Customer-Controlled AI Inference

Privacy Policy

Effective Date:

1. Data Handling Overview

Our platform handles highly sensitive intellectual property (your source code). We design PRInspector to process only what a review needs and keep as little as possible. PRInspector is currently in pre-launch development; this policy describes how the product is designed to handle data and will be updated before launch if anything changes.

2. How Code Is Handled During a Review

We do not clone your repositories or store your source files. When a pull request is opened or updated, our worker fetches the pull request's changed files and their .diff from GitHub and processes them for that review. Diffs are not written to our database.

What we do store are the review results: the findings (file path, line number, the issue described and any suggested fix, which can include short code excerpts), the quality score and summary, and pull request metadata (see section 4).

3. AI Inference and Model Training

Our Commitment

We do not use your source code, architecture, or PR metadata to train AI models.

Your inference configuration determines how code is processed. Diffnix supports self-hosted inference for customers requiring private deployments: models are served through Ollama on infrastructure you (or we, on your behalf) run, and code is sent only to that model server. Optional cloud AI providers are planned for teams that choose hosted inference. If you enable one, the code under review will be sent to that provider for processing under its own terms and data-retention policy, which we will identify before you turn it on.

4. Metadata Collection

To operate our dashboard and routing, we store aggregate metadata:

  • GitHub OAuth Credentials Used to generate access tokens required to post inline comments on PRs.
  • Repository Identifiers Repository names and IDs required for webhook routing.
  • Pull Request Metadata PR ID, author, timestamps, and state to populate analytics and history.
  • Review Results Findings, scores and summaries for each reviewed pull request.

5. Website Visitors, Sign-ups & Surveys

When you use our website — separately from the product — we collect the information below. We never sell it, and we don't share it with advertisers.

Early-access sign-ups (Get Started)

When you join the early-access list, we store:

  • Details you enter Your name and email address, and optionally your company, role, team size and what you'd like Diffnix to help with.
  • How you found us The referring website, the first page you visited, the page you signed up from, and any campaign (UTM) tags in the link.

We use this to contact you about early access, to prioritise which teams we onboard, and to understand which channels bring people to Diffnix.

Website survey

Our short website survey is anonymous and optional. It doesn't ask for your name or email and isn't linked to an early-access sign-up. We store:

  • Your answers Including partial answers, so you can pick up where you left off, and whether you completed or postponed the survey.

We use survey answers only in aggregate, to decide what to build.

Contact form and investor inquiries

If you message us through our contact page, we store your name, email address, the optional company and topic you choose, and your message. If you contact us through our investors page, we store the contact details, firm information, investment profile and message you submit. For both, we also store approximate location and device details derived from your request, and use everything only to respond to and follow up on your message.

Cookies and browser storage

  • Session storage Remembers the page and link that first brought you to our site during your visit, so a sign-up can be attributed correctly. It is cleared when you close the tab.
  • Analytics If we enable Google Analytics, it sets its own cookies to measure website traffic. We don't use advertising cookies.

Retention and your choices

We keep this information for as long as we need it for the purposes above, or until you ask us to delete it. You can ask to access, correct or delete your data at any time using the contact address below.

6. Third-Party Integrations

We use a limited set of third-party services only where needed to run the product:

  • GitHub To receive pull request events, read the changes under review, and post review comments, using only the permissions you grant our GitHub App.
  • Cloud AI providers (planned, optional) Only if your team chooses hosted inference. Self-hosted inference does not use one.
  • Payment processor (when paid plans launch) To process payments. We will not store your full card details.

All data exchanged with these services is encrypted in transit using TLS.

7. Government Requests

Because we don't store your source files, we have no repository code to hand over. We will only disclose the data we do hold (such as account details, metadata and review results) when required by a legally binding request, and where permitted we will notify you first.

8. Policy Updates

We may update this Privacy Policy to reflect new features or compliance requirements. All active users will receive email notification prior to any substantive changes.

Contact Us

For privacy inquiries or GDPR/CCPA data deletion requests, contact: [email protected].