How it works · logins and sessions
An AI browser agent that works in your own logins
An AI browser agent can use your logins without ever seeing your password: you sign in yourself in the agent's browser, and it works inside that signed-in session the way you would. Vassal Agent keeps those sessions on your own Mac, hands the window back to you for passwords, two-factor codes and CAPTCHAs, and asks before it sends, posts, pays or deletes anything in any account.
Why does an agent need your logins at all?
Most useful work sits behind a sign-in: your CRM, your ATS, LinkedIn, your email, supplier portals, job boards with saved searches. An AI that can only read public pages cannot do that work. The question is how to let it in without handing over the keys.
How does Vassal Agent use a login without the password?
- You sign in, yourself. Open the site in the app's browser and sign in the normal way, the first time a job needs it. If a job needs a site you are not signed in to, it opens the page and asks you to sign in.
- The session stays on your Mac. Like any browser, the app remembers that you are signed in. The cookies and saved passwords live in the app's profile on your computer, and Vassal Agent never receives them.
- The agent works in that session. It opens pages, reads them and fills in forms as you, at a human pace, with a live panel showing every page it opens.
- You stay in charge of the risky steps. Passwords, two-factor codes and CAPTCHAs go to you. Send, post, submit, pay and delete wait for your Allow.
Why a separate browser for the agent?
If an agent lives in your everyday browser, it can reach everything signed in there: personal email, banking, shopping, social accounts. Vassal Agent has its own browser, so it only sees the work sites you chose to sign in to inside the app. That is a smaller, clearer boundary than "everything I have ever logged in to".
What stops it doing something you did not want?
- An approval stop enforced by the app. Before anything that sends, posts, submits, pays, books or deletes, the app asks. The check is outside the AI, so a page that tries to trick the model (prompt injection) still cannot get past it.
- Rules per site. Never / Ask me first / Always allow, for a button, typing, opening a site or anything, on one site or all of them.
- An audit log. Every action is recorded before it runs, whether it was allowed or refused.
More detail is on Is it safe to let an AI use my browser?
What can it do once it is signed in?
Sweep job boards and directories you have accounts on, build lead or candidate lists from pages you can see, enter records in your CRM or ATS, draft replies in your webmail for you to send, fill in portal forms, and watch pages for changes. See recruiting, lead generation and form filling examples.
How do I try it?
Download Vassal Agent for Mac and use it free for 7 days with no card. After that it is $499 a month or $4,990 a year; see pricing.
Questions
Do I have to give the AI my passwords?
No. You sign in yourself, once, in the app's browser. The agent works in the session that sign-in creates. Passwords are never recorded in workflows, and the AI does not need them.
What happens with two-factor authentication?
When a site asks for a code, a passkey or a CAPTCHA, the agent stops and hands the window to you. You finish the step and it carries on.
Where are my logins stored?
On your Mac, in the app's own browser profile, the same way a normal browser keeps them. Vassal Agent does not receive your cookies, passwords or browsing.
Can I keep it out of certain accounts?
Yes. Rules in plain words set what it may do on each site: "Never touch my bank", "Ask me before anything on linkedin.com", "Never delete". Deny rules always win over allow rules.
Is it against a site's terms to use an agent in my account?
It depends on the site. Some sites restrict automated activity. Vassal Agent works at a human pace inside your own accounts, and its acceptable use policy forbids spam, evading rate limits or bot detection, and automation that breaks a site's terms. Read the terms of the sites you use it on.
Does a cloud AI agent work the same way?
Not quite. A cloud agent runs its browser on the vendor's servers, so you sign in there and your session lives on their machines. A desktop agent keeps the session on your own computer.
See it on your own work.
Try it free for 7 days with no card. After that it is $499 a month or $4,990 a year, with a private setup call and one custom workflow built with you.