Privacy

Your visit

The site keeps a first-party log of visits to operate and improve the showcase — no third-party trackers, no cross-site cookies, and no ad-tech. If you use the contact form, your message is relayed straight to me; the contact form and the “watch it work” simulator are protected from abuse by Cloudflare Turnstile. There is no public sign-in, so you never grant the agent any access by browsing, running the simulator, or sending a message.

Some links to the site carry a short, human-readable label of where they came from — for example ?from=my_linkedin, visible right in your address bar — so I can tell which channel a visit came through. It’s recorded with the visit below and visible only to me; a private one-to-one outreach email uses a short unique code instead. There are no other trackers.

When you run the “watch it work” simulator, the run is saved — that’s how you get a shareable result, and how the abuse limits work. It keeps the company, role, and job description you enter, a redacted activity trace, the run’s cost and runtime, and your IP address; it’s stored only in this service’s private database, visible only to me, and deleted after a short retention window. The public “recent runs” list shows aggregate cost and runtime only — never the text you typed.

The fuller, plain-English version of this lives in the About page.

Google account data

To run my job search, the agent connects to my own Google account using OAuth. It requests these permissions:

• gmail.modify — read, organize, and send email: spotting recruiter and job-related messages, labeling them, and drafting and sending outreach.
• calendar.events.owned — create and manage events on my own calendars, to schedule interviews.
• drive.file — create and manage only the Drive files the agent itself makes (for example, interview-prep documents) — never anything else in my Drive.
• basic profile (openid, email, profile) — to sign the agent in to the right account.

How that data is handled:

• It is used only to operate features of this service for me — triaging job-related mail, researching roles, drafting and sending outreach, scheduling interviews, and preparing interview-prep documents.
• To research and draft text, message content may be sent to an AI model (Anthropic’s Claude, via the Portkey gateway) to return a result. Under those providers’ API terms, the content is not retained to train their models.
• It is stored only in this service’s own private database, on infrastructure I control.
• It is never sold, never shared for advertising, and never used to build advertising profiles.
• No human other than me reads it.

This service’s use of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.

Retention & deletion

The agent keeps only what it needs to act on my job search (for example, the application-pipeline records it builds). I can revoke its access to my Google account at any time from my Google Account’s security settings, which immediately stops all further access.

Changes & contact

I may update this policy; the date above reflects the latest version. See also the Terms of Service. Questions can go through the contact form.