Ghostbusters Team

An immune system for hypergrowth

Deel built an internal special-projects unit its founders call "Ghostbusters" roughly three years into a run from zero to $1B ARR in five years. Co-founder Shuo Wang describes it as a response to two coupled pathologies that arrive with scale.¹ The first is founder distance from individual contributors. As layers stacked (managers, directors, senior directors, heads), the founders stopped talking with ICs directly. Wang's tell was performative reporting: managers booking calendar time to tell her things she already knew. Her stated reaction was that she needed to know what was happening daily, and that the pattern had to stop immediately.

The second pathology is the pure people-manager. In Wang's account, managers whose only job is managing people "create a lot of levels and unnecessary projects," then seek larger scope, which pulls in more headcount and more non-critical work. Deel's counter-rule, as she frames it, is that managers must be experts who also do the work, not people-managers. Alex Bouaziz, her co-founder, runs the same doctrine at the personal level, staying in the details rather than managing purely through reports.

What the unit does

A Ghostbuster embeds in any function and asks a fixed set of questions: what are you working on each day, where is the low-efficiency part, why are you not productive, and is there a project I can help with without hiring additional people.¹ The last clause is load-bearing. The unit is explicitly framed against the reflexive "we need more people" answer, hunting the underlying inefficiency instead.

Two structural choices define it. Ghostbusters report directly to the founders (Wang, Bouaziz, or Dan Westgarth) rather than to the managers-of-managers, so they move fast and are not filtered by the chain they inspect. The unit also functions as a talent escalator: Wang's canonical example is an operator who joined as a junior SDR, moved into business operations, ran special projects as a Ghostbuster, and became head of crypto within about a year.

Structure over exhortation

The pattern's distinguishing feature is that it answers bureaucracy with an org unit rather than a cultural plea. Instead of asking a scaling company to "stay lean," Deel installed a standing team whose job is to find and remove inefficiency from outside the hierarchy it examines. In Wang's framing this keeps the company's execution speed alive as headcount grows, and it shares the instinct behind founder sign-off on hiring and skip-level sampling: keep founders in direct contact with ground truth as layers accumulate. It is the organizational counterpart to Bouaziz's personal practice of staying in the details, and to the broader Deel habit of refusing to accept a stated solution before understanding the problem.

The dependency it carries

A unit that bypasses the management chain and reports to founders can enable or corrode, depending on how it is received. Wang's telling frames the interaction as "what can I help with," but a team with a mandate to audit any department and escalate over its leaders can also read as a feared internal inspectorate. The model works at Deel because it is positioned as help rather than policing, which makes it dependent on founder judgment to keep it enabling. The pattern travels less as a rule than as a diagnosis: when a founder can no longer see the work through the management layers, the answer here was to build a team that could.