How Cairn Handles Your Data

This page is the plain-English version of our Privacy Policy. It walks through exactly what happens to your data at each step. If something here conflicts with the Privacy Policy, the Privacy Policy controls — but we try hard to keep them in sync.

When you sign up

You authenticate through Clerk, our identity provider. Clerk gives us an email address and a user identifier. If you use GitHub to sign in, Clerk also passes along a GitHub identifier. We store those on a users record so every piece of your content can be scoped to you.

We do not ask for or store passwords — Clerk handles that.

When you upload a resume or other material

We store the original file you uploaded. We also extract structured data from it (work history, skills, education) and store that separately so other parts of the app can reason about it. Both the original and the extracted version stay on our database and belong to your account only.

When you add an opportunity (job posting)

You can paste a URL or a block of text. We may fetch the public page at that URL to enrich what we know about the role. We may also run a short AI-driven research pass on the company. The source text, our parsed fields (role, company, location, comp, required skills), and any research we generated all live under your user ID.

When you generate prep materials, a resume, a cover letter, or an interview guide

For every generation, we build a prompt from your profile and the opportunity in question, send it to Claude (Anthropic's API), and save the result. The inputs we send typically include:

• The relevant parts of your profile (skills, work history, positioning).
• The opportunity the output is about (role, company, JD text).
• The specific question or command you gave the coach.

Anthropic processes that request under their terms. The output is saved to your account so you can return to it.

When you chat with the coach

Messages you send are stored so the conversation is continuous. To answer you, we send the relevant context to Claude, which returns a response. The response is stored alongside your message.

What we log for cost and debugging

Every call to Claude is recorded in our usage_logs table with: the model used, how many input and output tokens the call used, our estimated cost in U.S. dollars, and a timestamp. This is tied to your user ID so we can debug issues and so you can see your own usage on the Profile page. We do not store the prompt contents in these usage records — only the accounting metadata.

What we do not do

• We do not sell your data.
• We do not use your content to train general-purpose AI models.
• We do not hand your content to advertisers, data brokers, or analytics vendors.
• We do not run behavioral trackers, ad pixels, or cross-site tracking.

Deletion

You can delete your account from the profile page. Deletion removes your uploaded materials, generated content, opportunities, and coach messages from our active database within 30 days. Aggregated usage records (token counts and cost, without your content) may be retained longer for financial reconciliation. Backups may take additional time to fully purge.

Third parties in the loop

• Clerk — handles sign-in and session cookies.
• Anthropic (Claude) — runs the AI that produces the generated content.
• Railway — hosts the app and the PostgreSQL database where your content lives.

These are the only providers that touch your data as part of normal operation.

Questions

If anything here is unclear, or you want to see exactly what we have about you, email velayudamarjuna@gmail.com.