For recruiters and staffing firms

An AI agent for recruiters and staffing agencies

An AI agent for recruiting does the browser work between you and a placement: sweeping job boards for companies that are hiring, building candidate lists, researching people before a call and drafting the first note. Vassal Agent does this in your own signed-in browser and stops for your approval before any message is sent.

Updated September 24, 2026

What recruiting work can an AI agent do?

Most recruiting hours are not spent on judgment calls. They go to finding, reading, copying and checking: which companies just opened a role, who the hiring manager is, whether a candidate's last three jobs fit the brief, what goes in the ATS. Those steps all happen in a browser, which is where Vassal Agent works.

1. Job-board sweeps for business development

Staffing firms win clients by reaching companies at the moment they are hiring. Tell the agent the roles you place and the kind of company you want, and it checks the boards and careers pages you name, collects each opening, and returns a table: company, role, location, date posted, link, and the likely hiring manager where it can be found on a public page.

Try: "We place sales leaders at venture-backed startups. Find companies in New York hiring a head of sales or VP of sales this month, with the job link and the founder's name."

2. Candidate sourcing

Give it the brief and the places to look: search results you can see while signed in, company team pages, GitHub, conference speaker lists, portfolios. It reads each profile and keeps the people who fit, with a one-line reason for each and a link to the source, so you can check its work in seconds.

3. Candidate and client research before a call

Before a screen or a client meeting, ask for a short brief: career history from public profiles, recent posts or talks, the company's latest funding news and open roles. It reads the pages and writes a summary with links, so you walk in prepared without twenty open tabs.

4. Outreach drafts that wait for approval

For each person on a list, it writes a first note in your voice that refers to something specific about them, and opens it in the channel you use. Then it stops. You read the draft, edit it if you like, and press Allow to send, or skip it. It never sends in bulk on its own.

5. ATS and CRM entry

It copies candidate or client details from a profile into your ATS or CRM in the browser, field by field, and stops before saving if you want to check. If your ATS has no import for a source you use, this is often the biggest time saver.

6. Watching for changes

Point it at the careers pages of your target accounts, or a board filtered to your niche, and it tells you when a new role appears. Runs can be scheduled every morning, with results waiting in the app's Inbox.

How it works for a recruiting firm

1. Sign in once. Open the job boards, LinkedIn, your ATS and your email in the app's browser and sign in as usual. Sessions stay on your computer.
2. Describe the search. Type what you want in plain words, as you would brief a researcher.
3. Watch or walk away. A live panel shows every page it opens. Results arrive as a table you can copy, save as a spreadsheet or send to Google Sheets.
4. Approve what leaves. Messages, form submissions and deletes wait for your Allow.
5. Keep what works. Do a search once while it watches, and it becomes a workflow you can re-run or schedule.

What it will not do

• Send without you. Every send waits for approval. You can add rules like "Never press Send on linkedin.com" in plain words.
• Get around a site's limits. It works at a human pace inside your own account and does not evade rate limits, CAPTCHAs or bot detection. That is also in our acceptable use policy.
• Invent contacts. It reports what it found on real pages, with links. It has no contact database of its own.
• Make the call. Judgment, relationships and the phone stay with you.

Why a browser agent instead of another recruiting tool?

Recruiting software usually does one thing: a sourcing database, a sequencer, a scraper for one site, an ATS. A browser agent works across all of them, because it uses the same websites you do, signed in as you. The trade-off is speed. A dedicated database returns ten thousand rows in a second; an agent reads pages one by one, the way a researcher would, and is better suited to a focused list of the right fifty people than a list of everyone.

It also means you can hand it the odd jobs no tool covers: a niche board, a client portal, a spreadsheet that needs checking against a website.

What it costs compared with the alternatives

Vassal Agent is $499 a month or $4,990 a year, with a private setup call and one custom workflow built with you in the first month. It runs on your own AI plan, such as Claude Pro at $20 a month. For comparison, a part-time researcher or virtual assistant for 20 hours a week costs roughly $700 to $2,200 a month at $8 to $25 an hour. The full comparison with a virtual assistant covers where a person is still the better choice.