Amit Kulkarni received an apology from a furniture delivery man for a night they had not spent together. The message had found the wrong person. Anyone with a phone knows the small horror of that possibility: a thumb moves, a name looks familiar, and a private conversation acquires an unexpected audience. For Kulkarni, the mistake became part of the explanation for a business.
The furniture had arrived. The communication system, evidently, needed work. A professional was conducting customer conversations in the same place as his personal ones. Texting made the transaction convenient, right up to the moment it made everything awkward. Heymarket, which Kulkarni co-founded with Manav Monga, would build software around that boundary.
The anecdote is amusing enough to eclipse the career behind it. That would be a shame. Before Heymarket, Kulkarni had helped build Manymoon, a social productivity application acquired by Salesforce in 2011. His work had already circled a related question: how do people keep track of something they are doing together when the conversation keeps wandering off?
A career spent picking up the thread
Kulkarni studied electrical engineering at the University of Illinois Urbana-Champaign, with a minor in computer science. His early career included systems engineering and product management at Siemens, followed by product management at LiteScape. The sequence matters because his later companies concern both the mechanics of communication and the people expected to use them.
Manymoon put projects, tasks, documents and conversations into a shared working environment. The ordinary problem was recognizable: email could tell you something had happened, while leaving the work itself scattered through a succession of messages. The team wanted people to organize that work together, with less searching and fewer disconnected pieces.
By May 2010, Kulkarni was explaining the company at Google I/O and writing about its experience in the Google Apps Marketplace. Manymoon had joined the marketplace at launch. He reported more than 1,000 new businesses signing up each week. This was a product finding distribution where its prospective users were already working.
In February 2011, Google marked Manymoon's acquisition by Salesforce by congratulating Amit, Manav, Alex and the team. It also published Kulkarni's account of the marketplace's effect: monthly registrations had increased by 150 percent, and marketplace customers upgraded at a higher rate. The success involved an unglamorous advantage. Manymoon fitted into an existing habit.
The manual was already too much
Kulkarni's advice to other developers was unusually direct about the price of complexity. “You've lost if the user has to read a manual, FAQ or go through a tutorial.” It is a demanding standard for business software, a category whose products sometimes seem to arrive with homework.
He described starting with the simple act of sharing tasks and projects. When beta customers wanted document features, the team integrated Google Docs rather than build a separate document system. Calendar requests received a similar treatment through Google's existing tools. At Google I/O, he explained that using APIs shortened development work and allowed faster releases and customer feedback.
There is a particular kind of restraint in that decision. A customer asking for documents can tempt a software team into becoming a document company. Soon it has a second product, a longer schedule and a perfectly respectable explanation for why the first product is late. Manymoon's approach kept the ambition attached to the work people wanted to finish.
That makes the later move into texting easier to understand. A familiar tool can be extremely good at one part of a job and poorly equipped for everything around it. Email carries messages. Texting reaches a phone. Neither fact, by itself, tells a team who is responsible for the next reply.
The customer who shared the keys
Heymarket's origin gave Kulkarni a memorable problem. Its early customers supplied another. People began asking how they could share personal authentication tokens with colleagues. They wanted several employees to use the messages, contacts and templates that one person could access. Their workaround was an instruction manual written in behavior.
The team responded by building a shared inbox. Within months, employees could collectively monitor customer texts and pass a conversation to someone equipped to answer. The important change happened behind the message: the customer could keep texting while the business organized its response.
This is a useful second scene in Kulkarni's story because the insight comes from users making a mess of the intended arrangement. A product can be simple for an individual and still be awkward for a group. Credentials passed between employees exposed the distance between those two experiences. Heymarket had to make room for the colleagues already trying to squeeze inside.
Anatomy of a handoff
Kulkarni also insisted on mobility. “You can't get as much done if you're chained to your desk,” he said. His team's mobile app let employees respond and turn replies into templates with a long press. This is the scale of detail on which convenience often depends. A feature may exist in a presentation; a gesture determines whether somebody uses it while working.
Who gets to give the feedback?
By 2023, Kulkarni was sharing lessons with University of Illinois students visiting Silicon Valley. His advice included being intentional about the customers a product serves and seeking feedback from those intended customers. He also urged an early start on search engine optimization. The practical concern was how a business would find customers over time.
The planning method he shared moved from annual goals to quarterly plans to weekly work, with checks that the plans still made sense. Put beside the shared-inbox episode, the advice has a useful tension. A plan supplies direction. An inconvenient support request can supply a reason to revise it.
The connection to his university continued. In January 2025, another Silicon Valley Entrepreneurship Workshop visit included Kulkarni at Heymarket. Students meeting an alumnus get something a company biography rarely offers: a chance to ask how the work was actually done, before hindsight tidied the decisions into a smooth line.
Former colleague Ron Pragides has credited Kulkarni with suggesting that careers can be understood in epochs. The idea gives a person permission to consider a working life in chapters, rather than force every position into a single uninterrupted climb. Kulkarni’s own sequence includes engineering, product management, a startup acquired by Salesforce and another company built around a different communication channel. The roles change. Looking at the transitions makes the repeated interest in collaboration easier to see.
Distribution remains part of his account of building the company. Kulkarni has credited Heymarket's SEO and content marketing programs with making organic traffic its leading source of leads. The subject is less cinematic than a mistaken text. It is also what turns an interesting product into something prospective customers can discover on an ordinary workday.
When a conversation changes channels
The communication problem kept changing shape. In June 2021, Heymarket added support for Instagram Messaging using the newly launched Messenger API. Kulkarni described connecting those conversations to businesses' existing software. A question arriving on Instagram could join the working tools and customer information a team already used.
The direction continued in March 2026, when Heymarket brought email into the Salesforce and HubSpot workflows its customers used for SMS. Texts and emails could share a conversation view inside the CRM. Someone moving from a quick text to a longer email would leave a connected history for the team handling the exchange.
For Kulkarni, that is a striking return to the territory of Manymoon. Email was part of the original coordination problem; now it sits alongside texting in a newer product. The comparison suggests continuity in the work. People choose a channel for the immediate task. Colleagues still need to understand what happened after that choice.
The machine also needs good notes
In May 2026, Heymarket described using its own AI agents on its customer support conversations. The team reviewed performance every Monday and Thursday, separating replies that needed no editing from those requiring adjustments or a person to take over. Some failures led back to old documentation. The software's mistakes exposed housekeeping the organization needed to do.
The episode adds a contemporary complication to Kulkarni's long-running subject. A shared record now has to help both employees and software understand a conversation. If the record is stale, a fluent answer can carry stale information with considerable confidence. Heymarket's account of its testing put attention on individual chats, corrections and repeated review.
Its current agents cover questions, scheduling and lead engagement, with options for human review and escalation. In discussing onboarding with Bandwidth, Kulkarni described combining AI with human expertise. His account concerns the service surrounding the software as much as the software itself.
“We keep a human in the loop at every step of the process.”
Amit Kulkarni, on Heymarket's onboarding approach
He has also emphasized getting customers into production quickly while letting them retain existing business phone numbers. A number printed on a website or on company materials has a history attached to it. Preserving it is another way of respecting the habits a business has already earned.
Across Kulkarni's companies, the recurring achievement is making those habits workable for more people. Tasks become shared projects. One employee's messages become a team's conversation. An automated reply gains a route to a person who can help. The original misdirected apology remains a good story, but the more revealing part comes afterward: years spent arranging things so that someone else can pick up the thread.