Overview
The Candidate Match Rule Builder allows administrators to create candidate matching criteria using natural language instead of manually configuring individual matching rules.
With this feature, users can simply describe the candidate requirements they want in plain language, and Sense AI will automatically interpret the request and convert it into matching criteria.
This makes it easier and faster to configure Candidate Match nodes in AI Recruiter (AIR) and Workflows without requiring manual rule creation.
Availability
Beta Feature
Candidate Match Rule Builder is currently available as a Beta release.
While the feature has undergone extensive testing, AI-generated criteria may occasionally require review and validation before use.
How It Works
When configuring a Candidate Match node, users can describe candidate requirements using natural language.
For example:
- Find candidates within 50 miles of the job location.
- Find candidates within the same state as the job.
- Find candidates with 5 or more years of experience.
- Find candidates with specific certifications or skills.
The system analyzes the request and converts it into supported candidate matching criteria.
The generated criteria are then displayed for review before being added to the workflow.
Creating Match Criteria Using Natural Language
To create candidate matching criteria:
- Open your AI Recruiter workflow or Workflow containing a Candidate Match node.
- Navigate to the Candidate Match node.
- Click Edit/Add Criteria.
- Enter your requirements in natural language.
- Click Generate Criteria.
- Review the criteria generated by the system.
- The system will provide quick suggestions that you can use to generate criteria.
- If the results are correct, click Add Criteria.
- Continue building and configuring your workflow/Sequence.
The generated criteria are applied as candidate matching filters for the Candidate Match node.
Using Suggestions
The Candidate Match Rule Builder provides suggested example queries to help users get started.
These suggestions can be used as-is or modified to meet your organization's requirements.
Suggestions are designed to:
- Demonstrate supported filtering capabilities.
- Help users understand how natural language requests are interpreted.
- Provide examples of successful query structures.
Reviewing Generated Criteria
After clicking Generate Criteria, the system evaluates the request and displays:
- Criteria that were successfully interpreted and converted into matching rules.
- Requests that could not be converted into supported matching criteria.
This allows users to verify exactly what rules will be applied before adding them to the Candidate Match node.
If part of the request cannot be converted into supported criteria, the system identifies the unsupported portion so it can be revised.
Preview Matching Candidates
After configuring candidate matching criteria, administrators can preview candidate matches before activating the workflow.
To preview matches:
- Open the Candidate Match node.
- Select the Preview Matches tab.
- Search for and select an active job.
- Review the matching candidates returned by the system.
Previewing matches helps verify that the configured criteria are producing the expected candidate results.
Supported Candidate Fields
The Candidate Match Rule Builder currently supports the following candidate and job attributes:
Location
- Zip
- City
- State
- Country
Job Preferences
- Worksite Options
- Shift
- Work Authorization
Experience & Qualifications
- Years of Experience
- Skills
- Title
- Certifications
- State Licenses
- Languages
Industry & Specialization
- Industry
- Discipline
- Specialty
Compensation & Assignment Details
- Pay Rate Minimum
- Contract Duration Minimum
Important: These fields must be properly mapped to the required Job and Candidate designations on the ATS Entity page before the agency is onboarded to Candidate Matching.
Supported Filter Operations
The Candidate Match Rule Builder can generate rules using the following filter operators:
- Equals
- Not Equals
- In
- Greater Than
- Greater Than or Equal To
- Less Than
- Less Than or Equal To
- Exists
- Geo Within (Location Radius Filtering)
What Is Included in This Release
This release allows users to create and manage candidate matching filters using natural language.
The AI converts requests into supported matching criteria and applies them as candidate matching filters within Candidate Match nodes.
These filters work alongside any existing default matching rules configured for the agency.
What Is Not Included
The following capabilities are not supported in the current release:
Ranking Customization
Users cannot:
- Modify candidate ranking logic.
- Assign weights to skills.
- Configure custom ranking scores.
- Create ranking-based matching rules.
The Candidate Match Rule Builder only creates matching filters and does not change how candidates are ranked.
Beta Considerations
Because this feature uses AI to interpret natural language, users should review generated criteria before saving them.
You may occasionally encounter situations where:
- The generated criteria do not fully reflect the intended request.
- The request references fields that are not currently supported.
- The AI cannot generate criteria for part of the request.
If this occurs, revise the request using supported fields and filters, then generate the criteria again.
Frequently Asked Questions
1. Can I write matching rules in plain English?
Yes. The Candidate Match Rule Builder allows you to describe candidate requirements using natural language, and the system converts them into supported matching criteria.
2. Can I preview candidates before activating the workflow?
Yes. Use the Preview Matches tab within the Candidate Match node to review matching candidates for an active job.
3. Can I modify candidate ranking using natural language?
No. This release only supports creating candidate matching filters. Ranking customization is not currently supported.
4. What happens if the system cannot interpret part of my request?
The system displays which criteria were successfully generated and identifies any portions of the request that could not be converted into supported matching rules.