At Handshake, we are committed to making the hiring process easier, more efficient, and more effective for employers.To support this mission, we use artificial intelligence (AI) in various features within the Handshake platform.
This article provides a detailed overview of how these AI-powered features function, what they are designed to do, and how they benefit you as an employer using Handshake.
Our use of AI is governed by a commitment to transparency, fairness, and responsible innovation. We follow these principles:
- We never make hiring decisions on behalf of the employer. We keep a "human in the loop" and do not use AI for automated decision making.
- We do not share data with third parties for AI training purposes. We do not export employer data to third party AI vendors for training purposes, nor do we ingest employer data from ATS or other systems for training purposes.
-
Governance and control: We maintain oversight of all AI models through internal governance frameworks and ensure compliance with applicable privacy and regulatory requirements. We use observability tools and follow a quality assurance process to monitor performance. Where appropriate, we also conduct independent AI bias audits.
- The latest summary of results conducted by independent auditor Holistic AI can be found here.
Which Handshake features use AI?
1. AI-Powered Autofill
Handshake uses AI to reduce manual work for recruiters by extracting data from job posting text to pre-populate fields in Handshake.
- Job Form Autofill
- How the AI use case is triggered: Job form autofill is primarily triggered by employers through the Handshake application when they start posting a job. Employers have the option to use AI to break down their job posting text into the required Handshake job fields. The autofill option is selected by default, but employers can choose to de-select it. Job form autofill is also used to fill out fields for publicly posted jobs that are imported into Handshake.
- Employer control: Employers retain full control, with the ability to edit or disregard AI-generated suggestions.
2. Campaign Segment Autofill
- How the AI use-case is triggered: When an employer starts creating a campaign and attaches a job post, Campaign Segment Autofill pre-populates suggested targeting filters based on the job requirements. Employers can review, edit, or add to these filters before finalizing the campaign.
- Employer control: Employers retain full control, with the ability to edit or disregard AI-generated suggestions.
- Data and privacy: These features do not access or rely on any personal data. Only job details are processed.
Note: In the future, we plan to extend AI-powered autofill capabilities across other areas of the product, following similar model & data use and privacy considerations above.
2. AI-Assisted Draft Messaging
Handshake offers AI-powered features to assist employers in crafting messages to reduce manual work in message creation and improve personalization of messages:
1. AI-Assisted Direct Messaging
- How the AI use-case is triggered: When an employer opens the message composer to draft a message to a single candidate from a relevant product area, such as Matches or Applicants, Handshake offers an AI-generated draft personalized to the candidate and job.
- Employer control: Employers retain full control, with the ability to edit or disregard AI-generated suggestions.
2. Automated Message Generation for Automated Job Promotions (AJPs)
- How the AI use-case is triggered: When Handshake Plus employers create an Automated Job Promotion, Handshake offers an AI-generated draft message personalized to the candidate and job.
- Employer control: Employers retain full control, with the ability to edit or disregard AI-generated suggestions.
- Data and privacy: We obtain applicable consents from educational institutions and students to share student information with employers. Sensitive demographic information such as race or gender is not used to generate messages.
Note: In the future, we plan to extend AI-assisted messaging capabilities across other areas of the product, following similar model & data use and privacy considerations above.
3. AI-Powered Audience Selection
Handshake provides AI-powered features to select student audiences most relevant for employer campaigns and other promotions:
1. Campaign Relevance
-
How the AI use-case is triggered: When an employer initiates a campaign, AI is used to select the student audience most likely to find it relevant
2. Dynamic Targeting in Campaigns
- How the AI use-case is triggered: When an employer initiates a job campaign, they have the option to further optimize campaign delivery by delivering outreach in batches to students with the highest predicted likelihood to apply.
- Employer control: Campaign Relevance is used across all campaigns, but employers have the option to choose whether to use Dynamic Targeting.
- Data and privacy: Employer and job details, and candidates’ interest and activity patterns (e.g., previous job clicks or applications) are used to surface opportunities to a relevant student audience. Data is pseudonymized before being ingested into the applicable underlying AI models powering Campaign Relevance and Dynamic Targeting, meaning it cannot be traced to the individual.
Note: In the future, we plan to extend AI-powered audience selection across other product areas where promoted opportunities are surfaced.
4. AI-Powered Relevance
Handshake uses AI to surface the most relevant candidates to employers across the product including:
1. Job Matches
- How the AI use-case is triggered: When an employer views candidates on Job Matches, candidates are sorted based on Best Match by default.
- Employer control: Employers can change the sorting to alternative criteria like name and recent activity.
2. Talent List
- How the AI use-case is triggered: When an employer views candidates on Talent List, candidates are sorted by an evaluation of the candidate’s relative qualifications and interest in the employer’s job. For example, we consider whether the candidate has engaged with the employer in the past through messaging, following, or applying, and whether their profile matches any active jobs the employer has posted
- Employer control: Employers can view alternative lists of candidates by viewing saved searches or filtering by segments or other candidate criteria.
3. Applicant Management (available to Plus and Pro)
How the AI use-case is triggered: When an employer views applicants on the Applicants tab, they can define which criteria to use to rank those applicants.
Employer control: Employers can choose to sort applicants by application date or name instead.
Data and privacy (applies to Job Matches, Talent List, and Applicant Management)
Employer and job details, and candidates’ interest and activity patterns (e.g., previous job clicks or applications) are used to present the most relevant profiles. No protected categories (e.g. race or gender) are used to determine a candidate’s relevance.
Note: In the future, we plan to extend AI-powered relevance across other areas of the product, following similar model & data use and privacy considerations above.