Daniel Ha loves to argue. He loves it so much that his last company was an internet commenting platform that allowed people to banter about news. But he knows his contrarian bent can be tiresome for the people around him. 

So when AI came along, and others started thinking about how to build friends, romantic partners, or tutors, Ha, 39, saw an opportunity to make what he’d always wanted: a debate opponent who loves arguing as much as he does.

His early tinkering with AI to create a “digital version” of his brain has grown into something much bigger. He now runs his life with a legion of AI personas that have changed his relationships with his job, family members, and even his unborn child. 

While Silicon Valley’s tech behemoths and scrappy startups race to build sophisticated AI agents that they say will become indispensable to everyday life, Ha is already living a version of that future. He spends his days consulting AI employees at his investment firm, talking to an AI doctor about his father’s late-stage cancer, and preparing for the birth of his first child with his AI parenting coach. 

“It’s been game-changing,” Ha said. “It’s like a superpower that I cannot imagine living without.”

Ha’s experiments with AI agents began this year. He was at a Mission coworking space called The Alchemist when a golden idea struck: What if he built AI employees to staff his new venture capital firm? 

Ha and a fellow investor, Gadi Borovich, had recently founded an early-stage venture firm called Antigravity Capital. But their company had an early-stage problem itself — a lack of capital to hire staffers and other investors.

So Ha set out to create a digital analog of a small firm. He built AI agents using Google’s Gemini 2.5 Pro model, grounding them in dozens of hours of recorded conversations between him and Borovich about their worldviews and investing goals. 

AI agents are software programs that carry out tasks or make decisions in pursuit of a specific goal. Unlike chatbots or search engines, agents are proactive and don’t need continual human oversight. 

After weeks of tweaking, Ha and Borovich had an army of agents at Antigravity that they collectively christened their “diligence engine” and that now reviews every startup they consider investing in. 

There’s an “intake analyst” agent, which creates an initial report about the startup based on its pitch decks, business plans, and transcripts of Antigravity’s conversations with the team. There’s a “market researcher” agent, which considers the company’s product-market fit by creating industry-specific synthetic customers — you guessed it, also agents — to gauge demand. 

The resulting information is fed into the next tier of agents, which act as the digital equivalent of an angel and devil sitting on Ha and Borovich’s shoulders. The “optimistic investor” agent argues for a startup’s ability to eat the world, while the “pessimistic investor” agent describes its inevitable crash and burn. 

“One investor is the deal champion, and the other is the devil’s advocate,” Ha said. “Like that annoying colleague who says, ‘However …’”

Borovich, who worked with Ha to conceptualize the diligence engine, watched as it grew from an idea into an app on his iPad. 

“There were mornings when I woke up and saw Daniel had stayed up all night working on it,” Borovich said. “I saw the future of investing being built right before my eyes.” 

Ultimately, all the data has to go somewhere. The optimistic and pessimistic arguments are fed into an “arbiter agent,” which behaves like a wise third investor at the firm. Its job is to be “reverent” to Ha and Borovich’s worldview and make its decision based on their goal of funding industry-disrupting AI companies. This agent spits out a final report and a score between 1 and 100 measuring its confidence about investing in the startup. 

While the arbiter might sit at the top of the diligence food chain, Ha says he and Borovich don’t take its word as gospel. He’s aware that his system isn’t perfect and admits that they’ve invested in companies that received low scores from the arbiter. 

“It’s like having a smarter, more informed partner who’s done much more research,” Ha said. “But sometimes, we simply disagree.” 

Unlike Ha and Borovich, the diligence engine can’t have face-to-face meetings with founders at their hacker houses or huddle with them at the recent weeklong retreat Borovich organized at a Sea Cliff mansion. 

“What these AI personas can’t do is have dinner with a founder,” Ha said — interactions he finds particularly useful for understanding a person’s character and motivations. “The agents actually make me feel more human in the sense that I can focus on the things that only I can do.” 

The diligence engine provides another edge: With it, the venture firm more closely resembles the AI startups its principals want to invest in. Take Origami Agents, a Hayes Valley startup in Antigravity’s portfolio that’s building AI agents to take over the grunt work of human sales reps. 

“It’s really rare for investors to be so tech literate and build internal tools for their company,” said Origami cofounder Finn Mallery, who regularly talks through product challenges with Ha and Borovich over text and dinners at the Mission restaurant Base Camp. “They speak the same language as us.” 

When Ha clocks out of work, his AI squad tags along to Glen Park to lend a hand in his personal life. 

Last year, after his father was diagnosed with cancer, Ha began accompanying him to medical appointments. The oncologist rarely had time to address every question with the level of detail he and his father wanted. So Ha fine-tuned a large language model with thousands of oncology research papers and fed it detailed notes about his father’s diagnosis and health history.

“This isn’t for emotional support,” Ha said. “I’m trying to find the most fact-based, objective truth as possible.”

Now, every time there’s a development in his dad’s treatment or a new test result, Ha feeds it into the model, which considers the information and explains its conclusion. The digital doctor then helps Ha prepare for the next meeting with the real-life oncologist, suggesting questions and treatments to ask about based on the latest research. 

“So much of what we’ve done in treatment has been born out of the control we take and agency that we request,” Ha said. “Instead of being a passenger in this really shitty ride of cancer treatment, I feel like an educated ally.” 

When Ha found out late last year that his wife, Lisa, was pregnant with their first child, the feeling was a familiar mix of joy and terror. Most expectant parents use books or blogs to assuage their anxieties and prepare for the birth, but Ha and Lisa turned to something with a little more processing power.

They began having long conversations about what kind of parents they wanted to be. “We were talking about our parenting styles, but we had nothing to judge it against,” Ha said. In typical fashion, his answer was to personalize a chatbot with volumes of parenting research to serve as a mediator. 

Initially, Lisa had a different perspective. 

“At first, I rolled my eyes many, many times,” Lisa said. But she decided to humor him, and they began consulting their on-demand parenting coach a few nights a week with burning questions: when to introduce screen time to the child, say, and what kinds of adjectives they should use when issuing compliments. After learning from the AI coach that fetal hearing starts developing around 16 weeks of pregnancy, Ha began reading an AI summary of current events and sports news to Lisa’s womb every night. They also consulted with the parenting coach over a disagreement about whether to sleep in the same bed with the baby or buy a crib. After reviewing the research, they decided on the latter. 

“We litigate our arguments using deep research all the time,” Ha said. “It’s been nice to have a therapist with us in the room to help us unpack these ideas.” 

Lisa, once a skeptic, agrees. “I can see why it’s so intriguing to him,” she said. 

Ha doesn’t fear the AI characters chaperoning him through life are dulling his critical thinking. In fact, he believes they’ve sharpened his mind. But he does worry about younger generations outsourcing their brainpower completely to AI. 

All that leads to a key question: Will he let his child run its life with agents? 

“Absolutely not,” Ha said. “Maybe after they turn 25.”