AI job alerts that filter by eligibility, not just keywords

What standard job alerts get wrong

A keyword alert for 'software engineer' in a major city can return dozens of postings a day. Many will require local candidates only, exclude visa sponsorship, demand ten years of experience for a role titled 'junior', or list a time zone that conflicts with where you live. None of that is visible in the alert email. You find out only after you click through and read the full posting.

The result is alert fatigue. Most people either ignore the alerts after a week or spend an hour each morning triaging postings that were never realistic. Neither outcome helps a job search.

• Keyword matches do not check visa or sponsorship requirements
• Seniority labels are inconsistent across employers, so 'mid-level' can mean two years or eight
• Remote postings often have hidden location restrictions such as 'US only' or 'must overlap with GMT+1'
• Salary ranges, where disclosed, are rarely surfaced in alert emails

What eligibility-first filtering actually checks

An eligibility-first alert reads the full job posting the way a recruiter would, then compares it against your actual profile before deciding whether to surface it. The checks that matter most are location and work authorisation, experience level, required skills versus your confirmed skills, and time zone overlap for remote roles.

In practice this means a posting that says 'no sponsorship, must be based in Ontario' is filtered out for a candidate on a work permit in Texas before it ever reaches them. A posting that says 'senior' but lists only three years of required experience is matched against your actual years, not the label.

• Location: city, state or province, country, and any 'must be local' language
• Work authorisation: sponsorship offered, citizen or permanent resident only, or open to all
• Experience level: years required versus years you have confirmed
• Skills: hard requirements versus nice-to-haves, checked against your resume
• Time zone: overlap windows for async or hybrid remote roles

How to set up smarter alerts manually

If you are using standard job boards, you can reduce noise by being more precise in your search strings before saving an alert. Use Boolean operators where the platform supports them, and save separate alerts for each job title variant rather than one broad alert.

On LinkedIn, use the 'All filters' panel before saving an alert. Set experience level, job type, and on-site or remote. On Indeed in the US, Canada, and Australia, the 'Employer type' and 'Encouraged to apply' filters help narrow results. On Reed and Totaljobs in the UK, the salary range filter removes a large share of irrelevant postings. None of these platforms filter by visa sponsorship automatically, so you still need to check that manually.

1. Write the exact job title in quotes to avoid broad matches, for example "data analyst" not data analyst
2. Add a minus sign before terms you want to exclude, such as -director if you are not at that level
3. Set a salary floor where the platform allows it to remove postings that are below your range
4. Save one alert per title variant rather than combining them, so you can turn off noisy ones
5. Set alert frequency to weekly rather than daily until you have confirmed the alert is well-targeted

When AI reads the full posting for you

Break-In Bad's assistant Omar reads the complete job posting, including the requirements buried in the middle paragraphs that most candidates skip. She checks location, visa language, time zone, and experience level against what you have told her about yourself, then scores each role. Postings that score 75 to 100 percent match are flagged for automatic application. Postings between 45 and 74 percent are surfaced for your review. Postings below 45 percent are set aside with a note on what would need to change before they are worth pursuing.

This means the alerts you receive are already pre-screened against your actual eligibility, not just your keywords. You spend time on roles where you meet the stated requirements, not on roles where you will be screened out in the first thirty seconds of a recruiter's review.

How to write a job alert query that ages well

Job titles shift over time. 'Growth hacker' became 'growth marketer'. 'Machine learning engineer' is now sometimes listed as 'AI engineer'. If your alert is tied to a single title, you will miss postings for the same role under a newer label.

A more durable approach is to anchor your alert on a core skill or tool rather than a title, then review the results weekly and adjust. For example, an alert for 'dbt' or 'Snowflake' in a data role will catch postings regardless of whether the employer calls the position analyst, engineer, or specialist. Combine this with a location or remote filter and you have a narrower, more stable alert.

• Use a core tool or technology as the primary keyword rather than a job title
• Add the industry as a secondary filter if your experience is sector-specific
• Review and update your alerts every four to six weeks as titles evolve
• Delete alerts that have not produced a relevant result in three weeks

Managing alert volume without missing good roles

The goal is not zero alerts. It is a manageable number of relevant ones. A reasonable target for an active search is five to fifteen genuinely eligible postings per week, depending on how specialised your role is. If you are seeing more than that, your filters are too broad. If you are seeing fewer than five, your filters or your target market may be too narrow.

Set a fixed time each week to process alerts rather than checking them as they arrive. Batch processing takes less time and reduces the anxiety of feeling like you are always behind. If a platform lets you pause an alert without deleting it, use that feature during weeks when you are not actively applying.