How AI job search works

Type naturally

Type what you are really looking for - e.g. "cooking for a kindergarten Brno" or "junior developer Prague". JobsAI combines classic full-text search with AI understanding of meaning, so it finds listings that do not contain your exact words.

When we find nothing

Results only include listings that really match your search or are close to it in meaning. If the catalogue has no such listings, JobsAI shows nothing rather than random jobs with a similar-sounding word. In that case try a broader term (for example "mechanic" instead of a narrow specialisation), another name for the profession, or adjust the filters.

Refine with filters

Narrow results in the left panel: region, employment type (HPP, DPP, DPC, part-time job, freelance), salary, remote work or publication date. Filters combine with AI search.

Map of results

On the web, you display the results on a map using the button for showing the results on a map below the result list. In the mobile app, from version 1.2.0 there is a List / Map switch. The map shows the first 50 results of the current search with the same query and filters; listings without coordinates appear in the list but not on the map.

Tailored recommendations

The Recommendations page needs no query - it works from your profile: preferred occupations, skills, regions, employment types and industries. The more complete your profile, the better the matches.

What the match score means

The match score in recommendations indicates how similar a listing is to your profile or query. It is not your probability of being hired or confirmation from the employer that all requirements are met.

Check the listing and, where needed, ask the employer about shifts, base pay, the workplace and the start date. Distance in kilometres does not indicate commuting time.

AI transparency

Every recommendation comes with an explanation of why AI suggested it. How our AI works and what rights you have is described on the About AI page.

Fit with your work requirements

Comparison with saved work requirements can show a match, mismatch or missing information for each condition. A stated pay range alone does not establish a guaranteed minimum or base pay excluding bonuses. Some conditions cannot be compared reliably and remain unknown.

Recommendations separate jobs meeting required conditions according to available information from jobs needing verification. Semantic similarity percentage is a separate measure. An empty result does not mean suitable work does not exist, and requirements are not relaxed automatically.

Employers cannot see your private requirements, and the comparison does not prevent you from applying.