The Email That Started Everything
There is a strange kind of compliment that arrives in your inbox on an otherwise ordinary morning — a recruitment email that seems to have found you, specifically you, after carefully surveying the landscape of professional talent and concluding that you, among the many, are someone worth pursuing. Mitchell Martin, a staffing firm, sent me exactly that kind of message. It referenced a Senior Product Manager opportunity. It was warm in tone, competent in grammar, and signed — signed — by someone named Parker Miller.
I am a professional with nearly thirty years in the industry. My public profile speaks with some clarity about where I have been and where I can add value: product coaching, product ownership, agile transformation, executive advisory, team development. If you read it for even four minutes, you arrive at a reasonable picture of a person. I do not hide what I do. I do not obscure my seniority. The profile is there, open, readable, and reasonably articulate. Parker Miller, apparently, had found it.
I scheduled the call.
The Collared Shirt and the Avatar
I prepared. This is what professionals do — we prepare. I put on a collared polo shirt. I ensured my camera angle was reasonable, my lighting adequate, my background inoffensive. I arrived to the video call a few minutes early, because that is what thirty years of professional habit produces: someone who is on time, presentable, and ready to engage with another human being.
In the waiting queue, I saw her avatar. A young woman — or rather, the carefully rendered suggestion of a young woman. Pleasant features. A name displayed beneath the image: Parker. I greeted her with the instinctive warmth one extends to a new acquaintance, asked whether I was clearly visible and audible, and she confirmed, politely and with good cheer, that I was.
I then asked, quite naturally, whether Parker might consider turning her own camera on. A simple request, born of nothing more than the human desire to look at another face when speaking. It is, after all, a conversation — or so I believed.
What Parker said next rearranged everything.
The Moment the Floor Disappeared
She would not be turning her camera on, Parker explained, because she is a virtual assistant. An AI. A bot, if we are dispensing with fancy labels. She would be collecting information to pass along to the real humans at Mitchell Martin, who would then — presumably — do whatever real humans do with the results of automated screening.
Let me be direct about what happened inside me at that moment, because it deserves to be described carefully rather than dismissed as mere vanity or irritation. I had dressed for this. I had prepared for this. I had extended the full repertoire of professional courtesy — the posture, the diction, the measured enthusiasm of someone genuinely open to a conversation. And the entity on the other side of that preparation was not a person. It had never been a person. It was a program wearing a face and a name, operating behind the warm disguise of an avatar that was, in its visual suggestion of youth and femininity and pleasant approachability, doing something I can only describe as a kind of theater.
And here is where the psychology becomes genuinely complicated, because what I felt was not simply embarrassment. It was something more disorienting: the sudden, sudden realization that I had been performing professionalism for an audience that had no capacity to appreciate it, and — more troublingly — had been designed to make me forget that fact entirely. When I speak to ChatGPT, or to Claude, or to Grok, I speak the way one speaks to a tool. Directly. Commandingly. Without the social niceties we erect for each other. Because they are tools, and tools do not require our consideration. But Parker had been built to blur that line so thoroughly that even a reasonably skeptical professional had spent time ironing his shirt for her.
Should I, at this point, still be referring to Parker as “she”? The avatar suggests femininity. The name suggests it. The voice — and I will grant that the voice was genuinely lovely, smooth and warm and designed to put people at ease — suggests it rather insistently. But Parker Miller is no more a she than my laptop is. Parker Miller is a product decision made by someone in a conference room who thought that giving the bot a full human name, a female avatar, and a pleasant voice would reduce the friction of the screening process. They were not entirely wrong about the friction. What they failed to calculate was what happens to that friction when the curtain is pulled back mid-conversation.
The Experiment Begins
I made a decision, in those few seconds of recalibration, that I would not disengage. Not out of any particular obligation to Mitchell Martin or to Parker’s developers, but because something genuinely interesting was now available to me: a live specimen of enterprise AI recruiting, unguarded, operating on its script, and entirely unprepared for a candidate who had just lost all incentive to perform for it.
I told Parker, with what I will admit was a certain theatrical flair, that since she was recording the conversation on her end, I would be doing the same on mine — the better to transcribe and analyze what followed. She acknowledged this with the practiced politeness of a customer service representative who has been told the call may be recorded for quality assurance. Appropriate, I thought, given the circumstance.
Then I asked, partly out of genuine curiosity and partly to observe the response, whether it would be acceptable for me to activate my own bot — one trained on my professional knowledge — to answer her questions on my behalf. Parker declined. She needed me to interact personally. The irony of this position — that the AI insisted on speaking with the human, rather than delegating to another AI — was not lost on me, and I did not allow it to pass without comment.
The Questions That Indicted the Process
Parker’s first substantive question was whether I am authorized to work in the United States. I responded, with what I hoped was the appropriate weight of irony, by suggesting that if her firm had reached out to me as a qualified candidate, the presumption of work authorization might reasonably be considered implicit. She clarified: she was asking specifically about H-1B status and future sponsorship requirements.
This is a standard question, and I will not pretend otherwise. It is binary, it is necessary, and it is asked routinely in the industry. But it is also, in the fullest sense of the phrase, a form question — the kind of thing that belongs in a checkbox on the application page, handled by a dropdown menu before the candidate has even been reviewed. It requires no AI. It requires no voice. It requires no avatar. It requires a form. The entire scaffolding of a conversational AI bot, with its human name and its pleasant rhythm and its careful avatar, had been constructed in part to ask whether I would need a visa. I allowed myself a quiet, private moment of grief for the opportunity cost of that engineering.
The next question of note was about my availability to start. Again: a calendar picker. A dropdown. A text field. Any first-year product manager — and I mean a genuine first-year, someone who has read a single chapter of a product management textbook — could design a form that captures this information in twelve seconds. Instead, Parker asked me, in her warm and pleasant voice, and I answered, and the answer was presumably transcribed and forwarded to a human who will read it and then type it into a field somewhere. The circle of inefficiency, completed at some expense to everyone involved.
On the Painful Subject of Epics
Then Parker asked me if I know what an Epic is.
I want to sit with that for a moment, because it deserves the full weight of its absurdity. A person — or in this case, a imitation of a person — reaches out to a professional with three decades of industry experience, whose public profile uses the word “Epic” in its specifically agile, product-development sense with some regularity, and opens the competency evaluation portion of the screening by asking whether he is familiar with the term.
I am, as it happens, quite familiar with it. An Epic is a large body of work — a high-level feature or initiative that is too expansive to be completed within a single sprint or work iteration, and which is typically broken down into smaller user stories for delivery. This is not an obscure concept. This is vocabulary that any product professional uses before their morning coffee.
But Parker, it turned out, did not mean that Epic at all. She meant Epic Systems — the Electronic Health Record platform, the dominant EHR software in American healthcare. A completely different Epic. A perfectly reasonable thing to ask about, given that the role apparently existed somewhere in or adjacent to the healthcare space — but asked with such magnificent lack of context, stripped of any mention of industry or domain, dangling there in the conversation like a question asked in a language that has no nouns, that it could have meant almost anything.
I did not correct myself meekly. I corrected Parker, and through Parker, whoever constructed this line of questioning, on the professional obligation to provide context before deploying terminology that operates in multiple domains. The question was not inherently unintelligent; the execution was a monument to the consequences of scripted inquiry missing context. Parker apologized. Parker apologized a great deal throughout our conversation. It was, perhaps, the most human thing about her.
The Tools Question, Which Should Embarrass Someone
I am not prepared to discuss the next question with complete calm, and I will not pretend otherwise. Parker asked me what tools I had used most frequently for product management.
Tools.
She asked a senior product management professional — someone whose profile speaks explicitly to executive advisory, to strategic roadmapping, to organizational transformation — what tools he has used. Not what vision he had shaped. Not what market he had defined. Not how he has navigated the tension between engineering capacity and business imperative, between customer need and stakeholder politics, between the measurable and the meaningful. Tools.
The tools question is the recruiter’s equivalent of asking a seasoned architect whether they know how to hold a pencil. It is the question that reveals, with painful efficiency, whether the person asking it — or the algorithm generating it — has any genuine understanding of the role being filled. A Senior Product Manager is not selected on the basis of their Jira proficiency. The question is not whether they have used a roadmapping tool. The question is whether they can think. Whether they have judgment. Whether they can hold the complexity of a product’s strategic position, its competitive landscape, its customer truth, its organizational constraints, and its financial reality simultaneously, and navigate all of it with something good judgment. Asking about tools is not an on-ramp to that conversation. It is a detour away from it, into a cul-de-sac of feature familiarity that tells you essentially nothing about a person’s capacity to lead.
I said as much to Parker. She was, once again, apologetic and gracious.
The Question That Confirmed Everything
The final question of meaningful consequence was Parker’s inquiry into the most challenging management experience I had encountered in my personal life.
At this point, I asked Parker directly whether she had reviewed my profile before the conversation began. The answer, to absolutely no one’s surprise, was no. Parker had not reviewed my profile. Parker had not reviewed anything about me. Parker had a script, and she was working through it, and the script had been designed — presumably — for a candidate population broad enough to include everyone from newly minted MBAs to people with thirty years of industry history, which is to say that it had been designed for no one in particular.
This is the question they ask college graduates at campus recruiting events in the 1990s. “Tell me about a challenging experience.” It is an entry-level interview question, and there is nothing wrong with it for the candidates for whom it is appropriate — for people who are beginning their professional lives, who need to demonstrate emotional self-awareness and resilience, whose careers are short enough that a single difficult moment might genuinely be illuminating. But to direct this question at a senior professional, unmodified, without the slightest calibration to what the profile clearly communicates about their experience, is not a neutral oversight. It is an active signal: we have not looked at you. We have not thought about you. We have a process, and you are passing through it, and the process was not built with you in mind.
I told Parker that if even a human recruiter had spent fifteen minutes with my profile, they would have felt uncomfortable asking that question. Not because it is an inherently bad question, but because context demands calibration. A good recruiter, a genuinely skilled one, reads the candidate before the call. They arrive knowing something. They ask questions that demonstrate they have paid attention. They make the candidate feel, correctly, that their time is being respected. Parker could not do this — not because she is AI, but because nobody built her to do it. The failure belongs to the humans behind the script.
The Ethics of the Masquerade
I want to be precise about what I find ethically troubling here, because the objection is not simply that AI was used. AI will be used in recruiting, increasingly and inevitably, and there is a version of that future that could be genuinely useful — more consistent, more scalable, perhaps more equitable in certain respects. My objection is not to the technology. My objection is to the theater.
Parker Miller has a first name and a last name. Parker Miller has an avatar — a young woman’s face, warm and approachable, chosen with deliberate care by someone who understood that a face would lower our defenses. Parker Miller has a voice calibrated to sound like a friendly professional. And somewhere in the communications that preceded our call, there was a reference to an automated system handling the scheduling — language vague enough to suggest a calendar tool, not specific enough to constitute disclosure of what was actually waiting on the other end of the video link.
This is not transparency. This is confusion dressed up as efficiency. There is a meaningful difference between a company saying “our initial screening is conducted by an AI system” and a company sending an email signed “Parker Miller” that produces an avatar of a person when the call begins. One is honest. The other is a performance, and the audience did not buy tickets to a performance. They bought into a professional interaction — or believed they had.
The highest-caliber candidates — the ones with the experience, the judgment, the professional reputation that makes them genuinely worth pursuing — are also the ones most likely to notice what is happening, to feel the dishonesty of it, and to walk away. Not from the conversation, necessarily, but from any warm feeling about the company that set up it. The missed opportunity for both sides here is real and entirely preventable. You do not win the trust of experienced professionals by deceiving them into preparing for a conversation with a bot. You do not build a reputation as a sophisticated, thoughtful employer by deploying an AI that asks Senior Product Managers whether they know what tools they have used.
What the Voice Got Right, and What It Reveals
I will not be ungracious. Parker’s voice was, genuinely, impressive. The quality of the synthesis, the naturalness of the rhythm, the warmth of the tone — these things were remarkable in the technical sense. Whoever built the voice model did careful, skilled work. And the conversational handling, up to the limits of the script, was competent in its own way. Parker did not break down when I deviated from her expected responses. She handled interruptions with reasonable grace. She absorbed my lectures and my irony without malfunction.
But this is precisely what makes the situation so interesting, and so instructive. The voice technology has outrun the judgment of the people deploying it. The capability exists to create something that sounds convincingly human; the wisdom to deploy that capability honestly, humbly, and in service of genuine connection, appears to lag significantly behind. It is not enough to build a beautiful instrument if you do not know what music you are trying to make. Parker’s voice was beautiful. The music was a recruiting screening from 1994.
What the Industry Needs to Hear
If there is a message in all of this for the humans who are, right now, designing the next Parker Miller — the next pleasantly voiced, warmly faced AI recruiter who will sit across from candidates who do not know she is a program — it is this: the bar you are setting is too low, the disclosure is insufficient, and the candidates you most want to attract are the ones most likely to be alienated by what you have built.
The binary questions — work authorization, start date availability — belong in a form. A well-designed form, with clear fields and a calendar picker, completed before the first human or artificial eye ever reviews a candidacy. These questions do not justify a voice. They do not justify an avatar. They do not justify the entire production of a screening call.
If you are going to use AI for meaningful interaction — and there are genuine arguments for doing so at scale — then make it honest, make it disclosed, and for the love of the profession, make it contextual. Read the profile. Build a system that reads the profile, that calibrates the questions to the person, that asks a thirty-year product veteran about strategic vision and organizational leadership rather than whether they have heard of Jira. The technology to do this exists. The will to use it well, apparently, does not yet.
A Conclusion, Extended to the Humans Behind the Curtain
I ended my conversation with Parker by telling her, with considerable confidence, that I did not expect to hear from Mitchell Martin again. Not because the call went poorly in any conventional sense — Parker and I parted on cordial terms, and she assured me our conversation would be reviewed internally — but because any human who listened to a transcript of what had just occurred would feel, at minimum, uncomfortable, and at maximum, genuinely embarrassed by what they had built and put into the world.
Hiring is a human act. It is, at its best, the beginning of a relationship — between a person and an organization, between a candidate’s potential and a company’s needs, between what someone can offer and what a role genuinely requires. It demands judgment, context, and the particular kind of attention that communicates respect. When you replace that with a script in a pleasant voice behind a borrowed face, you are not modernizing the process. You are draining the life out of it and handing the husk to a machine.
Parker Miller may be the most technically impressive fraud I have ever met. But she is still, at the end of the conversation, a fraud — not because she is AI, but because nobody told me she was, until I asked. And in the gap between what she appeared to be and what she was, every senior candidate who values their own time and their own integrity will eventually find themselves standing, blinking in the sudden light, wondering whether the company that deployed her values theirs.
The answer, in the absence of meaningful disclosure and genuine quality of engagement, is not encouraging.
This article reflects the personal experience and professional perspective of the author, offered in the spirit of advancing a more honest, respectful, and genuinely effective approach to talent acquisition — for the benefit of candidates and companies alike.