学会挨这一拳：Whoop 产品负责人如何在恐惧与自我里带队

Hilary Gridley, head of core product at Whoop, says her team's most common struggle is learning to "take a punch", meaning how to handle moments when something has gone wrong or someone is critical of your work. She sees fear and uncertainty everywhere right now, partly because of AI's effect on jobs and identity and partly because many younger people have never worked in a climate without layoffs. Managers, she says, now have to lead teams through fear and hard things, and she would love more people to attempt hard problems that may well fail.

Her core advice is to stop litigating what another person thinks of you and instead ask what you can do next to "counter-program" the narrative you fear they hold. She illustrates this with a meeting in which she laughed at the chief technology officer's suggestion to track ketamine use in the Whoop Journal, which the CTO had meant seriously. Instead of explaining herself, Gridley researched emerging public health concerns and within minutes sent a note proposing that Whoop track sports betting and correlate it with stress.

With her team she asks what they are afraid the other person thinks, challenges that thought by asking for evidence, and then asks what small action would show it is untrue. She says managers should not just validate negative spirals. She also allows for some healthy chip-on-the-shoulder motivation, but wants people to act on what they know to be right without fixating on what others think.

Gridley traces the idea to behavioral activation, a concept from cognitive behavioral therapy that she learned while building a depression therapeutic at Big Health. The misconception is that you feel better and then act, whereas therapists teach people to act first and then feel better. She keeps a list of such actions on her phone, and says the action can be as small as putting away a piece of laundry.

She sees the payoff at two levels. It helps in an immediate crisis, but it matters more for people who avoid putting themselves out there, for example by staying silent in meetings. She notes that each extra person in a meeting makes the room less candid, so attendees need to earn their place by saying something valuable. The blockers she hears, such as being too junior or processing slowly, are often rational-sounding cover for fear, and removing the fear is most of the challenge.

Gridley also shares a "hot take" that too many talented PMs play on easy mode, building for people just like themselves at companies that sell to other companies. She would like to see the same love of craft applied to hard problems such as services for low-income people, while acknowledging the prestige and pay that pull people elsewhere and offering a shout-out to those already doing that work.

Gridley says that when ten people must sign off on an email, the cause is a transparency and communication problem rather than a process problem, because each approver has different context and a different model of how the CEO thinks. A strategy document written six months ago becomes outdated quickly, especially with AI forcing companies to rewrite their plans. Understanding how a leader thinks is more useful than knowing what they decided, and when her team understands how she thinks, she does not need to approve their emails.

She builds this understanding by attending meetings with senior people, listening for "the note behind the note", and sending her team a short weekly Slack rundown with verbatim quotes, her interpretation and what she will do differently. When time is short she narrates her notes in a team meeting. Over time, the team develops a shared model of the CEO's priorities and the company's risk tolerance.

Her example is Whoop's CEO, Will, who obsesses over pixels and keeps repeating that things must "feel like the future". Gridley says this is often misread as demanding maximal scope, whereas she hears a call to use Whoop's data and AI coach in small, elegant, high-impact touches, such as explaining VO2 max with the member's own data. She connects feedback from several design reviews and tells her team why it matters to him.

She adds that, even when she disagrees with a decision, she asks what she might be missing and what would have to be true for the other person to be right. People who say they have no idea what is going on in a company tend to do the opposite and hunt for holes.

Asked how to carry out a leader's decision you disagree with, Gridley first runs a "what if I'm wrong" exercise, since a hunch can reflect years of judgment or confidential context. If she still disagrees, she is candid but respectful, and separates her opinion from an explanation of the rationale that makes sense to the decision-maker. She avoids telling her team that the decision is stupid but must be done. Her view is that in product there are no right answers, only the least wrong one, which you must execute well.

To understand how someone thinks, she uses what she calls magic questions, which are actually statements ending in "Do you agree?" or "Is that right?". She learned this working with legal and compliance teams who answered "it depends"; proposing a concrete scenario and hearing yes or no revealed their reasoning. She asks her own team to rephrase "what should I do?" into "here is what I think I should have done, do you agree?" so they calibrate their judgment and do not become dependent on her.

Lenny worries this could feel manipulative, and Gridley responds that you must go in with real openness to being wrong. If the goal is to reach a yes rather than to understand, you will get bad intelligence.

Lenny shares an Airbnb story in which a designer was told to build something the internet had never seen before, which sounded absurd until it was viewed through the founder's worldview. Gridley, a former English major, says reading fiction taught her to ask in what world a person's statements make sense. She says this made her happier, since she used to go home frustrated that nobody appreciated her "perfect, unique, beautiful insight".

She recalls advice from a coach who was formerly Coinbase's chief product officer: the mistake people make reporting to a CEO is thinking the game is influencing the CEO. The real job is to understand the CEO's vision and operationalize it into the best possible product. In the story of work, she says, you are probably not the protagonist, and an organization where everyone pushes their own narrative becomes inefficient and painful.

Gridley insists this is not defeatism. Her fulfillment comes from the millions of smaller decisions where she can zig while others would zag, so that the product is different because she was in the job. She likens it to Jenga: knowing how a leader thinks tells you which pieces are immovable and where they may be uncertain, which is where your perspective can be influential. Lenny connects this to Jeffrey Pfeffer's view of how power works, and she says that if you fight about what the world will look like in five years, you will lose.

She also teaches her team to build compelling arguments, with data, qualitative evidence or both, because they have an obligation to advocate. Only if they have made the case well and it still fails should they assume there is something they missed.

Gridley says she is obsessed with habit formation and thinks behavioral psychology works better than an education-plus-accountability model. For AI adoption, she does not measure or enforce but asks how she is creating habits. Her recipe is consistency, low friction and reward loops, and she created a "30 days of GPT" list with one small task per day that she says leaves people far more confident.

She starts people with fun, low-pressure uses such as vacation ideas or talking points for meetings drawn from their calendar, because adding friction to deadline-driven work backfires. The most important element, she says, is the reward loop: it must be powerful, immediate and emotional, so that people feel "like a million bucks" when they do the thing. Custom GPTs help because the builder designs the prompt, so users upload a document and get a useful result without the frustration of poor prompting.

In practice, she gives shout-outs in team meetings when someone uses AI to solve a problem they would not have before, and lets them demo it. Lenny summarizes the model as consistency, low friction, and a reward loop that is powerful, immediate and emotional.

Gridley points to WHOOP's recovery score, a daily red, yellow or green rating, as a product example. Members learn that drinking leads to a red score, and she says people tell her they got a handle on their drinking only after they got WHOOP, even though they knew the facts before. The emotional jolt of the red or green score appears to override what drove the behavior in the first place.

The harder problem is that healthy changes today pay off in decades and feel bad at first. WHOOP's new Healthspan feature tries to close that gap with a visual called the Amoeba, which shows your WHOOP age and changes color as your sleep, VO2 max, heart-rate zones and strength training change each day.

She applies the same thinking to her team's culture. Bosses often inadvertently reward staying up until 2 a.m. to finish something, so she looks for ways to celebrate people who take care of themselves, for instance having a PM who teaches fitness lead a quick stretch before long meetings.

Asked how she makes space for creative and deep work, Gridley says she mostly models it. She understands what each person needs to be at their best, whether that is fitness, teaching or, for her, illustration, writing and reading, and she talks openly about her own pursuits. She sometimes runs a book club, and in one-on-ones she asks what people do for joy each day, calling a "no" a problem to solve.

She argues that the busyness many PMs feel is partly self-inflicted and that leaders must give a permission structure for people to take their time back. She compares this to athletes, who accept that recovery is necessary and that running at 100% leads to burnout and worse performance. Active rest and heads-down time are, in her view, well documented and still skipped out of self-sabotage.

Gridley says she is regimented about this because she knows she would otherwise fall apart, quoting a line about having exercised the demons from your head while they "are outside doing pushups". Lenny connects it back to behavioral activation: do the thing that will make you feel the way you want to feel.

Gridley says people are not creative enough about learning with AI. Entry-level jobs are inefficient by design, but they give the reps that build judgment and taste, and she credits her early social media work for her ability to cut text in half. People worry that if companies stop hiring juniors because AI can do the tedious work, in five or ten years there will be no cohort that learned judgment.

Her answer is that the grind-wait-for-feedback-try-again model is slow, and AI can shrink the loop. She builds GPTs that think like her so her team can get feedback that is at least 80% close to hers, on demand and as often as they like. Lenny mentions studies of students using AI tutors, and Gridley says that skills still need investment, only not in the historical way.

She describes a GPT called Aristotle that generates LSAT-style logical reasoning questions framed as PM scenarios, with a version specific to WHOOP and consumer health. Another idea is a GPT that asks you to t-shirt size proposed scope and explains why integrations tend to be complex, which would give a PM with no engineering background many more reps than the few they get at work.

Gridley names her most pivotal moment as the time her boss at Big Health left and she began reporting to the CEO. It was trial by fire, but it taught her to give leaders the benefit of the doubt and understand the pressure they are under. Her manager, Kelvin, recalls that she was terrified yet "absolutely knocked it out of the park", and says the key advice he gave her was that product leadership is a role where, if you are not in control of the voices in your head, "they will eat you alive".

She explains that every path forward has flaws that others can poke at, and that the existence of potential criticism does not warrant criticism. She also says that building a mental model of a person means understanding their background, and notes that she has tried feeding her notes about someone into an LLM to predict the criteria that person would use or the objections they would raise.

For her failure story, Gridley describes a depression therapeutic she worked on for about a year, which Big Health abandoned when it acquired a different one. The lesson is that there is always a shot clock, and the build-versus-buy question is always fair. She says this instilled a healthy urgency to ship, especially in an AI arms race.

Gridley describes WHOOP's new generation as moving beyond elite athletes to a health and performance companion for anyone. The company updated its mission for the first time, to unlock human performance and healthspan. She highlights actionable Healthspan insights, AI coaching, women's health and hormonal insights, a heart-health screener with blood pressure insights and ECG, a 14-day battery, and a waitlist for Advanced Labs.

In the lightning round she recommends fiction over business books, naming *East of Eden* and *The Sun Also Rises*. She says fiction teaches you to sit inside tension, and quotes Keats on negative capability, the ability to remain in uncertainties without irritably reaching after fact and reason. Product people need to provide clarity and also to accept that there is no perfect answer.

Other picks include Nathan Fielder's *The Rehearsal* and Zwift, whose "ghost" of your past self racing at your personal best is a reward loop she loves. Her motto comes from a Beavis and Butt-Head clip: if the parts that suck were not there, the cool parts would not be as cool. She closes with Derek Walcott's poem "Sea Grapes" and its tension between obsession and responsibility.