Provenance Claims prepares SR&ED tax credit claims. Doing that means handling two sensitive things: your engineering records and your payroll figures. This page explains exactly what happens to both. It is written to be read, not to be survived.
There are no analytics, no cookies, no tracking pixels, and no third-party scripts. The refund calculator runs entirely in your browser; the numbers you enter are never transmitted anywhere, and the site's content security policy blocks outbound connections entirely.
Two things we do not control: this site is hosted on Netlify and our email is Google Workspace, both in the United States. Netlify logs visitor IP addresses and Google holds the content of any message you send us.
We do not need, and do not ask for, customer data held inside your systems, production databases, or personal information about your users.
Analysis is local-first. Your repository is processed on infrastructure we control, and the bulk of the work, including all initial classification, runs on a local model with no external network calls.
Drafting the claim narrative uses a frontier AI model operated by Anthropic, in the United States. Exactly these fields are sent, one row per piece of evidence: a ten-character commit identifier, the date, the commit subject line, a pseudonym such as P1 in place of the author's name, and a short classification label. Real names are not sent. Source files are not sent. Pull request bodies, file paths and salary figures are not sent.
Because this crosses the border, the information becomes subject to the laws of the United States and may be accessible to US law enforcement or authorities under those laws. This step happens only with your written consent, recorded in the engagement letter, and you may withdraw that consent at any time, in which case drafting runs entirely locally and we tell you that review will take longer.
We do not sell your data, and we do not permit it to be used to train anyone's models.
We retain a claim defense file: the evidence ledger, the drafted narratives, and the citations linking them. If the CRA reviews your claim, this file is what defends it, so it is kept for the CRA's record retention period of six years from the end of the tax year the claim relates to.
Working copies of repository data, including any cached pull request text, are deleted once the claim is filed. Deletion is performed by a dedicated command in our tooling and produces a dated deletion receipt that we will send you on request. You can ask us to delete the defense file earlier and we will, with the caveat in writing that doing so weakens your position if the claim is later reviewed.
Nobody else. We do not use client work in marketing without asking first, and the sample materials we show prospective clients are built from public open-source code, not from client engagements.
British Columbia's Personal Information Protection Act (PIPA) governs how we handle personal information, because we are a BC organization. Canada's federal PIPEDA applies additionally to information that crosses a provincial or national border, which for us means the drafting step described above.
Our privacy officer is the Director, reachable at privacy@provenanceclaims.ca.
Under BC PIPA you can ask what personal information we hold about you (section 23), ask for corrections (section 24), and withdraw consent for a particular use. Email the privacy officer above and you will get a real answer from a person within 30 business days, which is the period the Act allows.
If you are an employee of a client company and your name, salary or work history appears in a claim we prepared, those rights are yours directly. You do not have to go through your employer to exercise them.
If you are not satisfied with how we handle a request or a complaint, you can contact the Office of the Information and Privacy Commissioner for British Columbia at oipc.bc.ca.
Provenance Claims · Vancouver, British Columbia · ryan@provenanceclaims.ca