Key facts
| Access control | Scoped API keys with rotation; enterprise SSO and RBAC options |
| Tenancy | Separate keys and retrieval indexes per client tenant |
| Deployment | Cloud, VPC, on-prem or air-gapped per client requirement |
| Auditability | Request, model and response logging for client service reviews |
| Data grounding | Embeddings and RAG are live for runbooks and client documentation |
| Models | 30+ models behind one OpenAI-compatible API |
| Pricing model | Flat monthly self-serve plans; no per-token billing on self-serve |
| Endpoint roadmap | Files, batch, assistants and fine-tuning are coming soon |
TL;DR
- Isolate tenants with separate keys and indexes before adding features.
- Document subprocessors and data flows in a reusable client pack.
- Scope technician access per client and per ticket, with expiry.
- Log requests, tenants and reviewers so service disputes are traceable.
- Keep one named owner per client AI service and its exceptions.
How it works, step by step
- Inventory AI use cases across helpdesk, monitoring, onboarding and reporting.
- Classify data per client and record contractual handling restrictions.
- Choose deployment per client: shared cloud, dedicated VPC, on-prem or air-gapped.
- Issue per-tenant keys with rotation and central inventory.
- Define log fields and retention: request ID, model, sources, output, tenant, technician.
- Approve a model allow-list and document it in client assurance packs.
- Review consent and disclosure before extending AI to a new service line.
Try it yourself
Open the AI API key security checklist →
Multi-tenant client data
An MSP processes data from many client environments, each with its own contract, regulator and tolerance for AI. Classification should record the client, the data class and where processing is permitted. Public vendor documentation and internal runbooks are low risk; client tickets, logs, source code and production configuration are not.
Write the isolation rule down before automating anything, because one shared retrieval index across tenants is the most common and most damaging MSP design mistake.
Keys, tenants and least privilege
Issue a distinct API key per client tenant and per application, store them in a secrets manager, and rotate on a schedule. Keep retrieval indexes and conversation history separated per tenant so one client cannot retrieve another's content. Enterprise SSO and RBAC options keep console access aligned with technician changes, and third-line or subcontractor access should be scoped and time-bound.
Never place client identifiers or credentials in prompts; retrieve only the fields a task requires.
Residency, retention and client audit trails
Decide where processing happens and how long records live, per client. Region selection covers many residency needs; VPC, on-prem and air-gapped deployment covers clients who require data to stay in their environment or country. Retention is your policy applied to prompts, outputs, logs and retrieval indexes, and it may differ per contract.
Log enough to reconstruct a service action: request ID, model and version, retrieved sources, output, tenant and technician. See AI audit logs for a schema.
Model governance and white-label delivery
Keep an approved model list with evaluation evidence and re-test when versions change. Ground answers in current client documentation with citations so technicians can verify before acting. A named service owner should remain accountable per client, and the checklist should document how subprocessors are disclosed and how consent is captured. White-label delivery works because the API is OpenAI-compatible and keys can be scoped per tenant; keep it that way as you package new services.
Honest comparison
| Control area | Plugsky capability | Common gap | Owner |
|---|---|---|---|
| Tenancy | Separate keys and indexes per client tenant | One shared index for all clients | Architecture |
| Identity | Scoped keys, rotation, enterprise SSO and RBAC options | Shared technician credentials | Security engineering |
| Data boundary | Shared cloud, dedicated VPC, on-prem or air-gapped | Client rules undocumented | Service delivery |
| Retention | Configurable logging per client policy | Single blanket retention | Compliance |
| Audit trail | Request, tenant and technician logging | Actions not traceable | Internal audit |
| Disclosure | Model and endpoint docs for client assurance packs | Subprocessors undisclosed | Account management |
Frequently asked questions
Does Plugsky make us compliant?
No. Compliance is your program. Plugsky provides deployable controls - per-tenant keys, separated indexes, private deployment, logging - that you document in client assurance materials.
How do we keep client tenants isolated?
Use a distinct key and retrieval index per tenant, and ensure no shared workflow can query across tenants. Private deployment is available per client where contracts require it.
What should we log?
Request IDs, model names and versions, retrieved sources, outputs, the tenant and the technician, retained under the applicable contract so service actions can be reconstructed.
Can we white-label the service?
Yes. The API is OpenAI-compatible and keys are scoped per tenant, so you can deliver AI services under your own brand and contract terms.
How do we handle technician turnover?
Manage keys per tenant in a secrets manager, revoke on role changes, and connect SSO and RBAC options so console access follows your joiner-mover-leaver process.
What do clients ask in assurance reviews?
They ask which subprocessors process their data and where. Keep the model list and deployment architecture current so you can answer without a bespoke review each time.
Where should a pilot start?
Pilot on public vendor documentation and internal runbooks with the free plan, then extend to client-specific content with per-tenant keys and private deployment.