Rethinking Team Building: How a 30-Person Startup Serves 50 Million Users — Grant Lee, Gamma

A blank slide looks like an invitation to develop an idea. In practice, it often starts a different kind of work: designing boxes, aligning them, resizing them, and sorting out layers. Grant Lee encountered that frustration in consulting. It became the starting problem for Gamma, where he is co-founder and CEO: make creating and sharing content dramatically simpler.

Gamma’s response was to rethink the building blocks so that people could focus on content while retaining the tools to mold and shape its presentation. Content first does not mean giving up expressive control; it means making that control easier to use. The longer-term ambition is to help people stretch, shape, and share their ideas—tools for imagination that might, in turn, help more ideas become useful to others.

The same willingness to question familiar product conventions can apply to building a team. Founders routinely reason from first principles about what a product should do. Organizational design deserves that attention too, rather than automatically inheriting the structure of the companies that came before.

The conventional expansion pattern is recursive: hire a VP, who hires directors, who hire their own reports. Repeat that across functions and a small startup can become a large organization quickly. That pattern supplies a familiar way to scale, but it also makes each leadership hire the potential beginning of another layer of headcount.

At the time of the talk, Lee reported more than 50 million users and a team that had only recently reached 30 people. Those are reported users, not a stated count of active or paying users. Gamma’s experience is an evolving operating model, organized around three questions: whom to hire, how to manage, and how to prepare for growth.

Staying small still involves hiring. Lee estimated that Gamma had roughly one-tenth the headcount it might have needed if founded a few years earlier. That is a counterfactual estimate, not a measured productivity comparison. The aim is to change how much organizational expansion growth requires.

A full-stack engineer is a familiar example of a generalist, but the pattern extends beyond engineering. Gamma’s first hire, who became its head of design, combines visual design, coding, core UX work, and user research. Those capabilities let him connect a user’s needs with both the interface and the engineering required to deliver it.

Knowing how to code gives a designer more than the ability to produce a prototype. It builds an understanding of what an engineering counterpart can actually implement. A prototype can then embody something the team can ship to production, supported by practical knowledge of the constraints rather than just an attractive interaction.

The role also changes with the product. Initially, the design problem was to find the simplest useful UI and UX. As the product grew more complex, the work became a repeated loop of coding prototypes, putting them in users’ hands, running tests and interviews, collecting feedback, and iterating. At Gamma’s later scale, the same designer also provides guidance and mentorship across the team. Generalism includes the willingness to adapt the job itself.

Continuous learning needs a counterpart: teaching. Gamma asks candidates to teach someone else a new skill during the interview process. The exercise looks for domain depth, the ability to articulate an idea, and the ability to bring other people into that understanding. Breadth is useful when it connects real expertise to work across the team.

In American football, a quarterback or linebacker can read what is happening on the field and make an adjustment without waiting for every instruction from the head coach. That is the useful part of the player-coach analogy: the person making the local decision is also participating in the action.

Rapid changes in AI create a similar need to adapt. If every reprioritization requires a top-down mandate, the people closest to a new capability or constraint cannot respond directly. Lee describes all of Gamma’s core leaders as player-coaches. Its engineering leaders bring substantial management experience while continuing to code and participate in the daily work.

That participation supplies context for coaching and decisions. A leader who understands the work in flight can help reprioritize it, weigh technical trade-offs, mentor a colleague, and connect present responsibilities to longer-term career development. The model preserves technical expertise while still providing management support. Lee’s qualification is consequential: it was working for Gamma at the time, but he did not know how it would scale.

Preparing a small team to grow also means investing in brand and culture. Lee treats them as two expressions of the same values: culture is how the company operates, and brand should reflect that behavior. Each new person has a proportionally large influence on a small team, so a mismatch in values or working practices is harder to absorb than it might be in a larger organization.

Gamma maintains a living culture deck rather than treating culture as a founding document that stays fixed. The team repeatedly rewrites it around the values visible in how people behave, shares it with existing employees, and uses it to onboard new hires. The deck is an ongoing effort to make the team’s operating assumptions explicit.

Continuity preserves knowledge that would otherwise need to be rebuilt through retraining and repeated onboarding. Lee associates that continuity with productivity, transparency, and shared context, without assigning it a measured effect. Gamma reinforces it through three standing all-company meetings each week.

The show-and-tell sessions use Gamma itself, so sharing progress also means using the product as a customer would. These rituals make work visible across the company without requiring everyone to be directly involved in every project.

The purpose is to preserve the feeling of a small room working toward a large, long-term ambition. A one-person, billion-dollar startup may become possible; Lee questions whether it is desirable. Building with other people is part of the reward, and preserving the enjoyment of a small team is an explicit goal.

The question that began with a blank slide therefore extends to the company itself: could there be a better way? The familiar scaling playbook remains an option. The invitation is to test other ways of building teams and share what works, rather than assume that the inherited structure is the only viable one.

Asked what he would change about building the team amid rapid AI progress, Lee returns to the earliest work of finding a useful product. Gamma began roughly four years before the talk, before the recent AI wave. With or without AI, founders still have to navigate the idea maze: discover actual user needs and determine which problems deserve solving. Faster implementation makes it tempting to skip that discovery.

Reflecting on Gamma’s first AI launch, about two years before the talk, Lee wishes the team had taken more time to appreciate how quickly the landscape was changing before moving into continuous building. Infrastructure choices deserve attention early because they can become extremely difficult to unwind at scale. His advice is to be thoughtful about consequential decisions, not to slow every part of the company down.

A follow-up question asks for a specific infrastructure decision. Lee identifies experimentation infrastructure, but not an architecture he would reverse. Gamma had already invested in experimentation early, gradually adding to it. He would have put more effort and weight behind that investment sooner because experimentation supports velocity, especially with a large user base. He remains unsure whether doing so would have changed the outcome.

Adding communications, legal, or other specialists raises another question: how can an engineering-led company preserve its culture without making those people feel excluded? Lee’s response starts with hiring judgment. Founders or other leaders should first try doing the unfamiliar job themselves.

Lee initially handled marketing, sales, and customer experience. He had not previously hired for those functions, so doing the work gave him a baseline for understanding what good performance looked like. Even doing a job imperfectly exposed its nuances and made it easier to recognize someone who could do it well. Gamma then looks for player-coaches in those functions too, limiting the tendency for each hire to trigger a cascade of new teams. This is an approach to entering unfamiliar functions; Lee presents it as something Gamma is still learning.

Player-coaches need ownership and initiative, neither of which is reliably visible on a résumé. To investigate those qualities, Lee asks experienced candidates about their most challenging project or problem and how they solved it. The useful evidence lies in the account of the work, not just the finished result.

His follow-up sequence is a practical interview procedure:

1. Start with problem understanding. Ask how the candidate determined what was actually wrong before choosing a solution.
2. Go beneath the initial framing. Explore how the underlying problem differed from what appeared on the surface.
3. Reconstruct the attempts. Ask about the solutions they explored, not only the one that ultimately worked.
4. Keep asking why. Follow the reasoning into second- and third-order explanations.

Lee uses this depth as a signal of personal ownership: someone who explored the problem should be able to explain its layers and the choices made along the way. It is a hiring heuristic, not a universal test of agency.

The final hiring question concerns failures. Lee identifies a recurring combination: the role was ambiguous, and Gamma could not conduct a work trial. A trial lets a candidate perform the actual job before the longer-term hiring decision. It is easier to arrange with someone between jobs or already doing fractional work.

Without that shared experience, both the company and the candidate could enter a role without knowing what a good fit would mean. Lee describes possible work trials ranging from two days to three months, with Gamma defaulting to three months. He especially recommends them when the founders have not done the job themselves. The trial makes the work available for both sides to evaluate while the definition of the role is still uncertain.

Lee reported that all of Gamma’s five-plus work trials had worked out, while emphasizing that this was only a few data points. For ambiguous roles hired without trials, he described a high failure rate but supplied no numeric rate. Those observations do not establish a controlled comparison. They do identify the uncertainty Gamma was trying to resolve: whether either side understood the job well enough to recognize a good fit before committing to it.