Knowledge Base
> Getting Started
Setting up Core Criteria
Core Criteria are the rules the AI scores every candidate against. Set them on the job, change them whenever you need to, watch
| 5 Min Read
the pool re-rank in real time.
What Core Criteria are
Core Criteria are a short list of structured requirements you attach to a job. They are the only thing the AI looks at when
producing the Resume Match score, the AI Skill Match, the 3-column vetting view, and the hiring recommendation. No
criteria, no scoring.
You set them during step 2 of the job creation wizard and you can change them at any time afterward from the job page.
Core Criteria live on the job, not the workspace. Two roles with the same title can have completely different criteria. A senior
backend engineer in Riyadh and a senior backend engineer in Paris are not the same hire, and the AI should not pretend they
are.
[Illustration: A simple "job card" rendered in Forest and Signal Mint. To the right of the card, an arrow labeled "Drives" points
to four miniature output blocks: Resume Match, Skill Match, 3-column brief, Recommendation. Caption underneath in Sage
Gray: "One criteria set. Four outputs."]
What makes a criterion useful
The AI is only as specific as you are. A vague criteria set produces vague scores, vague briefs, and a reader who ends up
scrolling through the full pool anyway. Three rules that hold up across roles:
Be specific about the level, not just the skill. "Python" is a weak criterion because almost any candidate can claim it.
"Production Python in a service-oriented backend, three plus years" is a strong criterion because it can be evidenced or not.
Separate the must-haves from the nice-to-haves. Each criterion has a weight (High, Medium, Low). Use High sparingly. If
everything is High, nothing is. Most jobs have two or three High criteria; the rest are Medium or Low. The AI weights its
outputs accordingly.
Write criteria the way an interviewer would test them. If you cannot imagine the question that would confirm the criterion in
an interview, the criterion is too soft for the AI to score reliably. "Strong communicator" is hard to test; "experience presenting
to executive stakeholders" can be probed in five minutes.
How to change criteria mid-search
This is the single most underused feature in Talinty. The criteria you set on day one are almost never the criteria you actually
want by day fourteen. The product is built for that.
There are two ways to change criteria after the role is live.
Edit the Core Criteria set on the job page. Open the job, go to the Criteria section, edit, save. The entire pool re-ranks against
the new criteria within seconds. Resume Match and Skill Match update for every candidate. The 3-column briefs regenerate.
Type a new requirement in plain language inside any tab. This is the Dynamic AI Criteria Filter. In the filter bar at the top of a
candidate list, type something like "must have led a team of five or more" or "exclude candidates who have changed jobs
more than twice in three years". The AI interprets the requirement, applies it to the candidates in that tab, and re-ranks the
list immediately. Your underlying Core Criteria are unchanged; the tab gets the additional filter on top.
The plain-language filter is the fast version. Editing Core Criteria is the durable version. Use the filter when you are exploring;
commit to the Criteria edit when you have decided the requirement actually matters.
[Illustration: Two side-by-side panels. Left panel labeled "Edit Core Criteria" shows a job settings card being edited, arrow
pointing to a "pool re-ranking" indicator. Right panel labeled "Type into the filter bar" shows a search field with the example
query "led a team of 5+" and the same re-ranking indicator. Both panels use Signal Mint backgrounds with Forest typography.
A horizontal divider between them labeled "Same effect. Different commitment."]
What happens to candidates you already advanced
Changing criteria does not move candidates between stages. A candidate you already advanced to Interview will stay in
Interview even if the new criteria would have rejected them on Application. The scoring updates so you see the new
alignment, but the pipeline state is yours to manage.
This is intentional. The AI rescoring should not override decisions a human already made. If you want to reconsider an
advanced candidate after a criteria change, the new score will be visible on their card; you decide whether to move them.
When to leave criteria alone
A useful counter-rule: do not change Core Criteria more than three times in the first week of a role being open. Each change
re-ranks the pool, which changes the order recruiters see candidates in, which can make it hard to compare what you saw on
Monday to what you see on Thursday. Make changes deliberately, not reactively.
If you find yourself wanting to change criteria every day, the problem is usually upstream. The job description and the criteria
need a deeper rewrite, not a constant patch.