This article is based on the original version by Terrence O’Brien.
Splice CEO Kakul Srivastava on AI Emails and the Erosion of Real Conversation
As the head of Splice, Kakul Srivastava has spent years navigating the fine line between AI-powered creativity and authentic human expression. She also, as she admits, hates her cup warmer. But it’s her views on AI emails that are most provocative in 2026.
In a wide-ranging conversation, Srivastava argues that AI-generated emails are quietly killing conversation. “AI emails are optimized for efficiency, not connection,” she says. “They strip out the hesitations, the tangents, the small human flourishes that make a message feel like it came from a person rather than a template.”
Why AI Emails Feel Like a Dead End
Srivastava points out that most AI writing assistants in 2026 are trained to produce polished, predictable prose. That’s great for scheduling a meeting or confirming an order. But when every email sounds the same, the recipient stops reading closely. “You get a kind of conversational numbness,” she explains. “People skim because they already know the shape of the message. The back-and-forth that builds trust and nuance disappears.”
She contrasts this with the kind of communication that actually moves projects forward. “The best emails I get are imperfect. They have a question that isn’t fully resolved, a half-formed idea. That’s an invitation to respond, not just to acknowledge.”
Threading the Line on AI Music Tools
At Splice, Srivastava has tried to apply the same philosophy to AI music tools. Splice, known for its sample library and creative tools, has integrated AI features that assist rather than replace. “We don’t want to generate a finished track for you,” she says. “We want to help you get unstuck, to suggest a chord progression or a texture that you then make your own.”
That approach has drawn both praise and criticism. Some artists worry that any AI in the creative process devalues human work. Srivastava counters that the tool is only as good as the intention behind it. “If you use AI to avoid making decisions, you’ll get generic results. If you use it to explore, it can be a catalyst.”
She also notes that in 2026, the debate has matured. “Two years ago it was all or nothing. Now people are asking better questions: Who owns the output? How do we credit influence? How do we keep the human in the loop?”
The Cup Warmer and Other Small Frustrations
On a lighter note, Srivastava admits she hates her cup warmer. “It’s a perfect symbol of useless optimization,” she laughs. “It keeps my coffee warm but somehow makes it taste worse. Kind of like an AI email that’s technically correct but emotionally empty.”
That anecdote underscores her broader point: not every problem needs a technological fix. Sometimes the better move is to accept the messiness of human interaction—whether in a song, an email, or a morning cup of coffee.
What Comes Next for Conversational AI
Looking ahead, Srivastava hopes AI email tools will evolve to support conversation rather than replace it. “Imagine an assistant that flags when a message is too generic, or suggests adding a personal detail rather than smoothing it out. That would be a real step forward.”
Until then, she offers a simple rule: if you wouldn’t say it out loud to a colleague, don’t let an AI say it for you. “Conversation is supposed to be a little rough. That’s where the meaning lives.”
via The Verge AI