.png){kind=logo}
Now in private beta with high-growth startups across five sectors, the company is giving its CEO and co-founder a growing body of real-world agent behavior to inform what the product becomes next.
Proper Motion is now analyzing more than 1 million AI agent interactions each week as Advitya Chhabra and his team develop a platform centered on continuous learning for AI agents. The company is in private beta with high-growth startups across healthcare, legal, finance, go-to-market, and developer tooling, while its enterprise pipeline includes multiple Fortune 100 companies. For Chhabra, Proper Motion’s current stage provides something early startups rarely have in abundance: real-world behavior that can challenge assumptions about what the product should become.
Startup companies often look more certain from the outside than they feel from the inside. As CEO and co-founder of Proper Motion, Chhabra divides his time between customer discovery, product development, and strategic decisions about what the company should build next. The work is less about following a fixed roadmap than repeatedly testing whether the assumptions behind that roadmap still hold. “People see a launch, a polished demo, or a customer quote and assume the founders must know exactly where everything is going,” Chhabra said. “Most of the time, you are still learning what the right answer is. The job is to keep getting closer to it without becoming too attached to what you thought a month ago.”
That process is visible in Chhabra’s day-to-day work. He helps automate customer acquisition outreach, books discovery calls and demos, writes product code, and builds personalized demonstrations for individual conversations. He is also involved in team morale, product prioritization, disagreements between co-founders, and decisions about whether the company needs to change direction. “If I am talking to customers and then going back and building the demo myself, there is very little distance between those two parts of the work,” he said. “Being involved in both sides of that process means I am hearing what customers are saying while I am also working on what we show them next.”
That proximity becomes more consequential as the volume of product activity grows. More than 1 million weekly agent interactions give the Proper Motion team a substantial stream of behavior to examine as it studies where agents succeed and where they break down. Customer conversations can then be considered alongside what the system is actually doing, keeping product decisions close to both qualitative feedback and usage patterns.
Chhabra’s experience before Proper Motion reinforced his willingness to change course. After leaving Citadel Securities, he and two friends initially started a company. For Chhabra, the experience showed that a startup can change direction even after an earlier idea has begun attracting interest. “You can get signs that something is working and still decide it is not the thing you want to keep building,” Chhabra said. “Changing direction does not always happen because there was no interest. Sometimes you learn that another problem is the one you actually want to pursue.”
Proper Motion’s current direction is centered on continuous learning for AI agents. Chhabra wants to understand how customers are using those systems and where they break down, then use those observations to inform what the company builds next. The private-beta companies span industries with different workflows and expectations, giving the team exposure to how agents perform across healthcare, legal, finance, go-to-market, and developer-tooling environments.
The company has also earned a place in a16z Speedrun, Andreessen Horowitz’s early-stage startup program. The program describes itself as highly selective, with an acceptance rate below 0.4 percent and recent cohorts generally comprising only 60 to 70 companies. For its SR006 cohort, Speedrun said it evaluated more than 19,000 startup pitches and admitted fewer than 0.4 percent of them. The program currently invests up to $1 million in participating startups.
By April 2026, a16z reported that Speedrun had backed more than 250 companies and invested more than $300 million across its first six cohorts. Nearly 10,000 investors engaged with companies in the preceding batch. Proper Motion entered that network while still in private beta, with its product development closely tied to the interactions and customer conversations the team is collecting.
The enterprise pipeline adds another dimension to that feedback loop. Multiple Fortune 100 companies are now among the organizations in Proper Motion’s pipeline, extending the company’s commercial conversations beyond the high-growth startups already using the private beta. Those discussions expose Chhabra and his team to another set of requirements as they determine which product assumptions hold across larger and more complex organizations.
Moving quickly inside that environment does not eliminate uncertainty. Chhabra still has to make product and company decisions before every question has been answered, then revisit those decisions as the company collects more evidence. The difference is that Proper Motion now has a growing volume of actual agent interactions and customer activity feeding that process.
Chhabra’s role also includes navigating differences between co-founders as product priorities and company direction evolve. Those discussions sit alongside customer discovery and technical development as another part of determining what the company should build and when an earlier assumption should be reconsidered.
Chhabra remains close to the work on both sides of Proper Motion. He hears customer reactions directly, works on product development, and participates in the strategic decisions that connect those inputs. That proximity leaves fewer layers between what users are experiencing and decisions about what the company should change next.
For Chhabra, the moving target is no longer simply an idea being debated inside a young company. Proper Motion now has enough live product activity and customer contact for assumptions to be tested against use instead of theory. The company remains in private beta, but the feedback loop is no longer hypothetical. What users and agents do can now influence what the team builds next.
