Indeed puts two AI agents in front of job seekers and recruiters

Two agents become the new front door

The headline product news in Indeed's interview is a pair of named agents launched in recent months: Career Scout, described as a personal career coach for job seekers, and Talent Scout, aimed at automating the time-consuming parts of recruiting for employers. Both sit on top of what Hulce calls over a hundred AI-powered features already spanning job search and hiring.

This is a frontend reframing more than a backend one. Indeed says it has used AI for nearly 20 years to power matching; what changed is that the matching now surfaces as an agent a person can converse with, rather than a ranked list they scroll. Indeed confirms OpenAI APIs power Career Scout and Talent Scout alongside Invite-to-Apply.

The two-sided design matters. Career Scout faces the job seeker, Talent Scout faces the employer, and Hulce claims outcomes improve most when both sides use AI capabilities together — a marketplace bet that the interface has to be built twice, once for each participant.

The recommendation surface, quantified

The interview leans on specific numbers to argue the interface layer is working. In Sponsored Jobs, roughly 70% of sponsored applications now come from AI-powered recommendations. In Smart Sourcing, employers using AI capabilities are said to hire 40% faster. Premium Sponsored Jobs, described as used by hundreds of thousands of employers, is credited with moving forward three times more applicants and nearly 60% faster time to hire versus non-sponsored jobs.

On the seeker side, Indeed cites early Career Scout testing where people found jobs they were excited about seven times faster and were 38% more likely to get hired. And a recommendation-to-action figure stands out: when Indeed surfaces highly matched candidates and an employer reaches out, those candidates are described as 15 times more likely to apply than if they'd found the job themselves.

One named customer anchors the claims. BrightSpring Health Services is said to have filled 45% more hard-to-fill healthcare roles in four weeks while saving teams eight hours a week during a Talent Scout test. These are Indeed's figures, offered without methodology, but they are notably concrete for a vendor interview.

Transparency treated as an interface requirement

The most interesting design constraint in the piece is the insistence that recommendations explain themselves. Hulce says job-seeker recommendations are meant to help people understand why a role is a good fit, and that for employers the criteria stay theirs and the experience 'never feels like a black box.'

Our goal is for AI to feel like it gives employers superpowers, while they always stay in control. The process is transparent, the criteria are theirs, and the decisions remain human—it never feels like a black box.Montana Labs

That framing puts explainability and human sign-off inside the product surface rather than in a policy document. Indeed positions AI as sourcing, screening, and evaluation support while employers remain the final decision-makers — a deliberate limit on how much the agent is allowed to decide on the user's behalf.

What this signals for teams building agent-shaped product surfaces

Indeed's account is a case of wrapping a mature matching system in a conversational layer without discarding the existing product line. The hundred-plus features, Sponsored Jobs, and Smart Sourcing all persist; the agents are additive front ends that route users into the same underlying connections.

For anyone shipping similar experiences, the specific lesson here is the pairing of agentic UI with visible reasoning and a hard human-decision boundary. Indeed is betting that a career coach or recruiting assistant only earns trust in a high-stakes domain like hiring if it shows why it recommended something and stops short of deciding — and that the measurable payoff shows up as application and hire rates, not just time saved.