Key facts
| API compatibility | OpenAI-compatible chat, embeddings and function calling |
| Use cases | Permit and HSE document Q&A, maintenance log summaries, field knowledge search |
| Data controls | API keys, RBAC, SSO, audit logs and region selection |
| Deployment | Plugsky cloud, your VPC, on-prem and air-gapped options |
| Models | 30+ models, from free aliases to frontier reasoning |
| 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
- Ground field and HSE answers in your own documents instead of generic model memory.
- Keep your OpenAI SDK code; change the base URL and model name.
- Run in the cloud, your VPC, on-prem or air-gapped for remote and regulated sites.
- Start on the free plan with plugsky-micro and plugsky-lite, no card required.
- Keep OT and control-system data out of prompts; send documents and logs instead.
How it works, step by step
- Pick one bounded workflow, such as HSE incident report drafting or permit Q&A.
- Classify the data and keep OT, seismic and personal data out unless there is a documented need and a lawful basis.
- Ground answers with embeddings and RAG over the document set you are allowed to send.
- Build against the OpenAI-compatible endpoint and keep prompts, tools and evals in version control.
- Choose a deployment model (cloud, VPC, on-prem or air-gapped) and the region your policy requires.
- Pilot on a small labelled set, measure accuracy and review time, then expand to the next workflow.
Try it yourself
Open the private LLM deployment estimator →
Where an AI API fits in oil and gas
Oil and gas generates dense technical text: permits, leases, HSE reports, rotating-equipment logs and geoscience studies. The highest-value first use cases are retrieval-heavy rather than generative:
- Document Q&A: ask a question across permits, contracts and standards, with citations back to the source page.
- Report drafting: turn raw incident notes into a structured HSE draft that a supervisor reviews and signs.
- Maintenance knowledge: surface the relevant procedure or prior work order for a tag or asset id.
- Extraction: pull structured fields from supplier invoices, inspection sheets and lab reports.
Each keeps a human in the approval path and sends documents rather than control signals.
Security, data handling and OT boundaries
Treat the AI layer as an IT system that must not reach into OT. Practical controls:
- Scope API keys per application, rotate them, and use RBAC and SSO for people.
- Turn on audit logs so every request has an owner, a time and a model.
- Select the region where prompts and logs are processed, and check retention settings against your policy.
- Exclude well-control, SCADA and personal data from prompts; if a workflow truly needs it, escalate to a specialist deployment with a documented basis.
Plugsky supplies the controls; your security team still owns classification and approval.
Deployment options for remote and regulated sites
Offshore and remote sites often have constrained connectivity, and some assets are formally restricted. A single OpenAI-compatible interface keeps the application portable across deployment modes: Plugsky's cloud, your own VPC, on-prem racks, or air-gapped environments where no traffic leaves the site. Because the API shape does not change, the same retrieval code, prompts and tests move with it, which lowers the cost of starting in the cloud and moving to a stricter environment later.
From pilot to production
Start free with plugsky-micro or plugsky-lite to prove retrieval quality, then use the 14-day full-access trial for frontier models on the same workload. Score a labelled set of real questions (answer accuracy, citation support, review minutes saved) and compare two or three models before committing. Pin model versions for controlled workflows, monitor for drift as documents change, and add a second workflow only after the first is stable. Self-serve plans are flat monthly with fair-use usage; enterprise options cover private and air-gapped deployments.
Honest comparison
| Capability | Plugsky | Typical per-token API | Building in-house |
|---|---|---|---|
| API compatibility | OpenAI-compatible chat, embeddings and tools | Usually compatible | Full rewrite |
| Deployment | Cloud, VPC, on-prem and air-gapped | Mostly cloud-only | You operate GPUs and serving |
| Data residency | Region selection and sovereign options | Limited regions | You control fully |
| Pricing | Flat monthly self-serve, fair-use usage | Per-token, harder to forecast | 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 you change the base URL and model name and keep the SDK and prompt structure you already have.
Is there a free plan?
Yes. The free plan includes two free models, plugsky-micro and plugsky-lite, and does not require a credit card.
Can we deploy air-gapped?
Yes, air-gapped deployment is available for enterprise setups where no traffic may leave the site. Check the docs and talk to the team about the specifics.
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 options.
Which endpoints are live today?
Chat completions, streaming, JSON mode, function calling, embeddings, RAG and agents are live. Audio, images, moderation, files, batch, fine-tuning, assistants and the responses API are coming soon.
Do we have to send OT or SCADA data?
No. Keep control-system data out of prompts and retrieve from documents and logs instead. Escalate only if a workflow has a documented need and basis.
How should we evaluate quality?
Build a small labelled set of real questions, compare two or three models, track citation support and reviewer time saved, then pin the model version you approve.