Jean-Denis Greze on Hiring, Product Differentiation and Leading Engineering at Plaid

Harry Stebbings asks Jean-Denis Greze about the biggest break in his career. Greze says joining Plaid to lead engineering gave him opportunities as the company grew from around 40 people and less than $10 million in annual recurring revenue. But he identifies his earlier Dropbox job as the lucky break: two interviewers saw strengths in an unconventional candidate despite mixed interviews and advocated hiring him.

Greze says a resume is a limited signal. Repeated job changes or a long period without growth can prompt questions, while projects, a GitHub profile or a blog may reveal curiosity. He wants the interviewing team to meet the person and ask whether their experience fits the actual work, rather than pursuing an abstract idea of the best engineer.

For the first 15 or so employees, Greze favors hiring people who solve immediate problems. When management needs emerge, he contrasts a more experienced “four-year” candidate with a “two-year” candidate who can lead the next stage. He has shifted toward the latter: a leader accustomed to a large team may no longer enjoy or excel at the work of a much smaller one.

When a leader's fit changes as a company scales, Greze says the person may grow under a new manager or may be better suited to a different role elsewhere. He urges founders to give people a fair chance rather than freeze them in an old image, but says waiting too long without addressing a gap can cost a company momentum.

He distinguishes being good at firing from making the conversation pleasant. In his view the essential work is identifying a mismatch, giving a person a chance to improve and making a considered decision for the business. He says humane delivery and an appropriate transition matter, while acknowledging that the person may still feel the decision is unfair.

Greze stresses that difficult feedback should precede a termination decision, so the person is not surprised by a concern the manager has concealed for months. Once a decision is final, he favors being clear and working on the transition. Stebbings distinguishes that earlier feedback from debating the decision during the termination meeting.

One hiring mistake, Greze says, is failing to define the role precisely. A data-science leader for fraud and risk, for example, may need both domain experience and the ability to think creatively. A broad title alone does not tell an interview panel what evidence to seek.

He assigns interviewers specific questions tied to the role instead of having every leader ask the same generic career questions. He also values comparison across candidates so the panel understands what strong performance looks like. After hiring, he recommends reviewing explicit outcomes within a few months rather than assuming onboarding alone means the person is succeeding.

Stebbings sees intense title negotiation as a warning sign. Greze says he has favored low-ego organizations where ideas matter more than titles, but adds a qualification: titles can carry real external value, including for people from groups the industry has historically treated unfairly. As a company grows, the cost of fighting every industry convention can outweigh the benefit.

He argues that many organizational choices have several workable options, provided hiring and culture match the choice. By contrast, a poor product strategy or the wrong market can determine the business's outcome. Speed matters because it gives a company more chances to learn, but in financial infrastructure, he says, teams must move quickly without breaking systems that handle people's money.

On remote work, Greze thinks the industry has spent too much time debating one universal model. Some people prefer remote work and others an office; a company should choose deliberately and hire accordingly. He considers in-person work useful for highly creative, uncertain work close to customers, while established teams with clear roadmaps may execute incremental work effectively remotely.

He reports no difference in a work-quality measure on his team but says belonging and enjoyment scores have fallen. Stebbings strongly disagrees about remote productivity, creativity and retention. Greze agrees that culture needs attention and suggests hybrid arrangements may preserve some in-person benefits, while maintaining his view on productivity for well-defined work. These are their experiences and opinions, not a controlled study.

Greze says startup-building knowledge has spread well beyond Silicon Valley, even though more experienced operators remain concentrated there. In 2023 he observes a strong generative-AI community in San Francisco and considers proximity an advantage, while saying he does not know whether building a successful company elsewhere is necessarily harder.

Discussing work-life balance, Greze offers a provocative interpretation: when an industry becomes more accommodating of different schedules, people who choose to work exceptionally hard may stand out. He compares this with his impression of French work culture. These are his generalizations and personal views, not evidence that long hours guarantee better results.

He says high performance can come in different forms and that people can have fulfilling, well-paid careers without making work their sole priority. Asked whether someone can build an exceptional company with work-life balance, he recognizes leaders who delegate and motivate effectively, but believes such cases are uncommon. Stebbings largely agrees; neither presents this as a universal rule.

Stebbings asks why Greze doubts product differentiation is always durable. Greze says cloud products made large improvements over older software: he contrasts accepting payments before Stripe with integrating its API. In his analogy, the first improvement moves a product from one to 90 out of 100, while later refinements may move it from 90 to 93; the numbers are illustrative.

He argues that interfaces and features can be copied and that an integrated suite can combine good-enough versions of specialized tools. He cites Rippling as an example of a company pursuing integration. His prediction is that, in some mature software markets, coordination and a lower combined cost may matter more than a small advantage in each individual interface.

Stebbings asks whether new interfaces such as spatial computing and AI could create another large leap. Greze agrees that these technologies might change the picture, but says he does not yet know. His earlier argument concerns established product waves, not a claim that future user-experience innovation has ended.

Greze says Salesforce remains an important data store but expects challengers to attack parts of its customer experience. Asked for a view, he says he is bearish on its ability to remain number one over the next decade. This is his personal, forward-looking opinion in 2023, not a demonstrated outcome or a recommendation.

On a North Star metric, Greze says measurement helps but cannot make the difficult product decisions for a team. Plaid tracks whether someone attempting to connect a bank account can connect the right one for the use case. Improving that conversion can involve both incremental changes and longer, uncertain projects.

He compares the approaches to traveling from Paris to Milan: walking produces visible daily progress, while taking a flight may initially look like moving away from the destination and waiting. A dashboard can favor the predictable walker even when a more ambitious project could ultimately have a greater payoff. Metrics need counter-metrics, leading indicators and judgment.

Greze says Plaid achieved its annual conversion goal in six months largely through incremental improvements, while continuing to pursue a larger opportunity. He acknowledges the difficulty of deciding how much to invest in near-term wins versus work that may create a more durable advantage. His account of Plaid's performance is not independently verified and he has a commercial interest in the company.

For an early startup, he warns that focusing excessively on competitors can distract from finding a product and route to market. At greater scale, competitors are more consequential and distribution, reputation and sometimes network effects become important. He says his own angel investing favors founders and ideas and acknowledges that his approach is not optimized solely for financial returns.

Greze argues that venture investors fund many founders in the hope that a few achieve outsized outcomes. A founder with years of runway but weak product-market fit should ask whether continuing is the best use of their time, whether to change direction or discuss returning capital. He describes these as difficult choices for a venture-funded business, not proof that a slower-growing company has no value.

He treats product-market fit as a matter of degree rather than a binary state. As funding costs and cash-flow priorities change, a product that once seemed valuable may fall down a customer's list. He uses Snowflake as an example, saying his team would scrutinize whether storing additional data justified its cost under a higher hurdle rate.

Greze interprets apparent slowing in existing-customer spending growth as consistent with tighter customer economics. The example is his reading of the company and his own purchasing reasoning, not an independently verified explanation of Snowflake's financial performance. Asked for a market view after hearing an approximate valuation, he says “short,” then adds that he is a long-term S&P 500 investor; it is a 2023 opinion, not a personalized recommendation.

On generative AI, Greze says he lacks a reliable mental model of what the technology can and cannot turn into great products. It can handle tasks such as ambiguous language and image generation that software previously struggled with, but he wants to understand the limits. He regards developing that intuition as important for product and engineering leaders.

Reflecting on career mistakes, he says he learned during four years as a lawyer but regards the detour as costly in time. At Dropbox, he also led work on a Slack-like communication product that failed to find its way; he believes the team was too distant from customers. He says both experiences taught him lessons.

In the quick-fire round, Greze says angel investing has shown him how hard it is to predict a founder's capacity to grow. He names Rupa Health and Atlas among investments whose founders and missions interest him, explicitly revealing his financial relationships. He also urges listeners to check whether the facts behind generic startup advice actually match their own situation.

He speaks favorably of First Round and Sequoia based on his experience and relationships. His career advice is to look for work that creates impact for the business even when leadership is not paying attention. He wants people who worked with him to feel they did their best work up to that point; Stebbings thanks him and closes the interview.