Key facts
| API compatibility | OpenAI-compatible chat, embeddings and function calling |
| Use cases | Citizen service Q&A, policy search, form and document processing, multilingual support |
| Sovereignty | Region selection plus VPC, on-prem and air-gapped deployment options |
| Access control | API keys, RBAC, SSO and audit logs for attributable requests |
| Models | 30+ models, including multilingual options for citizen services |
| Pricing model | Flat monthly self-serve plans with unlimited fair-use usage |
| Free tier | plugsky-micro and plugsky-lite on the free plan, no card required |
| Product status | Chat, streaming, JSON mode, function calling, embeddings, RAG and agents are live |
TL;DR
- Use retrieval over published policy so answers cite the official source.
- Keep citizens informed without exposing personal case data to prompts.
- Region selection, audit logs and on-prem options support sovereignty mandates.
- Multilingual models extend reach without a separate translation stack.
- Prototype free, then procure with a documented evaluation.
How it works, step by step
- Choose a public-facing workflow such as policy Q&A or a service finder that only touches published material.
- Index the official documents that define the correct answer and require citations in every response.
- Keep case management and personal identifiers out of prompts; integrate by reference, not by copying records.
- Configure region, retention, RBAC and audit logging to match the mandate before any pilot goes live.
- Run a documented evaluation with a labelled question set and a human escalation path for unresolved cases.
- Move to on-prem or air-gapped deployment where the mandate requires it, without rewriting the application.
Try it yourself
Open the sovereign AI readiness score →
Where an AI API helps public services
Public sector work is document-defined, which makes grounded retrieval the right pattern:
- Citizen service Q&A: answer questions about eligibility, process and documents with a link to the official source.
- Policy search: let staff find the current rule across legislation, circulars and internal guidance.
- Form and document processing: extract structured fields from submissions for an officer to confirm.
- Multilingual service: respond in the citizen's language using a multilingual model rather than a separate pipeline.
Publish an escalation path, because confidence and clarity matter more than speed here.
Sovereignty, access and accountability
Three requirements recur in public sector procurement: data stays in an approved jurisdiction, access is attributable, and the system can be audited. Plugsky provides region selection, scoped API keys, RBAC and SSO for staff, and audit logs for every request. Deployment can be upgraded to a private VPC, on-prem or air-gapped environment where the mandate requires it. As with any deployment, the controls are platform features; your authority still defines the legal basis, retention rules and records schedule that govern the workflow.
Deployment and procurement path
Start with a low-risk, public-document pilot on the free plan using plugsky-micro or plugsky-lite. Use the 14-day full-access trial to benchmark a frontier model on the same question set, documenting accuracy, citation support and escalation rate. That evidence is what procurement teams need before a wider rollout. Because the API is OpenAI-compatible, the same application then moves to a paid plan, a private VPC, on-prem hardware or an air-gapped environment, which keeps the evaluation valid across deployment choices.
Limits to plan for
An AI API does not adjudicate cases, make determinations or replace statutory process. Keep it to retrieval, drafting and extraction, with a named officer accountable for decisions. Expect some questions to fall outside the indexed corpus, and design the fallback before launch: a human handoff, a form, or a phone number. Review prompts and documents whenever policy changes, and log the model version behind each response so the record stays defensible. Where language quality is uneven, test the multilingual models on your own corpus rather than assuming coverage.
Honest comparison
| Capability | Plugsky | Typical per-token API | Building in-house |
|---|---|---|---|
| API compatibility | OpenAI-compatible chat, embeddings and tools | Usually compatible | Full rewrite |
| Sovereignty | Region selection, VPC, on-prem and air-gapped | Often limited regions | You control fully |
| Audit and access | API keys, RBAC, SSO and audit logs | Varies by provider | You build all of it |
| Pricing | Flat monthly self-serve, fair-use usage | Per-token, harder to budget | GPU plus operations cost |
| Model choice | 30+ models behind one API | Varies by provider | You host every model |
Frequently asked questions
Can we keep our existing OpenAI SDK code?
Yes. Plugsky exposes an OpenAI-compatible API, so the change is a base URL and model name while your SDK and integration code stay as they are.
Is there a free plan?
Yes. The free plan includes plugsky-micro and plugsky-lite with no credit card, suitable for a public-document pilot.
Can data stay in our jurisdiction?
Region selection and sovereign deployment options are available, including VPC, on-prem and air-gapped environments. Confirm the specifics for your mandate with the team.
Do you support air-gapped deployment?
Yes, air-gapped deployment is available for enterprise and government setups where no traffic may leave the environment.
How does pricing work?
Self-serve plans are flat monthly with unlimited fair-use usage. See the live pricing page for current plans and enterprise terms.
Which endpoints are live today?
Chat, streaming, JSON mode, function calling, embeddings, RAG and agents are live. Audio, images, moderation, files, batch, fine-tuning, assistants and responses are coming soon.
Does the AI make decisions on cases?
No. Keep it to retrieval and drafting with a named officer accountable for the decision, and log the model version and sources behind each response.