Latest / Future of Work Tech with Fexingo: Remote Tools, AI Productivity, and Workplace Software / How AI Is Automating Your Employee Contract Negotiations
Transcript
- Lucas: So you get an offer letter from a company you're excited about. The salary is good but not great. You write back, politely asking if there's any flexibility. And the reply comes in thirty seconds — friendly, specific, addressing your concerns point by point. But it's not a person. It's an AI. Luna: I've heard of AI screening résumés, AI scheduling interviews. But AI actually negotiating the terms of employment? That feels like a different order of magnitude. Lucas: If today's conversation gives you a new lens on where this technology is heading, this show stays ad-free because listeners like you chip in. If you found value in this, you can support us at buy me a coffee dot com slash fexingo. Keeps the episodes independent and focused on what matters. Luna: Yeah, it's a small gesture that makes a big difference for the show. Lucas: Right. So back to the story. The company I'm talking about is a mid-sized tech firm — I won't name them because they're not public about this yet — but they deployed an AI negotiation assistant for their 2025 compensation cycle. That's about six months ago now. They handled over 800 offer letters and counteroffers through the system. Luna: And the AI was the primary negotiator on the company side? Not just drafting language? Lucas: Correct. The system was authorized to make real-time adjustments within a predefined band. So if a candidate asked for ten percent more, the AI could counter with five percent and a signing bonus, all without a human in the loop. Luna: What was the actual outcome? Did candidates accept faster? Lucas: They saw a twelve percent reduction in time to acceptance. The AI could respond instantly, which meant candidates never sat around wondering if their email was ignored. But also, the system was trained on historical offer data and market comps, so it was making pretty rational counteroffers. Luna: Still, there's something unsettling about negotiating with a machine that knows your market value better than you do. Lucas: And that's exactly the ethical friction. Employment lawyer I spoke to called this 'the most underregulated frontier in HR tech.' Because there's a fundamental data asymmetry — the AI has access to your résumé, your interview feedback, your salary history if it's legal in that jurisdiction, and even sentiment analysis from your email tone. Luna: Wait — sentiment analysis? So the system is reading between the lines of what the candidate writes? Lucas: Exactly. One vendor in this space, I won't name them either, but they market a feature that detects 'candidate enthusiasm level' based on response time, word choice, and question types. If you seem hesitant, the AI might proactively offer a better perk — extra equity or a flexible schedule — before you even ask. Luna: That's almost like the system is preempting the negotiation. But if a human being were doing that, it would be considered a skillful tactic. When a machine does it, it feels like manipulation. Lucas: Right. And the big question is: should candidates know they're talking to an AI? Most companies using these systems do not disclose it during the negotiation phase. Luna: That seems like a legal time bomb. If a candidate later finds out they were negotiating with a bot and not a human, could they argue bad faith? Lucas: Some employment lawyers think yes. The argument would be that the company had an unfair information advantage — not just data, but the ability to simulate empathy without actually feeling it. But the case law doesn't really exist yet. Luna: So what does the actual interaction look like from the candidate's side? Does the AI sign with a name? Use human language patterns? Lucas: Typically it uses a real person's name — usually a recruiter or HR manager — and the tone is intentionally warm. The system I saw demo'd uses first-person pronouns, casual phrasing like 'I hear you on that,' and even varies sentence length to avoid sounding robotic. Luna: So it's designed to pass as human. That's the part that makes me uncomfortable. Lucas: And the vendors know it. One pitch I heard was that AI negotiators are actually less biased than human recruiters — they don't get tired, they don't hold grudges, they apply the same logic to every candidate. So in theory, it could reduce inequity in offer outcomes. Luna: But only if the training data itself isn't biased. If the historical offers the AI learned from had gender or racial disparities, the system would just automate that bias at scale. Lucas: Right. And that's the double-edged sword. The firm I mentioned did audit their model for exactly that — they found that their human recruiters had been offering slightly lower starting salaries to candidates from non-elite universities. The AI eliminated that gap. But that's because they checked. Not every company will. Luna: So the best-case scenario is a fairer, faster process. Worst-case, it's an opaque system that exploits information asymmetry and entrenches existing biases. Lucas: And in between, there's the practical reality: most candidates are not going to walk away from a good offer just because they suspect an AI was involved. So adoption will likely increase quietly. Luna: Is there any regulation on the horizon? The EU AI Act classifies employment-related AI as high-risk, but that's more about screening and evaluation, not negotiation specifically. Lucas: The EU AI Act does cover 'employment, workers management, and access to self-employment' — so a negotiation system could fall under that if it materially affects the terms of employment. But enforcement is still fuzzy. In the US, there's no federal law yet, though a few states have introduced bills requiring disclosure when AI is used in hiring-related communications. Luna: So for now, it's the wild west. What would you advise someone who gets an offer and wants to negotiate? Assume it's a person or assume it's a bot? Lucas: I'd say assume it's a system that has more data than you do. So do your own research on market rates. Use sites like Levels.fyi or Glassdoor. And be specific in your counter — if you ask for something concrete and well-justified, the AI will likely flag it as a reasonable request rather than try to lowball you. Luna: That's actually empowering. The more rational and data-driven you are, the harder it is for the AI to say no. Lucas: Exactly. The AI is optimizing for acceptance within a budget. If you make your case convincingly, you become the optimal candidate to say yes to. Luna: I wonder if we'll eventually see consumer-side AI agents that negotiate back on behalf of the candidate. Like two bots haggling over your salary. Lucas: That is absolutely coming. There are already startups building personal negotiation assistants that you can authorize to counteroffer on your behalf. I expect we'll see the first lawsuit over a bot vs bot negotiation within two years. Luna: It's a very strange future. But also kind of inevitable. The question is whether we build guardrails before or after the first scandal. Lucas: History suggests after. But I'm hopeful that episodes like this — and conversations among listeners — can push for transparency sooner rather than later. Luna: So the takeaway: if you're job hunting, arm yourself with data. And if you're a company deploying this tech, disclose it and audit it. Otherwise you're just asking for trouble. Lucas: That's a good summary. And as always, we'll keep watching this space.