Sahil Aggarwal and the cost of starting over

On Christmas Day in 2021, Sahil Aggarwal and his co-founders received a term sheet. They had gathered in Kasauli, in India’s Himalayan foothills, for planning sessions and hiking after months apart. Their young company, Rattle, had built software to spare salespeople some of the clerical work surrounding a sale. Now an investor wanted to finance its next chapter. A holiday in the hills had acquired a rather consequential attachment.

The company went on to announce a $26 million Series A in April 2022. A few years later, Aggarwal would decide that its product needed to be rebuilt around a different technology. The interesting distance in his story is the one between those decisions: accepting the money to expand a business, then accepting that expansion along the same path might no longer be the answer.

He is now building Von, an AI intelligence layer for revenue teams. Rattle remains essential to the story because it gave him customers, experience and something difficult to surrender: evidence that his original idea worked.

The product was the hobby

Aggarwal studied engineering at Thapar University and began his career at ZS Associates in New Delhi in 2013. Product work drew him away from consulting. His subsequent stops included Gigster, his own development business WinTerrain, the visitor-identification startup Leadworx and, in 2019, Mutiny in San Francisco. At Mutiny, he was employee eight.

Those names describe a working education in making software, selling it and seeing how people use it. His enthusiasm had a less corporate expression: “Even on weekends, I look at products.” There are hobbies that take a person away from work. Aggarwal’s appears to have supplied additional office hours, without the inconvenience of calling them that.

An engineering background is useful context, but the problems he keeps returning to involve people. A salesperson has a conversation. The conversation produces information. Someone must translate that information into the formats a company’s systems demand. The sale may depend on judgment and trust; the machinery around it still wants its fields filled in.

A company carried across borders

Moving to San Francisco added another sort of apprenticeship. Aggarwal has spoken about the help he received from Indian founders already in the United States. Their advice ranged from finding a house and leasing a car to structuring a board and understanding which advisors to bring into a business.

He has also said he would have started his O-1 visa application earlier. His account of moving countries is full of such practical details. The founder crossing an ocean in a company biography also has to work out where to live when he gets there. Grand strategy eventually needs an address.

His advice on finding early American customers puts similar weight on other people. Experienced founders, CEOs and advisors can help a newcomer understand a market and make useful introductions. A product demonstration happens inside a relationship, however brief. Arriving with someone who can explain the room makes a difference.

There is a connection between that account and the software he builds. In both, the official structure tells only part of the story. A company has systems and procedures. A founder has incorporation documents and a pitch. The actual work moves through people who know something, make an introduction or remember what happened last time.

Selling the permission to stay

Rattle began in 2020 with Aggarwal, Apoorva Verma and Milan Singh. Its premise was easy to recognise if you had ever finished a call and then faced the job of telling several applications what had happened. Bring the process into Slack or Microsoft Teams, where people were already working, and let the information travel to the underlying systems.

The early product took six weeks to build. The founders initially let customers use it free. After four months, they asked for payment; around 80 percent agreed. That gave the team a practical answer to a question every new application faces: had it earned a place in the working day?

By the 2022 funding announcement, customers included Miro, ClickUp, Rippling, Front and Clearbit. Investors in the Series A included Insight Partners, GV and existing backers Sequoia Capital India and Lightspeed India Partners. The company was building connections between messaging interfaces and the systems customer-facing teams depended on.

Aggarwal also paid attention to how that company appeared in public. He has described an early emphasis on people, product and brand, and a later preference for smaller gatherings where the conversation could be more focused. In his view, identity needed thought before the marketing engine accelerated.

“Ideas can change a lot. You should pick a name that will not pigeonhole you.”

Sahil Aggarwal, on naming a business

The advice has a certain bite in retrospect. A name can leave room for a new idea. A software architecture can be considerably less accommodating.

A February 2024 Rattle feature shows how specific that work could become. Team members were required to connect their own Salesforce accounts before interacting with alerts. A button in Slack let them make the connection without a trip through the integrations settings. It also meant a record change could be associated with the person who made it. This is the everyday territory behind the broader pitch: a click removed here, a clearer trail there. Someone still has to account for who changed a customer record. The convenience has to survive contact with the company’s rules.

The carriage problem

By the end of 2024, Aggarwal had concluded that adding AI to Rattle was failing to produce the business he wanted. He and the team had been supporting the existing product while attempting to build a new AI agent. The effort divided attention without delivering the intended result.

He has used the image of attaching an engine to a carriage to describe the difficulty. It is a useful analogy for the decision he faced: the new capability changed what a product could be, while the old structure still determined how it behaved.

The rebuild had a cost that a product diagram cannot show. The company shrank from 70 employees to 15, with sales, marketing and customer-success roles cut. Aggarwal has spoken about the difficulty of telling people who had joined one vision that the company had to begin again. The word “pivot” is remarkably economical for something that can rearrange so many lives.

Public accounts place the rebuilding work in 2025. By 2026, Aggarwal was discussing Von in interviews as the next expression of his long-running interest in revenue workflows. More than 200 Rattle customers had existed before that change. Starting over did not mean starting without a past.

Dinner, then silence

One story Aggarwal told in a June 2026 interview takes place at a customer dinner in New York. A revenue leader pulled him aside after the meal and described a project that had taken a team three months. The customer began comparing that effort with what might have been possible using Von, then paused before thanking his host and leaving.

It is an anecdote told by the person selling the product. Its value lies in what it reveals about his ambition. Aggarwal wants a customer to reconsider how much work a team can take on. The reaction he describes is closer to recalculation than applause.

He is particularly interested in the productivity changes that software engineers have experienced with AI tools. His question is what an equivalent change would look like for the people responsible for taking a product to market. Preparing a forecast, examining a pipeline and assembling a presentation are different tasks, but they draw on overlapping knowledge about the same business.

The difficulty is making that knowledge usable across the boundaries of individual applications. A customer’s story might be distributed across a call recording, an email exchange and a CRM record. Each piece can be available while the useful explanation still has to be assembled by a person.

The work between the meetings

Von’s stated job is to connect those systems, learn the company’s business context and help complete work through plain-language requests. Its public positioning covers operational tasks such as pipeline analysis, forecast preparation, territory planning and Salesforce updates. The company calls its approach a GTM brain. The promise depends on how well it learns the business behind the request.

In his conversation with Seth Marrs, Aggarwal compared AI with electricity: a capability that could eventually run through business activity so thoroughly that using it would cease to be a separate category. In his discussion with Sandy Robinson, he argued for an intelligence layer above existing systems of record. These are predictions about where value will move, and the product is his attempt to act on them.

For all the scale of that ambition, one of his more revealing observations is small. He has described opening a Zoom call and receiving four different offers to take notes. Helpful software had become a crowd of eager volunteers. Each could perform a similar trick; he still thought of each as having a different place in his routine.

That distinction brings the story back to his early test at Rattle. Getting a feature into a product is one achievement. Getting the product into someone’s day requires another kind of understanding. Aggarwal has spent years watching that boundary. With Von, he has put a rebuilt business on the proposition that there is still room to move it.