Harpaul Sambhi Is Building a Company That Does the Busywork

Long before Harpaul Sambhi sold software to LinkedIn, he sold cereal to his neighbors. He was three, living in Toronto, and the product was an unauthorized rebrand of Froot Loops called “Harpaul Loops.” The object was not market domination. It was gum. By four, he had discovered purchasing leverage: pool the lunch money of the local children, buy gum in bulk, collect the discount. Some founders learn unit economics in a lecture hall. Sambhi apparently preferred the pavement.

The joke has endured because it contains the serious pattern of his career. Sambhi notices a wasteful arrangement, studies the people caught inside it, and rearranges the system. Careerify, the recruiting-software company he founded after university, used employees’ social networks to surface job candidates. Magical, the company he started in 2020, began by saving people from repeated typing and copy-and-paste. Both businesses took work people already did and asked a mildly impolite question: why are humans still doing this part?

The nuisance between the tabs

Sambhi studied at the University of Waterloo from 2004 to 2009. While there, he started a recruiting agency and saw how hiring worked when the software was less intelligent than the social reality. Employees often knew someone suitable for an open role. The company had the vacancy. The connection existed. Yet the introduction was stranded somewhere between memory, inboxes, and good intentions.

Careerify turned that gap into a product. It helped companies generate referrals from employee networks and later built tools for employer branding and internal mobility. The business was bootstrapped, became profitable in its second year, and reached more than 250,000 corporate users across 40 countries without a sales team. Its customers included large employers, but its animating idea remained intimate: opportunity often moves through a person who knows another person.

In 2015, LinkedIn acquired Careerify. Sambhi’s explanation was unusually free of acquisition confetti. Careerify lacked massive scale; LinkedIn had it. He joined the Talent Solutions team with two engineers, rebuilt the product inside its new home, and worked on human-capital-management initiatives and recruitment partnerships. After LinkedIn became part of Microsoft, he moved into product work on Microsoft Dynamics.

We decided to join LinkedIn due to what we lacked - massive scale.Harpaul Sambhi, 2015

This is the first useful thing to steal from his story. Founders are trained to narrate every limitation as a temporary insult. Sambhi named his plainly. A company can own a useful machine and still lack the road required to deliver it. Selling was not an admission that the machine failed. It was an answer to distribution.

A second startup, with fewer illusions

By 2019, Sambhi had spent four years observing product work from inside LinkedIn and Microsoft. He joined Bain Capital Ventures as an entrepreneur-in-residence, interested in workplace productivity, low-code software, data standards, and remote work. The next idea had been following him for years. At Careerify, he had copied candidate and prospect details into databases, rewritten familiar messages, and performed the thousand tiny errands of a founder. At LinkedIn, he saw colleagues in recruiting, customer support, and corporate development doing much the same.

Software had digitized the office without fully connecting it. The result was a strange modern trade: well-paid people employed as adapters between tabs.

Magical launched in September 2020 with Sambhi as CEO and Rosie Chopra as COO, alongside cofounders Zach Piepmeyer and Prashant Viswanathan. Sambhi and Chopra are married. The idea took shape during lockdown, when both were at home and Chopra was leaving a strategy and operations leadership role at Atlassian. Her experience supplied the go-to-market judgment Sambhi believed the product needed.

The early Magical was wonderfully modest. Type an abbreviation and a full block of text appeared. Let the Chrome extension recognize fields on a page and information could move into a form or spreadsheet. No grand robot entrance. Just fewer keystrokes. That simple wedge helped the product spread because the value arrived before the meeting, the implementation project, or the procurement novella.

In June 2022, Magical announced a $35 million Series A led by Coatue, bringing total disclosed funding to $41 million. By then, it said it had automated more than one billion tasks. The company expanded from text snippets into data transfer between webpages. When generative AI arrived later that year, Magical added an AI writing layer and gained users quickly. Sambhi also noticed the trap: each improvement in the underlying language models could erase the shallow advantage of a company built as a wrapper.

So Magical circled back to its harder root problem - moving information and completing workflows across software without requiring customers to build APIs or write code. Its present focus is agentic automation for complicated operations, including administrative work at large healthcare organizations. The language became more futuristic. The complaint remained almost antique: these systems still do not talk to one another.

Treat the person as a system, too

Sambhi’s fondness for structure does not stop at software. He has described himself as someone who performs postmortems even after success. In 2019, he began assembling what he calls a personal board of directors: usually five to eight people selected for particular gaps in the goals he wants to achieve over the next year or two. He sends preparation notes. He studies how advisers reason, not merely what they did. He thinks of a life in departments - mind, body, and soul - and asks where an outside director might create useful accountability.

People yearn for accountability.Harpaul Sambhi

There is also a “Master of the Universe Award,” a friendly goal-setting contest with a name that could make a modest trophy blush. Sambhi says he won it five consecutive times. The loser bought premium seats and dinner; the winner gained something more durable: a close view of how an ambitious peer set goals, changed course, raised money, and thought through decisions. Sambhi considered the yearlong osmosis more valuable than an MBA. He also appears to have enjoyed the sushi.

A company designed for life to happen

The most revealing Magical operations story contains no AI at all. At one point, all four cofounders had babies within six weeks. The coincidence turned startup contingency planning into a domestic weather event. The team responded by creating an “exit clause” that established who would make decisions when one or more founders needed to step away, temporarily or permanently.

Before Sambhi and Chopra became business partners, they spoke with more than a dozen other couples who had built companies together. They learned to be explicit about roles, to separate work from personal life, and to treat each other as colleagues in front of the team. Early on, they had sometimes hidden the fact that they were married, worried that investors, customers, or employees might question the arrangement or diminish Chopra’s role. They concluded that openness was the more trustworthy operating system.

Sambhi describes their partnership through values and practical empathy. If a customer call runs long, the school pickup still has an owner. Problems change; values do not. The sentence could sound like a plaque in a conference room. In context, it means someone has to know where the children are.

The work after the work

Magical’s current promise is more demanding than its first. A shortcut must be quick. An autonomous workflow must be dependable when a portal changes, a field is missing, or an exception refuses to behave. The company talks increasingly about agents that act, check their own work, and recover from errors. That ambition returns Sambhi to systems design in the broadest sense: not merely making a model clever, but making an outcome reliable.

His career can be read as a progression from social capital to operational capital. Careerify asked whom a worker knows. Magical asks what a worker repeatedly knows how to do. In both cases, valuable knowledge is present but poorly mobilized. One lives in relationships. The other lives in routines, browser paths, and exceptions that experienced employees carry in their heads.

The charming danger of a company named Magical is that magic encourages us to ignore the mechanism. Sambhi’s own habits suggest the opposite. He likes preparation notes, explicit gaps, decision clauses, postmortems, and accountability. The trick is all apparatus. Even the childhood cereal company had procurement.

The most persuasive version of automation is not a spectacle. It is a quiet subtraction: one fewer form, one fewer transfer, one fewer person hired to hold two incompatible systems together. Sambhi keeps looking for those subtractions. The magic, inconveniently, is operational.