AI Can Sell Products, But Harder to Deliver Trust
January 27, 2026
Table of Contents
This article was originally featured in The Venture Lens, on October 23rd 2025 by Alastair Goldfisher. To read the full article, please visit The Venture Lens.
Reprise CEO Sam Clemens talks about why AI may perfect the demo but still can’t replace the human connection that closes the deal.
It happened so quickly. There was a live demo taking place, an eager audience and then an AI system that went off script.
That’s what happened in September at Meta’s Connect 2025 event, when the company was in the middle of showing off its new AI headset. The $799 Meta Ray-Ban Display was billed as a breakthrough in wearable agentic AI. But it froze in mid-presentation.
The on-stage team was then stuck in that awkward limbo where the tech they promised to showcase tanks and refuses to cooperate. That kind of failure is familiar to anyone who’s ever been responsible for a demo.
Sam Clemens has seen versions of it throughout his career
Clemens, a repeat founder and former partner at the venture firm Accomplice, has spent decades on both sides where technology meets storytelling. At every company he’s worked in — from Upwork (nee Elance) to HubSpot to InsightSquared — he’s noticed the same problem always comes up: the demo that fails.
“You’d end up demoing on production, crossing your fingers nothing would break,” he said.
Now, as co-founder and CEO of Reprise, Clemens has built a platform designed to eliminate that anxiety altogether. Reprise allows sales and product teams create fully functional, AI-powered demo environments that look real, are predictable and don’t risk exposing live systems.
The Boston-based company has raised more than $80 million from ICONIQ Growth, Bain Capital Ventures, Glasswing Ventures and Accomplice, where Clemens is a venture partner.
“AI has made product demos harder, not easier,” Clemens said. “You can’t predict what it’ll say or do in front of a customer. It’s too damn smart sometimes. But that’s also the opportunity.”
This paradox is obvious and a little poetic: AI created the problem and AI is what’s solving it. Reprise uses machine learning to replicate complex software environments in minutes, allowing companies, such as clients Cloudera, Litera and Pendo, to showcase their products.
Clemens calls it a way to “take the prayer out of the process.”
He would know. After leaving his full-time investing role at Accomplice, he realized he missed the short feedback loops of operating, the time it takes to see whether something works. “In venture, the feedback loop is eight years,” he said. “As a founder, it’s a week. You ship something, you see if it works.”
That bias toward building has kept him close to the human side of technology. Even as Reprise uses AI to automate the technical setup of demos, Clemens resists the idea that selling itself can be automated.
“It’s designed to guess what you want to hear,” he said of AI. “That’s not conducive to building real trust.”
Trust remains the most fragile and essential part of the sales process, he said. When a potential client is about to make a multimillion-dollar purchase, they’re putting their career on the line.
“They need to trust the person on the other side, not a system that’s trying to sound agreeable,” he said.
Why Selling Still Needs a Human Voice
That view puts him at odds with startups pitching AI as a replacement for salespeople. Clemens doesn’t buy it. To him, the real future of go-to-market work is about dividing tasks, letting AI handle the mechanical, repetitive parts so humans can focus on the narrative.
AI should “take away the drudge work so humans can do what only humans can do,” he said.
It’s easy to see the appeal. As companies push toward efficiency, the temptation is to offload more and more to automation. But there’s a danger in that impulse, a risk of eroding the very relationships that drive enterprise sales.
“If you break sales down into its tasks, sure, half of them could be automated,” he said. “But the half that remains is the most human part.”
