Organize
Build the working library.
Group policy PDFs by publication family, search extracted text, and keep the current source set easy to navigate.
A Graywheel product
PolicyVector
Policy study, built around your source material.
A local-first workspace for policy-heavy study: organize PDFs, protect confidential material, and turn approved sources into grounded review on your own machine.
- Source-bound study
- Local document library
- Desktop and Reader workflows
From policy files to focused practice.
PolicyVector gives serious study work a clear path: build a reliable library, work directly from the source, and create the material you need to prepare.
Understand
Stay tied to the document.
Review source pages, build structured outlines, and ask questions against the policy material already in your library.
Practice
Study in the format that works.
Move into flashcards, multiple-choice tests, readback, and Oral Board practice without losing the source context.
One library. Every stage of preparation.
The product is organized around practical study jobs, not a pile of disconnected AI features.
Document library
Organize PDFs by publication, inspect the source, search across local text, and preserve frozen policy views for a study period.
Grounded review
Flashcards, tests, outlines, readback, and Oral Board practice stay tied to approved sources.
Reader handoff
Package selected study material into encrypted bundles for the native PolicyVector Reader on iOS and Android.
Flexible AI
Use supported local models for private workflows or bring your own cloud provider when the material is appropriate.
Use AI without losing control of the material.
Local-first AI keeps sensitive work on your computer. For non-confidential or approved redacted text, OpenAI's GPT models can act as a study partner.
Private, on-device
Ollama runs Qwen on your computer.
Ollama is the local engine that lets PolicyVector run AI without sending the prompt to a cloud service. Qwen3 is the language model doing the reading and writing work.
PolicyVector offers 4B, 8B, and 14B Qwen3 options and recommends one that fits your computer. The smaller model is lighter and faster; larger models can handle more demanding work but need more memory.
- Keep confidential prompts and source text on the machine.
- Use Local AI for supported policy questions and study-material generation.
Non-confidential text
GPT can be the second set of eyes.
PolicyVector connects to OpenAI's GPT models through the API—the same model family behind ChatGPT. Use it to explain, summarize, draft questions, and challenge your recall when the text is non-confidential or has been approved and redacted.
You provide your own OpenAI API key. A ChatGPT subscription is separate from API billing, so the app makes that setup and cost boundary clear before you connect it.
- Turn suitable text into a conversational study session.
- Keep cloud assistance optional and separate from confidential local work.
Clear boundary: PolicyVector does not send your library to OpenAI by default. Cloud assistance begins when you choose it for suitable material; local documents and study artifacts otherwise remain in the workspace.
Your working library stays yours.
Local-first means policy libraries and study artifacts stay on your device unless you choose to export. It is a product boundary, not a marketing flag.
Local-first
Documents, extracted text, study material, and progress are designed to live in the local workspace.
Deliberate sharing
Exports and connected services begin with a user action, so moving material out of the workspace is an explicit choice.
Built for serious preparation
Built for less generic productivity theater and more source-bound study workflow for policy-heavy work.
Turn the source material into a study system.
PolicyVector brings the policy library, grounded study tools, and Reader workflow into one focused product.