The ATM Heresy
Every era has its relationship the experts swore could never be automated. We VCs are not the first.
In 1969, Chemical Bank installed the first widely-cited American ATM at a Long Island branch. The slogan on the launch-day window promised that the bank would never close again. But selling the machine to other banks was a slog. Don Wetzel, the inventor, would later recall the reaction he got from his prospects across the country.
“Most everybody we talked to, the bankers, they thought, ‘Man we got tellers. We don’t need a mechanical teller.’”
The first machines didn’t come cheap. Wetzel could have positioned them as labor-replacement. He found out quickly that it was the wrong pitch.
“We never sold any banker on you can justify the cost of this machine because you can get rid of a teller or two, never. It was always on the premise that this will be better for your clients, you’ll get more clients.”
The argument the banks raised against ATMs was relational. The teller was the bank’s relationship with the customer. The bank lobby, the eye contact at the window, the personal banker who knew the depositor by name. You could not, the conventional wisdom went, replace any of that with a machine on the wall.
Then in January 1978, a blizzard dumped seventeen inches of snow on New York City. The banks closed. The ATMs didn’t. ATM use jumped twenty percent that month, and the resistance broke. By 1980 there were roughly fourteen thousand ATMs nationally, and adoption only accelerated from there. The teller relationship did not vanish, but the job changed. The teller stopped processing cash and became, in the historian James Bessen’s phrase, “more of a marketing person.” Branches grew. Per-branch teller counts fell. Tellers cross-sold mortgages and credit cards. The “personal banker” relationship was the explicit thing that survived. Banks even named the machines “Tillie” and “Buttons” to soften the loss, akin to today’s cute naming of AI agents. That survival held for roughly thirty years. Then the iPhone made the personal banker self-serve too, and branch counts started falling. The relationship migrated up the value stack, and waited.
The travel-agent industry got the same letter, twenty-five years later.
Microsoft launched Expedia in October 1996. The American Society of Travel Agents responded in Travel Weekly:
“There may be a small percentage of do-it-yourselfers who want to book electronically, but most people think their time is too valuable.”
A year later, ASTA’s senior VP for legal and industry affairs told the Senate Judiciary Committee that consumers “should not have to run a gauntlet of electronic gatekeepers” to reach their local travel agency’s website. By 2002, the major US airlines had cut their base agent commissions to zero. The agent share of US airline tickets had fallen from sixty-seven percent in 1999 to forty-six percent in 2002, per GAO. Bill Maloney, ASTA’s chief operating officer, sat with an interviewer that May and said the inversion out loud while it was still happening.
“Just as you saw the demise of general practitioners in medicine, you are increasingly seeing travel agents who are specializing.”
He was right. By 2025, US travel-agent employment was sixty percent below its 2000 peak, per BLS and Census data. The survivors were the luxury consultants, the virtuoso affiliates, the corporate travel managers, and the complex-itinerary specialists. The commodity agent died. The high-touch agent kept going, in smaller numbers, charging fees that the airlines used to pay through commissions. The relationship survived only at the tail.
Two stories, three decades apart, identical shape.
In both cases the industry’s defense was that the relationship was irreducible: customers wanted human contact, the slowest-adopting customers would always come back to us, the share we held today proved the model. In both cases that defense was sincerely meant, partly true, and strategically wrong. The unit count of relationship-based professionals fell by something like sixty percent over a generation. The relationship as a concept lived. The relationship as a job for most of the people who used to do it did not.
My previous post, The FICO Heresy, made the same case about decision automation: the loan officer’s gut, the captain’s hands on the yoke.
Venture capital is getting the same treatment now. Founders want a human in the room. The warm intro is the moat. Model output cannot read a founder the way a partner can. These are the arguments the ATM skeptics raised in the late 1960s and the travel-agent skeptics raised in the late 1990s, in different words.
The fairest version of the pushback is that venture relationships have more depth than a teller transaction or a Tuesday-night Cancún booking. A founder is making a ten-year commitment, not buying an airline ticket. Conceded. But Bill Maloney conceded the same in 2002 about luxury and corporate travel, and Wetzel’s banker prospects could have said the same about commercial lending versus walk-in cash withdrawals. The depth defense holds for the depth tier. It does not hold for the commodity tier, and the commodity tier is bigger than the venture industry admits. The warm intro, the first meeting, the routine diligence call: each of these is automatable. What survives is the high-touch tail: governance in a crisis, the lonely non-consensus bet, the founder-coaching call at midnight on a Sunday. That tail is real, and it will survive, perhaps for a while. Yet it will employ a fraction of the partners working in venture today.
The general practitioner is dying. The specialist is surviving. Maloney said that from inside ASTA, watching it happen inside an industry being unmade. The personal banker that grew out of the ATM transition was a new role, named after the fact, that nobody in 1969 could have described. In a similar vein, Expedia and Booking were not staffed by travel agents.
The shape that survives an automation transition is rarely a shape the existing practitioners can name from inside it. The firm that emerges from an automation transition is rarely the one the existing practice would have designed for itself.
The founder relationship matters. It does, and it will.
But the question for any venture firm is not which kind of practitioner to be. It is whether to build the firm that looks entirely different than the one you are sitting in now.
