LinkedIn is hoping to use AI to make it easier for job recruiters and qualified candidates to find one another.

A new AI Hiring Assistant will soon let recruiters supply job descriptions and notes on qualifications, then automatically generate a short list of LinkedIn users who look like a good match for the job. And as recruiters go through the list of suggested candidates, the AI will indicate which of a job’s required and preferred qualifications each person appears to have.

“It’s telling you this person might be a top fit,” says Hari Srinivasan, LinkedIn’s vice president of product. “This person might be a fit.”

The Assistant is designed to automate what in the past might have been tedious manual searches through LinkedIn and other candidate databases and, since the AI is able to parse the text of job descriptions and user profiles, it can ideally help surface potential hires that might not show up through simple keyword matching. So far, it’s been rolled out to a select group of LinkedIn customers, including AMD, Canva, Siemens and Zurich Insurance, with the goal of making it more broadly available to recruiters.

Users can adjust what qualifications the AI is looking for and even provide examples of LinkedIn profiles that would make a good match. They can also enter or copy and paste other descriptions of what a company is looking for, like email correspondence from a hiring manager. Businesses can also connect the system to a supported applicant tracking system, which can let them resurface candidates who previously applied for another position.

As recruiters verify the AI’s suggested candidates are a good match, they can also have the AI draft messages to them, referencing specific elements of their background and skills that make them seem like a good fit for the job. They can also optionally have the AI system handle certain screening tasks, like making sure interested candidates have basic qualifications for the job.

In testing so far, those AI messages appear to be performing well—hirers who use AI-assisted messages get a 44% higher acceptance rate and get their LinkedIn outreach messages accepted 11% faster than those who don’t use AI to draft messages, the company says, and AI-assisted search sessions see an 18% higher message acceptance rates than manual ones.

And the AI also frees up more time for recruiters to focus on working with promising candidates and converting them to hires, which many say they wish they had more time to focus on, Srinivasan says. “But they’re spending more of their time on the much more repetitive and tedious tasks,” he says.

The Hiring Assistant is part of an overall push by LinkedIn to improve the quality of business-to-candidate matching, in a time when online search and application tools can incentivize both job seekers and recruiters to cast a wide net, meaning both sides waste time on connections that ultimately won’t lead to a hire.

“Quantity actually wastes everybody’s time,” says Rohan Rajiv, LinkedIn’s head of career products. “Job seekers send hundreds of applicant applications to jobs where they are under fit and won’t hear back, and in doing so, hirers actually struggle to get back to qualified candidates.”

The company is also testing AI tools that will give job seekers more insight into jobs posted on the platform for which they could be a good potential match—including whether, based on their LinkedIn profiles, they appear to be a strong applicant and meet the job’s listed qualifications. Users who find the system isn’t picking up on their qualifications can update their profiles to emphasize their strengths but, like the Hiring Assistant, the AI is designed to go beyond basic keyword matches, so users shouldn’t need to include every possible synonym for a skill in their profiles to be deemed a good match.

As with recruiters writing to potential candidates, jobseekers will also then be able to use LinkedIn’s AI to help draft a cover letter or customized resume for a particular position.

The system, which Rajiv expects will begin to roll out to LinkedIn Premium users within a few weeks, could help assuage concerns about opaque AI systems—or quick keyword searches and scans by harried recruiters—throwing out qualified candidates because their resumes didn’t include the right keyword.

“I think that is going to be a thing of the past, very, very soon,” says Rajiv. “So you can just focus on putting in what you actually did, and the system will understand you and actually share what it understands.”