policyratio
Policy citation specialists

We turn your policy library into a citation layer.

Your people ask in plain language and get the exact governing passage — with its section and a link to source — or an honest “not covered.” Every policy answer, traced to the sentence that governs it.

For organizations where thousands of frontline decisions must trace to a written rule.

Open lookup Talk to us about your policy set
What we specialize in

We do one thing: citation.

Not search, which hands you forty documents. Not a chatbot, which hands you a confident paraphrase. We return the published sentence that governs the question, and we make it traceable. Organizations bring us the policy set their operation runs on — benefits manuals, clinical protocols, procurement rules — and every answer becomes citable.

How we help

Who it’s for

How it works

Policy selection

Loaded corpora

The policy sets this engine can answer from, and where each one was published. A jurisdiction stays dark until its source has been ingested and every passage checked against it.

Policy sets loaded into policyratio, grouped by jurisdiction
Jurisdiction Policy set Source Status
CA — Canada Government of Alberta — Income and Employment Supports Open the lookup Published manual. Checked passage by passage against the publisher’s own pages. Loaded
No jurisdiction yet Your policy set Any manual your operation runs on. Nothing is answerable until it is ingested and verified. Not loaded

What a procurement review asks

Where do the words in an answer come from?
From the published document itself. The answer is a sentence from your policy, not a sentence about it, and every quote is machine-checked to be an exact substring of the source we ingested. We retrieve and cite; we are not an official publication of, and are not endorsed by, any publisher whose policy we cite.
Does a model write the answer?
No model writes, edits, or extends a quoted line. Ranking layers only choose among verbatim passages, and a quote that does not match its source is dropped rather than shown. Optional plain-language layers are separate, and they ship off.
What happens to what staff type in?
Direct identifiers — SIN, health number, phone, email, postal code, account IDs — are stripped in the browser and again on the server before anything is stored. Free text can still contain a name, so staff enter the policy question, not the person. Logs keep the scrubbed question, section identifiers, and quote hashes.
Who is liable for a decision?
The staff member. policyratio returns the governing rule and how the stated facts line up against it — decision support, not the official determination. The determination, and the accountability, stays with your staff.
What happens when nothing covers the question?
We say so and escalate. Below the relevance threshold the result is no matching policy — not the nearest thing, not a confident guess. A domain stays dark until it has been ingested and verified, so a gap in coverage looks like a gap instead of an answer.