small.
Model training
Commercial AI services
When AI providers are used, client work runs through commercial services selected and configured for business use. Provider terms, retention options, and data-handling settings are reviewed for the proposed workflow.
Storage
Storage-agnostic by design
Your files stay in storage you own — Google Drive, SharePoint, Dropbox, or on-prem. We sync to the workflow. We don’t migrate your business into ours.
Security is not one badge or one setting. It is the combination of access, isolation, action limits, logging, and ownership applied to the actual work.
02 / Scope-specific controls
Protection follows the data and the action.
Data boundaries
Start with the minimum access required.
We identify the data classes involved, the systems that hold them, the access level the work requires, and how long temporary working data should remain available.
Access begins as narrowly as practical and expands only when the approved scope requires it.
Action boundaries
Routine work and sensitive work are treated differently.
Read-only research, internal drafting, customer communication, payments, deletion, and credential changes do not carry the same risk. Each workflow defines which steps may run, which should draft, and which require a person.
Requirements involving regulated or contractually restricted data are reviewed before the work is accepted and reflected in the written scope.
03 / The ownership seam
Yours, ours, and the exit.
You own your data, results, and customer-facing accounts — exportable any day. Ridgeway owns the engine. Leaving gets you a portable export, a documented snapshot of what was built, and confirmation that access was revoked.
Read next
Data handling
The plain-language page on what we may access, where data lives, retention, and the exit package. It keeps its attorney-review draft banner — read it before you grant access.
Scope it
Ask before access is granted
We can map the proposed systems, data classes, access level, and exit path before any workflow is activated. Access starts read-only; write access is separately scoped.
Fit the work
Review capabilities and scope
See the digital projects and managed systems Ridgeway can deliver, then bring the specific data and access requirements to discovery.
Start here
Ask the security questions on call one.
Bring your IT person if you have one. We’ll walk the proposed systems, data, access, actions, and exit path before you grant anything.