Two stories broke within days of each other this summer, and neither one was framed as recruiting news. Visa told investors it is cutting about 2,600 jobs — roughly 7% of its workforce, concentrated in technology and product teams — in a restructuring CEO Ryan McInerney tied directly to AI reshaping how work gets done, with savings earmarked for growth areas like cross-border payments and stablecoin infrastructure (CNBC, Bloomberg). Around the same time, OpenAI announced it is building its own jobs platform — matching candidates to employers, launching by mid-2026, with early partners including Walmart, John Deere, Boston Consulting Group, Accenture, and Indeed, plus a certifications push aiming to train 10 million workers in AI skills by 2030 (OpenAI, Yahoo Finance).
Neither story, on its own, tells HR much. Together, they mark something real: 2026 is the year AI in hiring stopped being a feature you bolt onto a job board and became something closer to a colleague — one that acts on its own, reshapes headcount decisions, and now has a foundation-model company competing to build the infrastructure it runs on. That shift has a name in the industry — "agentic AI" — and it deserves more scrutiny than the term is currently getting.
The gap between "using AI" and "running on AI"
Start with the base rate. According to 2026 research from Aptitude Research and iCIMS, 69% of companies now use AI somewhere in talent acquisition, but only 18% report broad use across their hiring workflows (iCIMS newsroom). That gap — nearly universal dabbling, narrow deployment — is the honest starting point for any 2026 hiring-trends piece, including this one. Most organizations are not running agentic recruiting. Most are running a chatbot on top of a legacy ATS and calling it AI.
But the leading edge is moving fast, and it's moving toward autonomy specifically. Korn Ferry's 12th Annual Talent Acquisition Trends survey — more than 1,670 global talent leaders plus 230 of Korn Ferry's own talent experts — found that 84% of talent leaders plan to use AI in recruiting in 2026, and, more specifically, 52% plan to add autonomous agents to their teams this year, with some organizations already creating internal employee-style records for those agents (Korn Ferry). That's a different claim than "we use an AI tool." An assistant drafts an email or summarizes a resume when you ask it to. An agent sources candidates, screens them, drafts outreach, and books interviews across a multi-step sequence without a human approving each step along the way.
The distinction matters commercially too. Industry analysis from Gem's 2026 recruiting-software roundup puts a number on it: a recruiter working on an AI-native, full-stack platform can delegate an estimated 50–60% of manual work to the AI layer, while a recruiter on a retrofitted applicant tracking system with AI features bolted on delegates closer to 10%, largely because the underlying data model was never built for an agent to act inside it (Gem). Architecture, not marketing copy, is what determines whether "AI-powered" means autonomous action or a search bar with a chat window on top.
Visa is a preview, not an outlier
It's tempting to read the Visa cuts as a payments-industry story with no hiring angle. We think that's backwards. Visa's move is one of the clearest public data points yet on what "AI reshaping work" looks like once it leaves the pilot stage and hits an org chart: roughly 7% of a large, technically sophisticated workforce reduced, with the CEO naming AI as a driver (alongside, reporting indicates, other factors) and savings redirected into new growth bets rather than simply cut (CNBC).
For talent acquisition teams, that pattern creates two simultaneous jobs, not one. The first is using AI to hire faster and more accurately. The second, harder job is hiring into an organization whose shape AI is actively rewriting — filling different roles than the ones automated away, at a pace the org chart hasn't caught up to yet. A recruiting function still organized around last year's requisition mix, using tools that can't tell you which roles are shrinking versus growing inside your own company, is not positioned to do that second job. That's an argument for TA teams to have real visibility into workforce and skills data, not just a faster funnel — the "role redesign" problem is going to keep showing up in earnings calls before it shows up in HR strategy decks.
Why a foundation-model company building a jobs platform changes the stakes
OpenAI entering hiring directly — rather than just powering other companies' hiring tools — is a legitimizing and a complicating signal at once. Legitimizing, because when the company that popularized the current wave of AI decides matching people to jobs is worth building in-house, "AI does the matching" stops being a differentiator any single HR-tech vendor can claim and becomes a baseline expectation candidates and employers will bring to every platform. Complicating, because it raises a question the industry has mostly avoided answering with specifics: when an agent — not a person — is the one deciding which candidates surface, on what basis, whose model is that decision running on, and can anyone outside the vendor explain it after the fact?
That question isn't hypothetical regulatory hand-wringing. It's the practical difference between an agent you can manage and one you can only hope is working.
What buyers should actually demand from "agentic" hiring tools
Given where the market is heading, we think HR and TA leaders evaluating agentic recruiting tools in the second half of 2026 should be asking harder questions than "does it have AI":
- Where does the agent act autonomously, and where does a human have to sign off? Sourcing and scheduling are low-stakes places for an agent to act alone. Advance/reject decisions and final recommendations are not — those are the moments a human should still be in the loop, not rubber-stamping after the fact.
- What happens when the agent is wrong? Ask for a specific example of a candidate incorrectly screened out and what the recovery process looked like, not a hypothetical answer.
- Is the pipeline staged, or is one agent making an end-to-end call? A design where AI handles early-funnel work (initial resume review, a first screening conversation) that feeds into human-run technical assessment, live interviews, and final decisions is a fundamentally different — and more governable — thing than a single agent owning sourcing through offer. NiceHire's own hiring pipeline is built on that staged model deliberately: AI resume vetting and an AI screening interview sit early in the funnel, ahead of technical assessment, live interviews, and a final round that stay human-run. It's a narrower claim than "fully autonomous hiring," and we think narrower claims are the ones actually worth making in a market that's about to be full of the other kind.
- Does the candidate know they're dealing with an agent? As agentic tools take on more of the early conversation, simple disclosure — that a candidate is talking to an AI system, not a person — is a low bar that some tools still don't clear.
None of this is an argument against agentic AI in hiring. The productivity case is real, and the direction of travel — Korn Ferry's 52%, the delegation-rate gap between AI-native and retrofitted platforms, OpenAI's entry into the category — is not reversing. It's an argument for treating "agentic" as a design decision with real tradeoffs to evaluate, the same way you'd evaluate any system that gets to act on your behalf without asking first, rather than a checkbox that makes a hiring tool automatically better than the version that just answers questions when asked.
The organizations that get this right in 2026 won't be the ones with the most autonomous recruiting stack. They'll be the ones that can explain, for every consequential decision an agent makes about a candidate, exactly where a human was supposed to be — and was.
Sources
- CNBC — Visa is cutting 7% of employees in efficiency push as AI reshapes work
- Bloomberg — Visa to Cut 7% of Workforce as CEO Seeks to Revamp Company
- OpenAI — Expanding economic opportunity with AI
- Yahoo Finance — OpenAI to Launch AI Jobs Platform by 2026
- iCIMS / Aptitude Research — 2026 AI Adoption Report
- Korn Ferry — TA Trends 2026: Human-AI Power Couple, full report
- Gem — Top 12 recruiting software with AI capabilities in 2026
Ready to transform your hiring?
See how NiceHire's AI-powered hiring platform works for your team.
Get Started