Daniel Ho Learned Sales by Building the Follow-Up Machine

Daniel Ho arrived at the founder's sales call with the wrong sort of résumé. He knew economics, data science, supply chains, product management and the peculiar pleasure of making an expensive process measurable. He had worked around Walmart's shelves, Lucid Motors' cars, Amazon's machinery and Tesla's suppliers. He did not know what salespeople meant by “discovery.” When Apten signed its first leads, his pitch wandered, his follow-up wobbled and the calls felt like exams for a class he had never taken.

There was one useful constraint: nobody else was coming. “If I didn’t pick up the phone, nobody did,” he later wrote. A founder can describe himself as a product person right until the cash register needs a person. Ho picked up the phone.

That awkward beginning is the cleanest introduction to Apten, the company he co-founded with engineer Roshan Kumaraswamy in 2023. Apten makes AI agents for the long, leaky journey between somebody raising a hand and somebody making a decision. The agent can begin in webchat, continue by text, switch to voice, send an email, remember the earlier exchange and bring in a human. It automates the very habit Ho was learning by hand: return, remember, respond.

The operator's eye

Ho's route into conversational software ran through physical operations. At Walmart, Lucid Motors, Amazon and Tesla, he built or worked with supply-chain analytics. As a Tesla product manager, he built systems that automatically detected supplier issues and supported chargebacks, recovering costs at substantial annual scale. The objects were parts, defects, vendors and dollars. The instinct was consistent: find the point where information stops moving, then design a system that makes the next action difficult to forget.

A lead funnel has different nouns but familiar physics. A person submits a form. The business calls too late. The person replies to a text but prefers a call. A colleague enters midway without the context. A scheduled sequence sends the same cheery message despite everything the person has already said. Each handoff is a small supply-chain failure, only the inventory has opinions.

Apten's first public version concentrated on SMS. Businesses could give the agent knowledge, rules, conversation flows and follow-up schedules. It could categorize leads, create alerts and decide when to return. Then voice joined text. Email followed. Webchat completed the loop from the first browser visit to the later phone conversation. Four Launch YC posts now read like tree rings, each recording a wider definition of one conversation.

The practical idea is less theatrical than the phrase “AI agent” suggests. People change channels. Memory should not be punished for it. Apten lets a team define flows, pull information from other systems, personalize timing, transfer a conversation to a human and apply guardrails before a message goes out. The difficult part is not producing a sentence. It is knowing which sentence belongs here, after all the sentences that came before.

Two kinds of systems builder

Apten's founding pair met with complementary histories. Ho studied Economics and Data Science at Berkeley. Kumaraswamy studied electrical engineering and computer science there, then spent three years at AWS building sales-planning applications and data platforms for Amazon's cloud sales organization. Earlier, he worked on multi-source data capabilities at InfluxData. One founder had learned to inspect the operational bottleneck. The other had learned to make large, complicated data systems behave.

Their division of labor is visible in the company's public face. Ho talks with prospects, demonstrates workflows and translates repeated complaints into a product argument. Kumaraswamy leads the engineering behind message pipelines, voice agents, context management and orchestration. Apten's open engineering role describes an environment with full ownership and no bureaucracy, a phrase that becomes more literal in a company where the complete team still fits around a table.

The pairing matters because omnichannel software is easily mistaken for a collection of integrations. Its harder requirement is continuity. A call, a text and an email arrive as different technical events, yet the person on the other end experiences them as one relationship. Keeping those perspectives aligned requires both the operator's intolerance for dropped work and the engineer's respect for state. Ho and Kumaraswamy built Apten in the gap between the two.

“Founders aren’t specialists. They’re shape-shifters.”Daniel Ho

The expensive customer

Six months into Apten, Ho and the team confronted the reverse problem: a customer who produced too many next actions. The account was one of their largest and generated a steady volume of leads. On the surface, it looked like the flattering sort of proof an early company wants.

Underneath, requirements flipped back and forth. Payments were missed or delayed. The customer's internal process was disorganized. Apten offered highly personal onboarding and support, a reasonable advantage for a tiny company, but whole days began disappearing into change requests that sometimes merely restored an earlier configuration.

Ho refunded the latest month and voided the contract. Early revenue enjoys an almost ceremonial status. It validates the idea, extends the runway and makes the weekly chart incline in the emotionally correct direction. Yet a dollar can arrive carrying several dollars' worth of distraction. Ho's decision was an operator's audit of the true bill.

“The most important decisions for a startup are the hardest ones,” he wrote. “But not making them will kill your company.” The line is stern, though the action behind it was disciplined rather than dramatic. They returned money, ended the obligation and recovered the team's attention. Focus sometimes appears on a calendar as white space bought at retail.

Shipping faster than the explanation

By the summer of 2025, Apten had acquired a more cheerful operational problem. Ho said the small team shipped so quickly that customers were discovering unfamiliar parts of the interface and asking what they did. In a matter of months, the company added AI email, warm and cold transfers initiated from text or voice conversations, a HubSpot integration, local-presence dialing, per-flow knowledge-base syncs and multilingual voice support.

Founders like speed because it photographs well on a roadmap. Understanding develops off camera. A feature is technically available before it is conceptually placed inside a customer's day. The team had shortened the distance between idea and code so aggressively that a new queue formed between code and comprehension.

Ho's public demos help close that gap. In an early series called Tip Tuesdays with Edwin, he walked through how a new lead could arrive from a form, be imported with the right context, receive a personalized text, schedule a meeting and get an automatic follow-up if the conversation went quiet. Edwin was a penguin mascot, a touch of levity for software concerned with the grave business of remembering to text back.

The demonstrations also reveal Ho's evolution as a seller. The vocabulary is no longer a barrier. He begins with what prospects tell him: they can find an AI texting tool or an AI calling tool, but struggle to find one that can do both. He converts that complaint into a product principle. Meet the lead where the lead is comfortable, even when comfort moves.

▶ Watch Apten's omnichannel product walkthroughDaniel Ho · Product video · YouTube

The skill nobody could automate for him

The irony of Ho's story is pleasant: he builds automation for follow-up because he had to learn follow-up without automation. In his telling, the founder's transformation was not mystical. He called, stumbled, adjusted and called again. Apten's current sales role still carries that founder-led texture. The job post asks observant applicants to contact Ho as though he were a prospect. The application begins by doing a piece of the work.

Apten now describes itself as profitable, with more than 30 percent average month-over-month growth across a recent six-month period. Y Combinator lists a team of six, split between the demanding offices of a young company: an engineering role in San Francisco and a sales role in New York. Ho's own location is the New York metropolitan area. His career has moved from large organizations with immense processes to a small organization trying to give immense processes a conversational memory.

The aspiration is clear in Apten's product progression. The team wants an agent that does more than answer. It should choose a channel, carry context, schedule the next contact, use outside systems, catch risky messages and know when a person should take over. Ho often describes the target in human terms. The useful tension is that human conversation is not one capability. It is timing, memory, judgment and the grace to stop.

That last quality may be Ho's sharpest operating lesson. Follow up when the silence is ordinary. Switch channels when preference changes. Hand over when a human is needed. End the relationship when the customer costs the company its direction. Automation is usually sold as persistence without fatigue. Ho's story adds a more adult clause: persistence becomes useful only when paired with discernment.

He began as the analytically trained founder who did not know discovery. He became the seller because the company required one. The charm is not that every founder should become a lifelong salesperson. It is that a job can reveal its missing verb, and a person can decide to conjugate it in public.