The company that taught Salesforce to speak Slack had to learn a new language

The first useful thing Rattle did was almost comically modest. After a sales call, a rep could get a prompt in Slack, add notes, and update Salesforce without going hunting through Salesforce. Sahil Aggarwal, who had worked on revenue teams himself, understood the small indignity buried in that routine: the work had happened in one place, but the official account of it had to be typed somewhere else.

In 2020, he joined Apoorva Verma and Milan Singh to build a way around the detour. The three made a minimum viable product in six weeks. They gave it to early users for free, watched what happened, and after four months asked them to pay. According to investor Insight Partners, roughly 80% agreed. There are more elaborate ways to validate a product. Few are as plain as asking people whether they want to keep it once it has become part of their day.

The case of the missing update

Every sales manager wants a current forecast. Every rep has a queue of calls, messages, approvals, and follow-ups. Between those two facts sits an apparently simple request: please update the CRM. When a record is stale, the cost is rarely confined to a messy field. A manager may coach the wrong deal, finance may trust the wrong close date, and customer success may inherit an account without the conversation that explains it.

Rattle’s original answer was to place the action in the communication tool a team already opened all day. An alert could arrive in a channel. A rep could edit a field from Slack or Teams. An approval could be given without another login. A deal room could gather the account executive, manager, and other specialists around one opportunity. The clever part was the location: a CRM remained the system of record, while the moment of action moved closer to the people doing the work.

The distinction mattered in a crowded market. A native Salesforce-Slack connection or a general automation tool could send an alert. Rattle tried to make the alert the start of a complete, two-way workflow. Moveworks later said it had compared Rattle with Troops and preferred Rattle’s workflow builder and approach to data storage. That is a customer’s account, not a universal verdict, but it describes the contest Rattle was actually in: whoever made a revenue process easiest to execute would earn the place in the daily routine.

A useful product gets ambitious

The company expanded beyond the original prompt. It added Deal Rooms for the people working the same opportunity, Atlas to draw the revenue process and locate bottlenecks, meeting intelligence, approvals, reports, and automatic CRM entry. The buyers were usually revenue operations and sales leaders; the daily users included account executives, SDRs, managers, and customer-success teams. Rattle sold the software as a business subscription, and its customers supplied the proof points that make a workflow product legible.

At Qualified, a pipeline generation company, a revenue operations leader used Atlas to see where handoffs and alerts sat in the process. Qualified’s published case study reports 1.5 times SDR and BDR productivity, deal cycles 14% faster, and a 5% rise in win rate within a quarter. Those are the customer’s reported outcomes across its broader use of Rattle, not a controlled experiment isolating one feature. They still tell you what the buyer was paying for: a team that noticed the right things in time to act.

Customer-reported results from one quarter of Qualified’s Rattle use.

Mezmo offered a smaller, more tangible story. It first moved quote approvals into Rattle, then expanded to handoffs and pipeline alerts. In its published case study, lead response fell from one hour to fifteen minutes, and the go-to-market operations team saved about four hours a week. A reader can copy the method without buying the product: choose one repeated delay, send the decision to the people who can resolve it, and measure elapsed time before adding more rules.

“My theory is it’s all about workflows.”Sahil Aggarwal, on finding a product people would miss

Rattle’s growth brought a $26 million Series A in April 2022, led by Insight Partners, with GV and earlier backers participating. The company said it had raised $28.8 million altogether. The money supported expansion of its team and product. By late 2023, it had introduced a broader Rattle 2.0, built around process visualization, automation, collaboration, and AI. On paper, the progression was sensible: start with a painful action, then make the whole process visible.

The problem with making the click easier

Then the question changed. If software could read a call, compare it with a CRM record, spot the discrepancy, and propose a correction, a faster route to the update was no longer the whole prize. Rattle tried to bring AI into its existing product. Aggarwal later described the experience as attaching an engine to a horse-drawn carriage. By the end of 2024, he had decided the architecture had to be rebuilt around the new capability.

This was not a painless rename. Wall Street Journal reporting described a company that had grown to around 70 employees and then shrank to roughly 15 as Aggarwal cut sales, marketing, and customer-success roles to concentrate on the rebuild. The old product also had customers whose workflows could not simply be switched off. Rattle’s current statement is that it remains fully supported for those customers. The forward product, Von, was tested in February 2026 and introduced publicly that spring.

Von aims at a different layer of work. It connects CRM records, call recordings, email, calendars, and other revenue data. A user can ask it to examine deal risk, investigate why deals were lost, prepare a briefing, or perform administrative work in Salesforce. The company says changes to records require confirmation and follow the user’s permissions. Its current FAQ describes three to five days to build a business context graph after connection, with a free trial and pricing by inquiry. That is a materially larger commitment than adding another notification rule: it needs connected data, governance, and confidence in the actions it proposes.

What the first product still teaches

The Rattle story is useful because the first product was not a failure in the ordinary sense. It won customers, produced reported operating improvements, and became a subscription business. Its limitation emerged when the unit of value moved. Rattle sold a more convenient way to perform a task. Von is trying to perform much of the task. Whether Von becomes a durable business is still being tested, but the reason for the change is visible.

For a revenue leader, there are two practical questions. First, which delay is costing the team money now: finding the signal, getting a decision, updating the record, or understanding what the data means? A Rattle-style workflow is strongest when the process is known and the obstacle is getting people to execute it. Second, does the proposed AI system have enough clean, connected context to do useful analysis and enough controls to make its actions reviewable? When those conditions are missing, a confident answer can be a very efficient way to make a mistake.

For a founder, the lesson is more uncomfortable. Four out of five early users agreeing to pay is excellent evidence that a product matters. It says nothing about whether its central task will look the same four years later. Rattle’s founders first learned where to put the work. Then they had to decide which parts of the work a machine could do. In software, even a useful answer has a shelf life.