AI Sourcing is the default sourcing mode on a job. Kalent searches the pool, applies your filters, then evaluates each profile against your qualification criteria.
You can switch to classic sourcing at any time with the dedicated button.
The 4 steps
Describe your need. Paste a job description, describe the profile, or drop a PDF. Kalent extracts the search filters automatically.
Review filters. Adjust them. The estimated candidate pool updates in real time.
Qualification criteria. Review, add or remove what the AI should judge on each profile. You can also launch the analysis with no criteria.
Launch analysis. The search starts. Results appear as they come in. You can keep working in the meantime.
If you open Settings and change the need, the filters or the criteria, a new full analysis starts.
Filters and qualification criteria: two different roles
Filters decide who enters the pool: job title, location, years of experience, company size, and so on. A profile that misses a filter never shows up.
A qualification criterion is a requirement the AI reads on the full profile, once the person is already in the pool. It does not shrink the pool: it qualifies and ranks.
Put structured, factual constraints in filters. Put judgment calls in criteria: type of company, type of clients, management, what you want to avoid.
What makes a good criterion?
Write it as a sentence you could check on a LinkedIn profile. Be specific. You can add up to 10. 3 to 5 is often enough.
Examples (the ones Kalent offers when the list is empty):
Has managed a team of 5 or more people
Has experience in B2B SaaS
Has hands-on experience building artificial intelligence or machine learning systems
Has sold to enterprise accounts or mid-market
Has never worked in financial services or investment banking
Has hands-on transaction experience, not only rental management
Avoid profiles from the public or non-profit sector
You can use them as-is, click to add them, or write your own. Kalent also suggests criteria from your job description: review them before you launch.
How the AI assesses each talent
On each card you will see:
a summary sentence of the profile against your need
one line per criterion, with an icon and a short explanation, based on what is written (or missing) on the profile
The overall qualification then ranks the talent:
Strong. Every criterion is clearly met. This is the list shown first, your working shortlist. With no criteria, everyone who passed the filters is Strong.
Medium. A partial match, or not enough information to judge some criteria. These are not necessarily poor fits: many are worth a look.
Weak. At least one criterion is not met. These profiles diverge from what you are looking for.
A career fact that should appear on a profile (“has managed a team”) and is nowhere to be found is usually treated as not met. A skill people rarely spell out (“knows this tool”) with no mention on the profile will often land in Medium for lack of information, not Weak.
For an exclusion (“has never worked in finance”), the absence of that sector on the career path counts in the right direction.
Strong profiles appear as they are found. You can browse Medium and Weak in the left column. Talents you save or set aside leave this feed: you will find them in your saved profiles.
In detail: why the analysis stops, and how to continue
The analysis does not scan the whole pool in one go. It stops when it has found 10 strong qualifications, or after analysing a batch of profiles. That is intentional: you can already work on the best profiles, then ask for more.
When the batch is done, click Analyze more to run the next one and discover new profiles. If some medium qualifications remain, Kalent also offers to review them before you continue.
If the entire pool has been analyzed, there is no next batch for these filters. Widen the search with Adjust filters (broader location, extra job titles, fewer structured requirements). Adding more qualification criteria will not grow the pool.
If almost nobody is Strong, Kalent may show an alert and ask you to review your criteria. They may be too strict, or out of line with your filters.
Two extra cases:
Medium qualification at the end of the pool. Some profiles match your filters, but information is missing to judge every criterion. Kalent invites you to look at them: they are not necessarily poor fits.
Set aside. A few profiles were analysed without receiving a qualification. This happens when the LinkedIn profile is no longer available, or when the person is no longer in the pool after an update.
If the analysis fails, a Retry button relaunches the current batch.