Your HR Team Is Wasting 57% of Their Day on Admin. An AI Computer Use Agent Fixes That.

The Workday Lawsuit Should Terrify Every HR Leader

Let's start with the thing nobody wants to talk about at the HR tech conference. In May 2025, a federal court granted conditional certification for a collective action lawsuit against Workday, alleging that its AI-powered applicant screening system discriminated based on race, age, and disability. The EEOC backed the plaintiff. The judge noted the class could involve 'hundreds of millions' of affected applicants. Hundreds of millions. This is what happens when you build AI automation that operates as a black box, trained on historically biased hiring data, and then let it gatekeep who even gets seen by a human recruiter. The old promise of AI recruiting was 'remove human bias.' The reality is that you just laundered the bias through an algorithm and gave it legal cover. That's not automation. That's a liability factory. The lesson isn't 'don't use AI in HR.' The lesson is 'stop using AI tools that make opaque decisions you can't audit or control.'

What HR Teams Are Actually Doing All Day (It's Embarrassing)

• ●57% of HR time goes to administrative work, not strategic hiring or employee development (Pentabell, 2025)
• ●The average cost per hire hit $4,700 in 2023, up 14% from 2019, and it keeps climbing
• ●Companies using AI in recruiting report 77.9% reduction in hiring costs and 85.3% time savings (HeroHunt.ai, 2025)
• ●The average time-to-fill a position is 44 days, during which your best candidates are accepting offers elsewhere
• ●Recruiters manually review an average of 250 resumes per open role, spending roughly 6 seconds on each one
• ●Interview scheduling alone eats 30-60 minutes per candidate when done manually, across email threads that look like a hostage negotiation
• ●Onboarding paperwork, background check follow-ups, and ATS data entry account for hours of work that produces zero value for anyone

Why Traditional RPA and ATS Plugins Didn't Save You

The HR tech industry sold you a dream in the 2010s. Buy this ATS. Add this RPA bot. Integrate this Chrome extension. The result? Most companies now have five to eight HR tools that don't talk to each other, RPA bots that break every time a vendor updates their UI, and recruiters who spend half their time doing data entry between systems that were supposed to eliminate data entry. Classic RPA is fundamentally brittle. It records clicks and keystrokes on a fixed interface. The moment LinkedIn changes their layout, or your ATS pushes an update, the bot dies. Then someone from IT has to fix it. Then it breaks again. This is why a16z called out the shift to 'computer use agentic coworkers' as the actual step change in automation, not more point solutions bolted onto legacy systems. The difference is whether your automation can see a screen, reason about what it's looking at, and adapt. That's what separates a real computer use agent from a fragile macro.

What a Real Computer Use Agent Does for HR

Here's what AI computer use automation actually looks like when it works. A computer-using AI agent can open your ATS, read a job description, go to LinkedIn or a job board, search for matching candidates using real judgment about skills and context, pull their profiles, fill in your ATS fields, draft personalized outreach, and log everything. Without an API. Without a custom integration. Without a developer. It works on the actual screen the same way a human would, which means it works with every tool you already have. Interview scheduling? The agent reads the calendar, finds mutual availability, sends the invite, and updates the candidate record. Offer letter generation? It pulls the approved template, fills in the candidate-specific fields from your HRIS, routes it for e-signature, and files the signed copy. Background check follow-ups? Done. Onboarding checklist emails? Done. The stuff that was eating 57% of your team's day can run in the background while your recruiters do the one thing AI genuinely can't do: build real relationships with candidates who have options.

Why Coasty Is the Computer Use Agent HR Teams Are Actually Using

I'm not going to pretend every computer use agent is equal, because they're not. Coasty scores 82% on OSWorld, which is the standard benchmark for AI computer use performance. That's higher than every competitor, including Anthropic's Computer Use and OpenAI's Operator. The gap matters because HR workflows are messy. Candidates have weird resume formats. ATS interfaces have quirks. Job boards change their layouts constantly. A computer use agent that scores 60% on OSWorld will fail on roughly 4 out of 10 tasks. In a recruiting workflow, that means dropped candidates, missed follow-ups, and corrupted data. At 82%, Coasty handles the real-world messiness that breaks lesser tools. It controls actual desktops and browsers, not just API endpoints, so it works with whatever stack you're already running. You get a desktop app, cloud VMs for running parallel agent swarms (useful when you're processing 250 applications for a single role), and a free tier to start without a procurement battle. BYOK support means you're not locked into someone else's infrastructure decisions. The point isn't to replace your recruiters. It's to give them back the 57% of their day that's currently being consumed by work a computer should be doing.