Talent SummonerHow to review Talent Summoner's candidate filters before evaluation
A practical guide to checking agent-proposed minimum score, source and city-level filters before a Talent Summoner search is evaluated.
Talent Summoner's 21 July 2026 Smarter candidate discovery release added agent-proposed candidate filters to the sourcing workflow. After evaluation criteria are set, the agent pre-fills the sidebar's minimum score and source filters. You can edit those values before anything is evaluated. The same release consolidates location options to the city level. It does not say that a city is automatically pre-filled.
This page is a review guide for that shipped behavior. A filter sets search scope; it is not evidence that a person has a skill, lives in a city, has work authorization or wants contact.
What the release documents
The Talent Summoner Changelog records three separate changes under the 21 July release:
- The discovery engine surfaces more matching profiles per sweep.
- The agent proposes a minimum score and source values in the candidate sidebar after evaluation criteria are set.
- Location filter options are collapsed to the city level, removing duplicate district choices.
The changelog says the proposed minimum score and sources can be edited before evaluation. It does not promise a fixed threshold, universal source coverage, source freshness, a city default, ranking accuracy or a hiring outcome. Recheck the live page when labels or behavior change.
Review each proposed value
Use the role brief as the decision boundary, then inspect the proposal one value at a time.
| Filter | Question to ask | Boundary to keep visible |
|---|---|---|
| Minimum score | Does the threshold match the approved must-haves and the breadth needed for this search? | A score is a search input, not a quality verdict. Widening or narrowing it can change which evidence is found. |
| Sources | Are the listed sources appropriate for this role and available for this search? | An absent source is not evidence that a person lacks a skill. The release does not promise universal source coverage. |
| City-level location | Is the city part of the approved work scope, and is the option a useful narrowing value? | City-level options are a UI grouping. They do not prove residence, commute, authorization, availability or willingness to relocate. |
If a proposed value is unclear, edit it or leave it marked unknown before evaluation. Do not turn a preference into an exclusion without an explicit role-owner decision. Keep the original brief and the final filter selection together so another reviewer can trace the scope.
Run a bounded review
Before evaluating results, record:
- The role-brief version and the evaluation question.
- The proposed values and the values you changed.
- The selected sources and city-level location, if any.
- The reviewer, date and any unavailable or empty source state.
After evaluation, read each profile's evidence in context. Distinguish no result under this scope, source unavailable and search failed. A profile's source label or city signal is only an observation about where a public signal appeared. It cannot establish capability, identity, authorization, availability or interest by itself.
The candidate-sourcing workflow starts from a role description and public professional sources. The separate candidate-ranking workflow compares a supplied job description with uploaded CVs. Do not carry source-filter assumptions from sourcing into CV ranking, interview assessment or outreach.
Worked review record
This is a fictional example, not customer or product evidence. A recruiter is sourcing a bilingual backend engineer for a Hong Kong team. The agent proposes a minimum score, LinkedIn and GitHub, plus city-level location choices. The recruiter keeps both sources, checks the brief's must-haves and chooses a city only because the role owner approved that scope.
Role brief: backend engineer / version 3
Proposed minimum score: [record value]
Final minimum score: [record value and reason]
Proposed sources: [record values]
Final sources: [record values and reason]
City-level location: [selected city or none]
Unavailable or empty sources: [record explicitly]
Reviewer and date: [record]
Unknowns requiring follow-up: [record]If no GitHub result appears, write “no observable GitHub result under this scope.” Do not write “no engineering experience.” If a city option is unavailable or the state is unclear, record it as unavailable or unknown rather than inferring that the candidate is elsewhere.
Product and evidence boundaries
This page documents the dated changelog behavior and a safe review method. It does not establish an ATS, applicant pipeline, automatic rejection, candidate verification, universal source coverage, freshness guarantee, saved-filter history or automated contact. The 31 July 2026 Changelog separately documents human-approved AI Email Outreach; a filter does not send a message or grant permission to do so.
For a search review, retain the Changelog URL, checked date, role brief, proposed and final values, result limitations and the human decision. If the sidebar no longer matches the release wording, consult current help or support before relying on a setting.
Does Talent Summoner pre-fill candidate filters?
Yes. The 21 July 2026 Changelog says the agent pre-fills minimum score and source filters after evaluation criteria are set, and that you can edit them before evaluation.
Is a city automatically selected?
The release says location options are consolidated to the city level. It does not say that a city is automatically selected. Treat a city as selected only when the interface and approved role scope show it.
Does a source filter prove a candidate has a skill?
No. It narrows where the search looks. Inspect the underlying public evidence and record unknowns or follow-up questions.
Can I use a filter to reject or contact a candidate?
No. A filter is a discovery input. It does not create an automatic rejection rule or outreach permission. Human review and the documented workflow remain separate.
What should I record before evaluation?
Keep the role-brief version, proposed and final minimum score, source choices, city-level choice, reviewer, date and any unavailable or empty source state. This makes the search scope reproducible.
Next step: review the candidate-sourcing workflow and the dated Changelog before evaluating the next search.
