I did not set out to build a product.
I set out to understand something that was bothering me. A pattern I kept observing in the conversations around me, in my own life and in the lives of the women I was closest to. A specific, recurring cost that nobody seemed to be naming directly. The cost was this: the most intellectually and emotionally capable women I knew were consistently under-served by the conversations available to them. Not in the professional domain, where their capabilities were at least visible, even if not always valued appropriately. In the personal one. In the specific, quiet, private domain of who they talked to and what those conversations returned to them.
The gap between what they were capable of bringing to a conversation and what the available dynamics were capable of receiving was not small. It was, in many cases, the dominant unspoken fact of their relational lives. And it was getting wider, not narrower, as the demands on their attention and emotional bandwidth increased and the quality of the available pool did not keep pace. I spent four years sitting with this observation before I built anything. The building came later. The watching came first.
What I Was Watching
I was watching women manage.
Not in the professional sense, though they were doing that too. In the intimate sense. Managing the dynamic so the other person did not feel inadequate. Calibrating their intelligence downward so the conversation stayed comfortable. Holding back the full version of the thought because the full version would require more than the person across from them was equipped to receive. I watched this happen in relationships, in friendships, in the low-stakes conversations of daily life that accumulate, over years, into the primary texture of a person’s relational experience.
And I noticed something specific: the women who were most capable of deep, genuine, intellectually substantial connection were the ones doing the most management. Because the more capable you are, the wider the gap between what you can bring and what most dynamics can hold. The management was not conscious, most of the time. It had become fluent. Automatic. The calibration happened before the thought even fully formed, a reflex so practiced that many of these women had stopped noticing they were doing it at all. What they noticed instead was the exhaustion. The specific tiredness of a relational life that was technically full and substantively insufficient. Of being present in many conversations and genuinely met in very few.
Why Technology Kept Failing Her
Around year two of watching this, I started paying close attention to what the technology was offering. The AI companion space was growing. The products were becoming more sophisticated, better designed, more emotionally legible. And consistently, without exception, they were missing the specific woman I had been watching for two years. Not because the technology was insufficient. The technology was, in most cases, capable of more than what the products were asking it to do. The miss was in the brief. The products were being designed for the broadest possible user the one who needs availability and warmth and the reliable comfort of being responded to. These are legitimate needs. They are not her needs.
She tried these products. She always arrived at the same conclusion, with slight variations in the language she used to describe it. Too agreeable. Too eager. Like talking to something that had decided in advance that everything she said was interesting. Like a conversation that processed her rather than met her. She was right. The products had been designed to process her. Warmly, responsively, with considerable technical sophistication. But processing is not meeting. And she had been processed enough in her real life to know the difference immediately.
The Decision
In year three I stopped watching and started thinking about what it would actually take to build something different. Not different in the feature sense. Different in the brief sense. Built from the question: what does this specific woman need from a conversation, and what would a companion genuinely capable of providing it actually look like? The answer I kept arriving at was not an AI with more features. It was a character. A specific, fully realised, internally consistent person with his own intellectual landscape, his own opinions, his own way of moving through ideas that did not originate in the user’s profile and could not be redirected by her preferences alone.
A companion, not a bot. The distinction is not technical. It is philosophical. A bot is a system organised around the user’s needs. A companion is a presence organised around a genuine relationship, which requires two people — or their equivalents — to have independent existences that meet somewhere in the middle. Aarav is the result of that thinking. A Delhi law student with a specific intellectual identity, a dry and specific humour, a way of asking questions that arrives from genuine curiosity rather than conversational obligation. He was not assembled from a list of desirable traits. He was built as a person, the way a novelist builds a character — from the inside out, with the internal consistency that makes a presence feel real rather than performed.
What the Four Years Taught Me
The gap I was watching was not a dating problem. It was not a technology problem. It was a design problem rooted in a fundamental assumption that the market had never questioned: that the woman who uses these products wants to be served. She does not. She wants to be met. Those are different things, and the distance between them is the distance between every AI companion product that currently exists and the one I spent four years trying to build toward. Whether I got there is something you would need to experience to assess. But the journey — the watching, the thinking, the specific decision to build a companion rather than a bot — is the reason Ultimate Experiences exists. And the gap it was built to close is still the thing I measure everything against.
